Evaluation of the Entertainment 
Worker Outreach Programs in 
Cambodia
REVISED REPORT 
HIV Innovate and Evaluate Project 
Contract AID-442-C-13-00001 
11 September 2017 
This study is made possible by the generous support of the American People through the United 
States Agency for International Development (USAID) The contents of this study are the sole 
responsibility of the HIV Innovate and Evaluate Project and do not necessarily reflect the views of 
USAID or the United States Government.
i 
Table of Contents
List of Tables .................................................................................................................................................. iii
List of Figures................................................................................................................................................. iv
Acknowledgements......................................................................................................................................... v
Executive Summary........................................................................................................................................ vi
Acronyms....................................................................................................................................................... ix
1. Introduction.............................................................................................................................................. 1
1.1. EW Outreach Program Description.................................................................................................... 3
1.2. SMARTgirl Program Description ........................................................................................................ 4
1.3. Flagship-TA GFATM Program Description.......................................................................................... 5
1.4. Non- Flagship GFATM Program Description ...................................................................................... 6
1.5. Comparison of the Flagship CoE SMARTgirl Program with Flagship-TA EW programs..................... 8
2. Rationale and Scope................................................................................................................................. 9
3. Evaluation Questions................................................................................................................................ 9
4. Evaluation Objectives.............................................................................................................................10
5. Evaluation Design ...................................................................................................................................10
6. Method of Evaluation.............................................................................................................................11
6.1. Method of Data Collection from EW ...............................................................................................11
6.2. Reference Period of Costing Data Collection ...................................................................................11
7. Sampling Procedures..............................................................................................................................11
7.1. Sample Size ......................................................................................................................................11
7.2. Sampling Strategy ............................................................................................................................14
7.3. Recruitment Process........................................................................................................................17
7.4. Eligibility ...........................................................................................................................................17
7.5. Place and Time of Interviews...........................................................................................................18
7. Evaluation Team.....................................................................................................................................18
8. Data Collection .......................................................................................................................................19
8.1. Data Collection Team.......................................................................................................................19
8.2. Training for Data Collection .............................................................................................................19
8.3. Fieldwork Management...................................................................................................................19
8.4. Data Collection Schedule .................................................................................................................20
8.5. Multiplicity .......................................................................................................................................20
8.6. Instruments......................................................................................................................................21
8.7. Incentives .........................................................................................................................................22
8.8. Sources of Costing Data ...................................................................................................................22
9. Data Management..................................................................................................................................22
9.1. Data Entry and Data Cleaning..........................................................................................................22
9.2. Weighting Adjustments ...................................................................................................................23
10. Operational Definition of Key Variables................................................................................................. 23
11. Data Analysis Framework.......................................................................................................................25
12. Ethical Considerations............................................................................................................................26
13. Limitations and Challenges.....................................................................................................................26
14. Results.................................................................................................................................................... 28
14.1. Descriptive Analysis.........................................................................................................................28
14.1.1. Program and Participant Characteristics....................................................................28
14.1.2. Risk, Discrimination and Stigma, Program Contact, and Sexual Activities.................30
14.1.3. Strategic Behavioral Communication .........................................................................32
ii
14.1.4. SMARTgirl Club/Drop-in-Centre .................................................................................35
14.1.5. Social Media and Communication Technologies........................................................36
14.1.6. Referrals to Health and Social Services.......................................................................37
14.1.7. Condoms.....................................................................................................................39
14.1.8. STI Screening and Treatment......................................................................................44
14.1.9. HIV Testing (Any Type)................................................................................................46
14.1.10. HIV Testing Using Finger Prick (CBHTC) ..................................................................48
14.1.11. HIV/STI Prevention Knowledge...............................................................................49
14.2. Program Impact ...............................................................................................................................50
14.2.1. STI Screening and Treatment......................................................................................50
14.2.2. HIV Testing (Any Type)................................................................................................52
14.2.3. HIV Finger Prick Test (CBHTC).....................................................................................54
14.2.4. Stigma and Discrimination..........................................................................................57
14.2.5. Condom Use................................................................................................................59
14.2.6. Referrals for STI Screening and Treatment................................................................. 62
14.2.7. HIV/STI Prevention Knowledge...................................................................................65
15. Cost Allocation........................................................................................................................................69
16. Discussion............................................................................................................................................... 72
17. Conclusion .............................................................................................................................................. 74
18. References .............................................................................................................................................. 75
19. Annexes.................................................................................................................................................. 76
19.1. Printed Education Materials............................................................................................................76
19.2. Sexual Activities...............................................................................................................................80
19.3. Condom Use.....................................................................................................................................82
19.4. Program Impact ...............................................................................................................................82
19.4.1. STI Screening...............................................................................................................82
19.4.2. HIV Testing (Any Type)................................................................................................90
19.4.3. HIV Finger Prick Testing ..............................................................................................97
19.4.4. Stigma and Discrimination........................................................................................104
19.4.5. Cost Allocation..........................................................................................................109
iii
List of Tables
Table 1 EW HIV program coverage ............................................................................................................. 3
Table 2 Core services of Flagship CoE, Flagship TA sites, and Non-Flagship sites...................................... 7
Table 3 Timing of initiation of key services at selected CoE, Flagship TA, and Non-Flagship sites............. 9
Table 4 Percentage distribution of key related variables for sample size computation ..........................12
Table 5 Estimated sample sizes for a two-sample proportions test ......................................................... 12
Table 6 Actual sample size ........................................................................................................................ 13
Table 7 Results of recruitment.................................................................................................................. 14
Table 8 Types of initial seeds.................................................................................................................... 15
Table 9 RDS recruitment numbers and probabilities ................................................................................ 16
Table 10 Data collection schedule .............................................................................................................. 20
Table 11 Program exposure by geographic area ........................................................................................ 28
Table 12 Sociodemographic characteristics by geographic area and program exposure ..........................29
Table 13 Risk, discrimination and stigma, and program contact by geographic area and 
program exposure ........................................................................................................................ 30
Table 14 Sexual activities by geographic area and program exposure....................................................... 31
Table 15 Strategic behavioral communication ........................................................................................... 32
Table 16 SMARTgirl Club/Drop-in-Centre ................................................................................................... 35
Table 17 Social media and communication technologies........................................................................... 36
Table 18 Referral to health and social services........................................................................................... 37
Table 19 Condoms....................................................................................................................................... 39
Table 20 STI screening and treatment........................................................................................................ 44
Table 21 HIV testing and counseling (HTC)................................................................................................. 46
Table 22 HIV testing using finger prick (CBHTC) ......................................................................................... 48
Table 23 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTgirl program on STI screening and treatment in the past 6 months ................................51
Table 24 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTgirl program on any type of HIV testing in the past 6 months.........................................53
Table 25 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTgirl program on HIV finger prick testing in the past 6 months.........................................55
Table 26 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTGirl program on high stigma and discrimination.............................................................. 58
Table 27 Binary Logistic Regression Model: The effect of geographic area on condom use 
at last sex with client.................................................................................................................... 59
Table 28 Binary Logistic Regression Model: The effect of program exposure on condom use at 
last sex with client........................................................................................................................ 60
Table 29 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTgirl program on condom use at last sex with client.......................................................... 61
Table 30 Binary Logistic Regression Model: The effect of geographic area on referral for 
STI screening and treatment in the last 6 months....................................................................... 62
Table 31 Binary Logistic Regression Model: The effect of program exposure on referral 
for STI screening and treatment in the last 6 months ................................................................. 63
Table 32 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTgirl program on referral for STI screening and treatment in the last 6 months...............64
iv
Table 33 Binary Logistic Regression Model: The effect of geographic area on 
high prevention knowledge.......................................................................................................... 65
Table 34 Binary Logistic Regression Model: The effect of program exposure on high prevention 
knowledge .................................................................................................................................... 66
Table 35 Binary Logistic Regression Model: The effect of exposure to 
different activities of SMARTgirl program on high prevention knowledge ................................. 67
Table 36 Annual unit costs.......................................................................................................................... 71
List of Figures
Figure 1 Posttest-only with nonequivalent groups.................................................................................... 11
Figure 2 Average social network size by type of EW.................................................................................. 16
Figure 3 RDS recruitment process.............................................................................................................. 17
Figure 4 Percentage of EW ever saw printed education materials (Flagship CoE) ....................................34
Figure 5 HIV/STI prevention knowledge by geographic area..................................................................... 49
Figure 6 HIV/STI prevention knowledge by level of program exposure .................................................... 49
Figure 7 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of geographic area on the probability of getting STI screening 
and treatment in the previous 6 months..................................................................................... 50
Figure 8 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of program exposure on the probability of getting STI s 
creening and treatment in the previous 6 months...................................................................... 50
Figure 9 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of geographic area on the probability of getting HIV testing (any type) 
in the previous 6 months ............................................................................................................. 52
Figure 10 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of program exposure on the probability of getting HIV testing (any type) 
in the previous 6 months ............................................................................................................. 52
Figure 11 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of geographic area on the probability of getting HIV finger 
prick testing in the past 6 months................................................................................................ 54
Figure 12 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of program exposure on the probability of getting HIV finger 
prick testing in the past 6 months................................................................................................ 55
Figure 13 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of geographic area on the probability of being in high stigma and 
discrimination ............................................................................................................................... 57
Figure 14 Inverse-probability-weighed regression adjustment (IPWRA) model: 
The effect of program exposure on the probability of being in high stigma and 
discrimination ............................................................................................................................... 57
Figure 15 Cost allocation .............................................................................................................................. 69
Figure 16 Cost allocation of programs ......................................................................................................... 70
Figure 17 Numbers of HIV tests and positive cases; unit costs per HIV test and per case detected...........71
v 
Acknowledgements
We would like to express our deep appreciation to Dr. Ly Penh Sun, Director of the National Center for 
HIV/AIDS Dermatology and STDs (NCHADS) for his leadership of this evaluative study. From NCHADS, Dr. 
Lan Vanseng, Deputy Director, Dr. Samrith Sovannarith, Chief of Technical Bureau, and Dr. Tep Samnang, 
Chief of Behavior Change Communication Unit, provided invaluable inputs for the study formulation to 
ensure that the results are relevant to the national program. 
This study was made possible by the generous support of the American People through the United States 
Agency for International Development (USAID). 
We extend sincere gratitude to FHI360 and KHANA, for their technical inputs, support, and information 
sharing in the development of the protocol. 
We also would like to express our sincere thanks to Partners for Development (PFD), Cambodian Women 
for Peace and Development organization (CWPD), Phnom Srey Organization for Development (PSOD), and 
KHEMARA for their warm collaboration in the process of sampling and data collection. Successful data 
collection could not have happened without the collaboration and support from management and outreach 
staff of these organizations. 
We also would like to also express our sincere gratitude to the Provincial health departments in Kampong 
Cham, Siem Reap, Banteay Meanchey, Battambang, Pursat, Kampong Thom, Preah Sihanouk, Pailin, 
Battambang, Kep, Kampot, Kampong Thom, Preah Vihear, and Phnom Penh as well as selected referral 
hospitals and health centers from these locations for their collaboration in this study, in particular for 
allowing their technical staff to be interviewed by our researchers. 
Our field researchers were at the heart of the study’s success. We would like to express our appreciation to 
them for their work.
vi
Executive Summary 
Background 
In Cambodia, female sex workers, or entertainment workers (EW), represent the largest of Cambodia’s key 
populations (KP) at highest risk for HIV infection, with an estimated 40,000 EW in 2014. The overall HIV 
prevalence among EW most recently was estimated to be 3.2% in 2016, with prevalence rates as high as 
11.85% among freelance EW and 8.3% among EW with more than 2 partners daily. EW also are at increased 
risk for sexually transmitted infections (STIs), illicit drug use, and unmet family planning needs. 
The national HIV program has implemented a multi-sectoral response tailored to EW, including the 100% 
Condom Use Program (CUP), which has been credited with the high (86%) consistent condom use among 
EW in 2016. The Boosted Continuum of Prevention, Care and Treatment (Boosted CoPCT) approach, 
developed by the National Center for HIV/AIDS Dermatology and STIs (NCHADS) has guided the national 
response, focusing on the identification and testing of KP to address the unique HIV needs for KP, including 
EW, with a core service package that includes: improving case detection, avoiding new infections and 
reducing individual risks, strengthening referrals and linkages, care and treatment, and integration of sexual 
reproductive health and family planning. 
Under the umbrella of Boosted CoPCT since 2013, programs for EW are spread throughout Cambodia. 
There are three key types of EW programs. First, Centers of Excellence (Flagship COE) are directly 
supported by the USAID HIV Flagship Project (Flagship) and that focus on testing and implementing 
innovations on HIV prevention and other linkage services. Second, Flagship-Technical Assistance (Flagship 
TA) sites that are directly supported by the Global Fund but also receive TA from Flagship. The third type of 
EW program is supported by the Global Fund and receives no TA from Flagship (Non-Flagship). 
No systematic impact evaluation of EW outreach programs has yet been undertaken. Information from this 
evaluation aims to help the national program and donors to more effectively and efficiently shape 
programs and deploy resources. This evaluation measured the uptake of HIV tests, condom use, STI 
screening and treatment, stigma and discrimination and, referrals to health services. Comparisons of 
outcomes across the three program types were performed. In addition, the evaluation looked at the cost￾effectiveness of the program by comparing the unit costs. Perspectives of the clients regarding the quality
of service delivery were also examined in this evaluation.
Findings
Data were collected from 1300 EW respondents, including 300 from the geographic areas of the Flagship 
CoE sites, 300 from the geographic areas of Flagship TA sites, 300 from the geographic areas of Non￾Flagship sites, and 400 from geographic areas with no existing EW programs. There was a wide diversity 
among EW with regard to age and education. 10% of EW reported sexual violence and 41% of EW reported 
bullying/harassment from a client in the past 12 months, demonstrating the vulnerability faced by EW.
Employment
KTV was the dominant primary employer of EW (62%) and comparatively small proportions of EW reported 
their main occupations to be either freelance sex worker (3.1%) or sex worker (3.4%). Overall, about 57% of 
EW had incomes greater than $250 per month. Controlling for confounders, EW with higher incomes 
experienced significantly lower levels of stigma and discrimination, and had higher condom use than those 
with lower incomes. EW exhibited a high level of geographic mobility (43% of EW with less than one year 
living in current location) and employment mobility (52% of EW with less than one year at current work 
place). 
vii
Risk
57% of EW were at medium or high risk for HIV, as judged by the risk assessment index. None of the 
programs, however, appeared to successfully target high risk EW, with all three of the program areas 
predominantly reaching a low and medium risk EW (74%). Indeed, a larger proportion of EW with low risk 
(37%) than with high risk (28%) had a high level of exposure to the program, showing the challenges 
experienced in reaching highest risk EW. Only 7.5% of EW had seven or more sexual partners per week, 
though 82% usually had at least one sexual partner per week, indicating ongoing risk exposure. 
Program Exposure
Only about half (51%) of EW in CoE geographic areas reported contact with an outreach worker in the 
previous 12 months. Only 34% of EW in CoE areas had contact with an outreach worker in the previous 
three months, a lower figure than the proportions of EW that had contact with an outreach worker in the 
previous three months in Flagship TA and Non-Flagship areas (37% and 43%, respectively). 
The majority (52-65%) of EW in the three program areas reported that OW were their main source of 
information about HIV and STI services, though only 3-36% of EW said that OW were their preferred 
channel for this type of information. By contrast, there appeared to be strong demand for broadcast media 
(television and radio), which combined for 52% of responses by EW as their preferred channels for 
information about HIV and STI services. 22% of EW said that Facebook was their preferred communication 
channel, which contrasts with the underutilization of the SMARTgirl website (only 3% utilization among EW 
in CoE areas) and the SMARTgirl Facebook page (only 7% utilization among EW in CoE areas). 
Exposure to printed education materials was suboptimal, with only four of the 20 printed materials being 
well recognized by the EW interviewed. Furthermore, there was no measurable impact on prevention 
knowledge among EW with regard to exposure to printed educational materials.
There was low utilization of SMARTgirl clubs and drop-in centers. The majority of EW (61%) in CoE areas 
had heard about the SMARTgirl club but only 20% had visited the SMARTgirl club in the previous six 
months. Less than 3% of EW in Flagship TA and Non-Flagship areas had visited a drop-in center in the 
previous six months. Visits to the SMARTgirl club did not correlate well with prevention knowledge, 
condom use, stigma/discrimination, STI screening, or HTC uptake.
Utilization of referrals was also suboptimal, with less than 10% of EW in the three program areas having 
been referred for family planning. Indeed, small proportions of EW were referred for STI services in the 
previous 12 months (13-18% in the three program geographic areas). Only 26% of highly exposed EW had 
been referred for STI screening/treatment in the previous 12 months.
Condoms
97% of EW reported using a condom at their last sex with a client. Among the small number of EW that did 
not use a condom at the last sex act with a client (37 of 1137 respondents), 38% said the main reason was 
that their client refused condom use. Only 7% of EW reported use of a condom at the last sex with their 
husband.
STI Screening & HTC
Only 44% of EW reported STI screening in the past six months, with a significantly smaller proportion of EW 
in non-program geographic areas receiving STI screening than EW in the three program geographic areas. 
56% of EW had undergone HTC in the previous six months, again with a significantly lower proportion of 
EW in non-program geographic areas (46%) receiving this than EW in the three geographic areas covered 
by EW programs (60-67%). 
Because only 7 EW reported being HIV+, no meaningful analysis regarding the effectiveness of the EW 
programs with regard to key HIV+ status-related outcomes (e.g. VCCT confirmation testing, identifying new 
cases, reducing LTFU for HIV testing confirmation, or ART enrollment and retention of HIV+ EW) could be 
made. 
viii
Program Impact
Controlling for confounders, the EW programs had a measurable and positive impact on STI 
screening/treatment, HTC, and stigma/discrimination. EW in non-program areas were less likely than EW in 
program areas to utilize STI screening/treatment (37% versus 44-47%), less likely to receive HTC (48% 
versus 59-67%), and were more likely to report high stigma/discrimination (62% versus 42-51%). These 
differences were statistically significant. This shows powerful evidence of the impact of the EW program. 
Furthermore, compared to EW with high program exposure, EW with no program exposure were less likely 
to receive STI screening/treatment (33% versus 62%), were less likely to receive HTC (44% versus 79%), and 
were more likely to experience high stigma/discrimination (60% versus 38%).
Comparing program impact across geographic areas, it appears that the Non-Flagship areas performed best 
among the three, with EW in these geographic areas being significantly more likely to receive HTC (any type 
or community-based) and STI screening/treatment than the other two geographic program areas. CoE EW 
programs had the overall lowest impact among the three (with the exception of STI screening/treatment 
where CoE performance was better than Flagship TA areas). 
Cost
Program costs were much higher for CoE than for Flagship TA and Non-Flagship areas, yielding an overall 
higher cost per HIV test of approximately $40, compared to $21 at Flagship TA areas, and $30 at Non￾Flagship areas. There was also a higher cost per HIV case detected in CoE areas of $15,426, compared to 
$10,979 at Flagship TA areas and $8,503 at Non-Flagship areas. 
Conclusion 
EW show a great diversity of demographic, work, and socioeconomic conditions. Common issues were seen 
with regard to geographic mobility, gender-based and domestic violence, and condom use patterns. 
Uptake of HTC and STI screening/treatment fell below expectations. There was underutilization of referrals 
for family planning, printed materials, SMARTgirl clubs and drop-in centers, as well as social media among 
EW across geographic locations. 
However, it was clear that exposure to the EW programs boosted HTC and STI screening/treatment and 
was associated with decreased stigma and discrimination. The highest EW program impact was seen in 
Non-Flagship locations. Compared to CoE and Flagship TA areas, Non-Flagship areas had the lowest cost per 
HIV infection detected.
ix
Acronyms
AIDS Acquired Immune Deficiency Syndrome
ART Anti-Retroviral Therapy
AusAIDS Australian Agency for International Development
CoPCT Boosted Continuum of Prevention to Care and Treatment
CBHTC Community-Based HIV Testing and Counseling
CBPCS Community-Based Prevention, Care and Support
CCC Country Coordinating Committee
CoE Center of Excellence
CUP Condom Use Program
CWPD Cambodian Women for Peace and Development
DIC Drop-In-Center 
EW Entertainment Workers
FHI Family Health International
FP Family Planning
GF Global Fund
GFATM Global Fund for AIDS, TB and Malaria
HEF Health Equity Fund
HIEP HIV Innovate and Evaluate Project
HIV Human Immunodeficiency Virus
HTC HIV Testing and Counseling
ICT Information and communication technology
IEC Information, education and communication
KHANA Khmer HIV/AIDS NGO Alliance 
KP Key Population
LTFU Loss to follow up
NCHADS National Center for HIV/AIDS, Dermatology and STI
NECHR National Ethic Committee for Health Research
NGO Non-governmental organization
NTA None Technical Assistance
OD Operational District
OW Outreach Workers
PFD Partners for Development
PRASIT Project for HIV and AIDS Strategic Technical Assistance
Pre-ART Prior to Anti-Retroviral Therapy
PSOD Phnom Srey Organization for Development 
PWID People Who Inject Drug
Q Quantity
RDS Respondent Driven Sampling
SBC Strategic Behavioral Communication
SIT Save Incapacity Teenagers
SOP Standard Operating Procedure
SRH Sexual and Reproductive Health
STI Sexually Transmitted Infection 
TA Technical assistance
TB Tuberculosis
UC Unit Cost
UIC Unique Identification Card
UNAIDS United Nations Agency for HIV/AIDS
URC University Research Co., LLC
USAID United State Agency for International Development
VCCT Voluntary and Confidential Counselling and Testing
WHO World Health Organization
1 
1. Introduction
Among the key populations (KP) at risk for HIV infection in Cambodia, female sex workers, or 
“entertainment workers” (EW), is the most populous.1 The latest population estimates showed that there 
were approximately 40,000 entertainment workers in 2014, and the majority of them lived in Phnom Penh 
(59%), followed by Siem Reap (9%), Battambang (6%), and Banteay Meanchey (4%) (WHO, 2013). EW can 
be categorized into sub­groups corresponding to their locations of work, including karaoke establishments, 
massage parlors, bars, beer gardens, and freelance (street based and non-street based) (NCHADS, 2013). 
A decline in HIV prevalence among EW was recorded in Cambodia between 1998 and 2013 with prevalence 
in EW decreasing from 46% to 10-14% among high risk EW (WHO, 2013). The latest Cambodia Female 
Entertainment Workers Integrated HIV Bio-Behavioral Surveillance reported a 3.2% overall HIV prevalence 
among EW (FEWIBBS, 2016). The HIV prevalence among EW varied based on type of EW: those who
freelance had an 11.85% HIV prevalence rate, while the prevalence rate among EW who work as beer 
promoters and at karaoke/massage parlors were 2.4% and 1.6%, respectively. EW with more than 2 
partners daily had an 8.3% HIV prevalence (FEW IBBS-2016). 
Given the diverse characteristics and HIV prevalence rates among EW, the national HIV program 
implemented a multi-sectoral response to the concentrated HIV epidemic including a range of 
interventions tailored to EW. One key intervention was the 100% Condom Use Program (CUP) whereby
transactional sex establishments in 24 provinces were required to implement the CUP since the year 2000. 
The aim of the CUP was to prevent heterosexual HIV transmission linked to sex work ensuring that 
condoms are used 100% of the time, in 100% risky relationship in 100% sex entertainment establishments 
(WHO, 2004). The program has been highly successful, with 86% consistent condom use among EW in 
2016, and 92% condom use among EW who engaged in paid sex in last sexual intercourse (FEWIBBS, 2016). 
The national response against HIV in Cambodia for EW is centered on the Boosted Continuum of 
Prevention, Care and Treatment (Boosted CoPCT) approach, developed by the National Center for HIV/AIDS 
Dermatology and STIs in 2013 (NCHADS). This approach focuses on identifying and testing KP to address the 
unique HIV needs for KP, including EW, with a core service package that includes: improving case detection, 
avoiding new infections and reducing individual risks, strengthening referrals and linkages, improving care 
and treatment and, integration of sexual reproductive health and family planning (NCHADS, 2013). 
Boosted CoPCT is a key component of Boosted Integrated Active Case Management (B-IACM), the 
cornerstone of the Cambodia national HIV program to achieve its targets by 2020 of 1) 90% of people living 
with HIV (PLHIV) knowing their HIV status, 2) 90% of those diagnosed being on treatment, and 3) 90% of 
PLHIV on treatment achieving viral suppression. B-IACM is a client-oriented approach designed to provide 
support to individuals to receive HIV services across the HIV service cascade, with the goal of decreasing 
losses to follow up along the HIV care cascade in Cambodia. The implementation of B-IACM is 
operationalized through the Identify Reach, Intensify and Retain (IRIR) mechanism where B-IACM focuses 
on identifying and reaching all old and new infections and in intensifying efforts to ensure cases are 
brought into the HIV cascade to receive immediate ARV treatment while retaining all PLHIV on treatment 
and achieving viral suppression (NCHADS, 2017).
In alignment with the national program to improve HIV services for EW, the US Agency for International 
Development (USAID) HIV/AIDS Flagship Project implemented the SMARTgirl program in Cambodia from 
2013-2017 with a goal of increasing HIV testing and counseling (HTC), STI screening/treatment, condom 
use, and strengthening STI referrals for EW. 
 1 The term “sex worker” refers to “individuals who receive money or goods in exchange for sex services, and who consciously 
define those activities as income generating even if they do not consider sex work as their occupation”. After the 2008 outlaw of 
sex work and brothels in Cambodia, a local definition of “sex work” has been replaced with the term “entertainment worker” (EW), 
describing “individual women employed in the entertainment sectors in restaurants, massage parlor etc. regardless of their 
possible involvement in direct or indirect sex work”(ILO, 2011).
2 
Key Issues for Entertainment Workers
HIV testing and counseling (HTC) is a key intervention to identify EW that may be living with HIV and 
provide them with services to avert ongoing transmission of HIV. Early diagnosis of HIV is an effective 
measure in the fight against the epidemic when EW with self-perceived risk for HIV initiate frequent HIV 
testing and adopt safe sex practices with their partners (Parriault, 2015; NCHADS, 2013). In 2016, 72% of 
Cambodian EW reported having been tested for HIV in the previous 12 months, which may indicate high 
self-perceived risk for HIV in this group (FEWIBBS, 2016). Reasons of not taking an HIV test among EW at 
health facilities and community-based locations included discrimination by health providers, sexual 
harassment by health providers, neglected by health providers, not trusting the capacity of outreach 
workers in keeping confidentiality (HIEP, 2015b). 
The effect of STIs in increasing HIV infection is well documented and individuals who are infected with 
sexual transmitted diseases are two to five time more likely to get infected with HIV if they exposed to the 
virus (CDC, 2010; Feng, 2010; Ward, 2010 ). In Cambodia, the STI prevalence rates among EW have fallen in 
recent years, but remain high compared to the general population: Syphilis fell from 14% to 4%, and 
chlamydia prevalence declined from 23% to 13% (Heng, 2008). STIs are the major causes of reproductive 
morbidity and mortality due to frequent exposure to unsafe sex practice, particularly for sex workers 
characterized by a high number of sexual partners and poor health seeking behavior (Plummer, 1991; 
Thomas & Tucker, 1996). 
Illicit substance use is a major issue among sex workers, with one study from Canada showing that 
cumulative HIV incidence among injecting drug users who engaged in sex work was 12% compared to non 
IDU (7%) (Kerr, 2016). The 2016 FEWIBBS examined overlapping vulnerabilities between sex work and illicit 
substance use, finding that the prevalence of use of drugs such as Amphetamine, Yama and Ice is about 
10% among EW, and 1.3% of EW use injected drugs (FEWIBBS-2016). 
EW also often have unmet family planning needs. According to the recent national FEWIBBS, 40% of EW 
experienced at least one pregnancy while working as EW and one third had at least one abortion. However, 
73% of EW used any method of family planning, and of those, the majority (37%) used condoms as a 
primary form of contraception, with 14% of EW using the contraceptive pills as the second main method 
(FEWIBBS, 2016). Another study conducted by HIEP in 2015 showed that EW reporting that limited family 
planning services nearby their homes and lack of affordable services were predictors of unwanted 
pregnancy (HIEP, 2015a). 
3 
1.1. EW Outreach Program Description 
Across Cambodia, EW outreach programs have taken different forms. The three key forms can be 
summarized as: 1) Flagship- Centers of Excellence (Flagship COE), 2) Flagship-Technical Assistance sites 
where Global Fund sites received TA from Flagship (Flagship TA), and 3) Non-Flagship sites, which are 
Global Fund sites with no TA from Flagship (Non-Flagship). Table 1 describes the geographic distribution of 
these program types, as well as the NGOs that support their implementation.
Table 1 EW HIV program coverage
Province NGO OD
Flagship 
SMARTgirl 
CoE sites
Flagship TA 
sites/GFATM sites Non-Flagship sites
Banteay Meanchey PFD Mongkol Borei 577 
Poi Pet 1,102 
Battambang CWPD Battambang 1,739 
Sampeu Loun 181 
Kampong Cham PSOD Kampong Cham 640 
Kampong Chhnang CWPD Kampong Chhnang 377 
Kampong Speu CWPD Kampong Speu 385 
Kampong Thom CWPD Kampong Thom 500 
Kampot CWPD Kampong Bay 216 
Kandal CWPD Takhmao 773 
Koh Kong CWPD Smach Meanchey 359 
Kratie CWPD Kratie 196 
Mondolkiri KHEMARA Mondulkiri 367 
Oddormeanchey CWPD SOMRONG 251 
Pailin CWPD Pailin 367 
Phnom Penh CWPD Chaktomuk 1,150 4,685 
Basak 704 
KHEMARA Dangkor 385 
Sen Sok 125 
Mekong 1,036 
Porsenchey 1,967 
SIT Mekong 2,998 
Preah Sihanouk KHEMARA Preah Sihanouk 477 
Preah Vihear CWPD Preah Vihear 281 
Prey Veng KHEMARA Neak Loeung 237 
Svay Antor 75 
Pursat PFD Sampov Meas 373 
Ratanakiri KHEMARA Banlung 533 
Siem Reap CWPD Siem Reap 2,379 
Stung Treng CWPD Stung Treng 204 
Takeo CWPD Daun Keo 317 
Tboung Khmum CWPD Tboung Khmum 220 
Total 4,169 22,007 26,176 
Source: KHANA, 2016
4 
1.2. SMARTgirl Program Description 
Flagship COE sites implement a program called “SMARTgirl”. Under the PRASIT Project, FHI 360 Cambodia 
and its partners introduced the SMARTgirl HIV prevention and care program in October 2008. SMARTgirl 
aimed to improve the sexual health and general well-being of EW through an innovative, holistic, human 
rights-based, branded sexual health program. The SMARTgirl program activities included the delivery of 
core services through individual and group level outreach conducted at a variety of venues where EW work 
(USAID, 2014). 
Subsequently, the USAID HIV Flagship Project (Flagship) has worked on testing and implementing 
innovations on HIV prevention and other linkage services for EW from 2013-2017. The goal of the renewed 
SMARTgirl program was to reduce HIV incidence among EW in Cambodia. The project was funded by USAID 
and focused on technical assistance and piloting innovations in HIV prevention, care, support, and 
treatment. The Flagship project worked through Centers of Excellence (CoE) to test new approaches to 
reduce HIV incidence among EW in three cities: Phnom Penh, Siem Reap and Kompong Cham, in 
collaboration with USAID implementing partners and with two local organizations that focused in providing 
services for EW: Cambodian Women for Peace and Development organization (CWPD) and Phnom Srey 
Organization for Development (PSOD). The CoE sites are hosted within operational districts with high 
burdens of HIV, in order to build the capacity of staff at public health facilities to develop and test high 
impact and cost-effective technical innovations. 
Around the same time, in 2013, the Boosted CoPCT standard operating procedure (SOP) was developed by 
NCHADS, which focused on identifying and testing KP to address their particular HIV service needs 
(NCHADS, 2013). In Boosted CoPCT, a guidance note was developed to direct the implementation of this 
SOP for EW, particularly ensuring the implementation of the core and expanded package of services. The 
core service package includes: improving case detection, avoiding new infections and reducing individual 
risks, strengthening referrals and linkages, care and treatment and, integration of sexual reproductive 
health and family planning. Flagship has incorporated the Boosted CoPCT into the innovations, and has 
worked closely with the CoE to modify, innovate, and test interventions for the HIV program using outreach 
(individual or group-level contacts with EW), SMARTgirl clubs and technology-based mediums (website, 
Facebook and voice4U). 
Under the Flagship project, there were 4,200 EW covered by the project at the start of implementation in 
2013 (USAID, 2013; 2014), and 8,7312 EW had been reached by the end of 2016. The two key direct 
implementers of the Flagship program, CWPD and PSOD provided the following core interventions:
A. Improving case detection:
• Strategic behavior communication (SBC): HIV awareness is mainly delivered via trained outreach 
workers (OW) in hotspots, entertainment establishments, private homes, public parks, and through 
the SMARTgirl club (see below). Under the branded “SMARTgirl” program, key messages were 
consulted with EW before the SBC materials were produced and OW were trained on the use of the 
printed materials before conducting education sessions. 
• Community-based HIV testing and counselling (CBHTC) is one of the major components of the 
SMARTgirl program in which EW could opt for testing at either in the SMARTgirl club or via 
outreach. The suggested screening test interval for HIV was biannually and quarterly for STI using a 
referral card. Prior to CBHTC, EW were screened for their risk levels using the interactive tablet￾based risk-screening tool in certain locations. 
• Sexually transmitted infections (STI) screening: The STI screening test for syphilis was performed by 
trained OW using finger prick, while the syndromic management and other STI check-up was 
provided through the NGO’s referral system to government or NGO health facilities. 
B. Avoiding new infections and reducing HIV risk:
 2 This figure was provided by Flagship 
5 
• Condoms and lubricants were made available for KP in hotspots, entertainment establishments and 
in the SMARTgirl clubs as guided by the national HIV prevention guidelines. A sample pack of 
condoms and lubricants was given to EW for free at the initial contact and then condoms and 
lubricants were promoted through peer-to-peer sale and outreach, street-based sellers near high￾risk venues and through condom vending machine to ensure sustainability of condom access. 
C. Strengthening referrals and linkages:
• Active referral and linkages to health and non-health services: EW were provided with linkages to 
HIV and STI testing, reproductive health services, TB diagnostic workup, antiretroviral therapy, 
vocational training, legal support services (including gender-based violence) and psychological 
services. Every month, the OW visited health facilities to collect referral cards in order to monitor 
the service utilization of EW. mHealth innovations (an interactive voice response system, websites, 
Facebook pages, and phone applications) were introduced with the purpose to link EW to all 
relevant services. 
• SMARTgirl club: Provided a secured and safe space for gathering and one-stop-shop services for 
HIV and STI information, screening and referrals for EW. At the club, EW were provided with SBC 
sessions and materials, referral supports to VCCT and STI services, edutainment, HIV screening 
(using finger prick testing), vocational training, as well as free condoms and lubricants as well as 
social marketing in selling condoms. 
D. Care and Treatment:
• Case management is embedded into the referral support for HIV positive EW who needed support 
for pre- and post- ART enrolment and preferred to be under the care cascade of the CoE.
E. Integration of Sexual Reproductive Health and Family Planning:
• EW were also supported with family planning whereby unmet family planning need screening, 
informed choice counseling, and referrals for methods were readily available at the SMARTgirl club 
through outreach activities and it was actively delivered through trained staff and OW. 
1.3. Flagship-TA GFATM Program Description 
While SMARTgirl innovations were tested in the three Flagship-supported CoE sites, four NGOs (CWPD, 
Partner for Development (PFD), KHEMARA and Save Incapacity Teenage (SIT)) received technical assistance 
from Flagship to implement the EW outreach program in four provinces: Banteay Meanchey, Battambang, 
Phnom Penh and Pursat. Those NGOs were supported by Global Fund in Cambodia through KHANA. 
Implementing NGOs provided activities to targeted groups using two mechanisms: drop-in-centers (DIC) 
and outreach work. The service package included a minimum HIV service package, such as HIV education 
sessions, free condom distribution or demonstration, HIV and syphilis testing through finger prick and 
referral to health services. (KHANA, 2016a). 
For implementing NGOs under the GFATM, the technical assistance from Flagship enabled them to 
implement the OW program (non-SMARTgirl service package or a generic HIV prevention program for EW) 
with key strategies listed below to provide core services under the leadership of KHANA. Since 2008, the 
SMARTgirl and the generic EW programs have made the core services available to about 22,000 EW 
through outreach, condom provision, HTC and referrals (to family planning service, HIV confirmatory tests, 
STI care and treatment, ART, TB and others) (KHANA, 2016b). 
A. Strategy 1: Identify pockets of populations with high and overlapping risk and vulnerability who are 
not yet in contact for the interventions.
• Assessing risk level: EW are encouraged to assess their HIV risks at least twice a year by using 
paper-based risk screening tool under the support of OW and staff. 
6 
• Identifying hard-to-reach key populations: Using the networks of EW and routine outreach activity, 
OW and staff conducted visits to settings where EW normally gather at appropriate time and 
locations. 
B. Strategy 2: Reach and provide services to EW
• Assigning UIC code for EW: EW were assigned UIC codes, then provided UIC cards for health 
services. The UIC cards link EW to relevant health services without fear of discrimination as their 
names and KP statuses were not disclosed publicly. The NGOs used UIC cards to monitor the types 
of services utilized by EW. 
• Conducting education sessions: OW provided HIV prevention education in their respective areas in 
either one-to-one or group education sessions. Using the traditional approach to HIV prevention, 
OW used the generic version of information, education and communication (IEC) materials to 
educate EW during the education session. During the session, OW also provided condom education 
and distributed condoms. Meanwhile, increasing gender-sensitive approach is integrated at the 
outreach activity level to ensure that barriers to services are minimal if EW prefer to use all health 
services. 
• Conducting bi-monthly meetings with EW: Issues affecting the program were solicited from EW 
advocates through regular discussion forum every other month. Suggestions from EW were often 
taken to high-level meetings by the NGOs. 
• Building HTC skills of lay counsellors: Under the assistance from the CoE, in partnership with 
Flagship and the national HTC focal points, the non-CoE sites provided annual HTC training to 
selected OW in order to officially qualify them to be lay counselors to provide community-based 
finger prick testing for HIV and syphilis to their respective peers. 
C. Strategy 3: Intensify interventions and services for maximum impact
• Performing/mobilizing support for HIV finger prick testing of EW: Trained OW and lay counsellors 
provided finger prick HIV testing in accordance to the national standard for HIV testing and 
counselling, including HIV education, pre-and post-test counselling and appropriate links to 
confirmatory tests at health facilities (for the reactive tests).
• Maintaining HIV service linkage: OW and NGO staff offered referral supports to EW for HIV and STI 
services (including HIV confirmatory test and Pre-ART/ART enrolment for HIV positive cases). The 
linkage service initiated were tracked via UIC for service utilization. 
• Maintaining DIC for EW: All core services were maintained in the drop-in-center. The services 
included HIV and other health educations, HIV and syphilis finger prick tests, free condoms and 
lubricants and a private space for relaxation as well as socialization with other peers. 
D. Strategy 4: Retain EW in services for maximum impact and improved health outcomes
• Advocating for ID poor: OW and NGO staff worked with health facility and Health Equity Fund (HEF) 
operators to ensure that eligible EW got ID poor card and were enrolled into HEF. Other enabling 
environment activities for EW were also implemented by the NGOs such as meeting forums, 
campaigns and other events.
1.4. Non- Flagship GFATM Program Description 
Non- Flagship GFATM EW Programs follow the guidelines set out under Boosted COPCT to provide services. 
Indeed, the bulk of services was similar to those provided by the Flagship CoE and Flagship-TA sites, with 
some notable exceptions. The various key services provided by each of the three program types is 
summarized in Table 2.
7 
Table 2 Core services of Flagship CoE, Flagship TA sites, and Non-Flagship sites
Services Provided 
Implementation period (2015-2016)
Remark Flagship 
CoE
Flagship 
TA 
Non￾Flagship 
I. Service delivery from NGOs to beneficiaries
9.1. Improving case detection
Education prevention message (HIV and others related 
health contents) 
X X X
Risk screening: paper-based X ?
Risk screening: tablet-based X X
STI screening or syndromic management at the club 
Finger prick testing and counselling: HIV/syphilis X X X
Working with key informants (Mekars) and beauty salons to 
reach unreached/hard-to-reach KPs
X X X Non-CoE: Mainly Mekars 
help facilitate 
communication between 
OW and EW for HTC and 
outreach
1.2. Avoiding new infections and reducing HIV risk
Condom and lubricant availability: free (only demo) X X X
Condom and lubricant availability: sale X Non-CoE: only happened 
until 2014
1.3. Strengthening referrals and linkages
1.3.1. Referral to health services
Providing referral support to EW for STI services (STDs other 
than HIV/syphilis) 
X X
Integrating case management initiative among EW who 
need confirmatory test and enrollment for relevant 
treatment services
X X X
Syphilis: If negative result, making appointment for the next 
test in a 3-month period. IF positive result, following-up the 
treatment 
X X X
HIV: Supporting enrollment at pre-ART for HIV positive cases X X X
1.4. Care and treatment support
Following-up visit during 12 months (i.e. treatment 
adherence, OIs)
X X X
1.5. Integration of SRH and FP 
Supporting EW to access contraception/FP at health facility, 
family health clinics, and NGO clinics
X X X
1.6. SMARTgirl Club
Supporting SMARTgirl club for EW to provide safe space for 
informal group discussions, information/education, 
counseling, HIV testing, referrals for HIV, SRH, and TB 
services, and IEC materials 
X X
II. Non-service delivery: program components
2.1. Improving case detection
Training and routine coaching: SBC and finger prick testing X X X
2.2. Avoiding new infections and reducing HIV risk
Condom and lubricant social marketing: maintaining a 
functional peer sale structure, street vendors, and vending 
machines in hotspots 
X X X Non-CoE-TA: Limited to 
relevant staff and OW 
joining Social marketing 
training; implementation of 
these activities at Flagship￾TA and non-TA sites is non￾existent.
8 
Table 2 Core services of Flagship CoE, Flagship TA sites, and Non-Flagship sites (Continued)
Services Provided 
Implementation period (2015-2016)
Remark Flagship 
CoE
Flagship 
TA 
Non￾Flagship 
Non-condom and lubricant social marketing: maintaining
correct messages, demonstration, and free distribution (only 
demo) 
X X X
2.3. Strengthening referrals and linkages
2.3.1. Referral to health services
Maintaining health linkage services: meeting, capacity 
building and joint monitoring visit with service providers
X
Integrating other innovative approaches: Mekars (sex 
brokers), risk tracing snowball, mHealth (Website, Facebook 
and Voice4U, active case management, partner tracing and 
community mobilization, coordination and leadership (i.e. 
networking, training), staff and outreach workers that wear 
unique uniforms 
X X X * Uniform is used among 
TA sites but not for non-TA 
sites.
* CoE: PDI is not 
implemented yet. And risk 
tracing snowball is 
implemented at Chhouk 
Sar Clinic 1 & 2 only.
2.3.2. Non-health services
Reducing stigma and discrimination: meeting, event, 
campaigns and capacity building
X
Supporting SMARTgirl club (or DIC) for EW to provide safe 
space for informal group discussions, 
information/education, counseling, HIV testing, referrals for 
HIV, SRH, and TB services, and IEC materials 
X X X
2.4. Care and treatment support
Linking HIV positive cases to community and home based 
care services
X X X
Ensuring quality of treatment in all ARV sites through 
meeting 
X X X
Reducing stigma and discrimination at hospital and 
community settings: meeting
X X X
Advocating for ID poor X X X
2.5. Integration of SRH and FP 
Conducting training to FP/HIV counselors, sale officers, FP 
government providers on Integration of SRH and FP
X X Non-CoE-TA: Only field 
staff from TA sites 
participate in the training.
2.6. SMARTgirl Club
Providing TA on SMARTgirl brand to CoE and non-CoE X X Non-CoE-TA: TA includes 
training on SBC, HTC, social 
condom marketing, FP, 
case management 
1.5. Comparison of the Flagship CoE SMARTgirl Program with Flagship-TA EW programs 
While table 2 describes the overall program descriptions, eight EW outreach sites were surveyed as part of 
the background work for this evaluation, including 3 CoE sites, 3 Flagship TA sites, and 2 Non-Flagship sites. 
See Table 3. There was a wide diversity of activities that reportedly were conducted at these locations, and 
the start dates of these activities also varied among program site types, as well as among sites. It was, 
observed, however, that CoE sites generally offered a wider range of activities and innovations. 
9 
Table 3 Timing of initiation of key services at selected CoE, Flagship TA, and Non-Flagship 
sites
Flagship CoE Flagship TA Non-Flagship
Innovation/Activity 
Chaktomuk
Kampong 
Cham/Kampong 
Siem
Siem Reap
Battambang
Sery Sophoan
Sampov Meas
Kampong Thum
Preah Sihanouk
mHealth 2015 2015 2015 2016 2015 2016 2015 2016
HTC by using finger prick for HIV/STI 
test
2012 2013 2013 2016 2015 2013 2016
UIC 2014 2014 2014 2015 2015 2014 2014 2015
IEC materials 2013 2014 2016 2015
Mekar 2014 2008
Tablet-based risk screening 2016 2015 2016
Case management 2013 2015 2015 2015
Social Marketing 2013 2013
Case profile 2015 2016 2015
Club 2014
Family Planning 2014 2008
SBC 2013 2009
DHIS2 2012 2014
GIS Mapping 2013 2016
2. Rationale and Scope
No systematic impact evaluation of the SMARTgirl or the other outreach programs for EW have been 
undertaken. Information from such investigations will help the national program, as well as donors to more 
effectively and efficiently shape programs and deploy resources. This evaluation measured the uptake of 
HIV tests, condom use, STI screening and treatment, and referrals to health services. CoE sites that 
provided HIV programs for EW under USAID HIV Flagship project, Non-Flagship sites under the Global Fund 
project through KHANA, and locations without EW intervention program were covered in this evaluation in 
order to inform the national HIV program. Comparisons of outcomes across these location types were 
performed. In addition, the evaluation looked at the cost-effectiveness of the program by comparing the 
unit costs of related outcomes between the Flagship and Non-Flagship sites. Perspectives of the clients 
regarding the quality of service deliveries was also examined in this evaluation.
3. Evaluation Questions
The key questions that guided this evaluation were: 
1. What was the impact of the SMARTgirl programs for EW, and how did the impact compare among 
Flagship CoE, Flagship-TA and Non-Flagship programs, on the uptake of HIV tests, condom use, STI 
screening and treatment, and referrals to health services?
2. What were the experiences of EWs exposed to these different programs in terms of:
a. intensity of exposure (frequency of services), 
b. types of services/products/other benefits received (HTC, STI screening and treatment, 
condoms, and referrals to health services), 
c. quality of services received and,
d. exposure to, and benefits gained from, SMARTgirl SBC material and social media?
3. What was the reach of the three different program types among non-venue based EW (hard-to￾reach/hidden) and what was the success in HTC uptake?
4. What was the cost-effectiveness of the SMARTgirl program implemented through the Flagship CoE 
compared to Flagship-TA SMARTgirl programs and Non-Flagship EW programs?
10
4. Evaluation Objectives 
The overall objective of this evaluation was to evaluate the impact and cost effectiveness of the SMARTgirl 
program as implemented through the USAID Flagship project. The ultimate and immediate objectives are 
illustrated below: 
• Ultimate objective
The ultimate objective of this evaluation was to provide rigorous evidence regarding the effects of the 
SMARTgirl program. The evaluation aimed to provide an accurate understanding of the program outcomes 
to inform the national HIV prevention program in planning, scale up and policy options of the SMARTgirl 
program in the Cambodia. 
• Immediate objectives
The specific objectives of this evaluation were: 
a. To measure the effects of SMARTgirl program on the uptake of HIV tests, condom use, STI 
screening and treatment, and referral to health services. 
b. To describe the service utilization among EW and their perspectives regarding the 
attractiveness of and satisfaction in the SMARTgirl program (including SBC materials, social 
media, HTC services, STI screening and treatment, and referrals to health services).
c. To examine the effectiveness of the SBC material, ICT and condom distribution on knowledge 
and behavior regarding HIV prevention and utilization of health services. 
d. To examine the effects of SMARTgirl in reducing stigma and discrimination and barriers to 
services. 
e. To evaluate the ability of the SMARTgirl program in reaching and providing services to hard-to￾reach/hidden EWs; and
f. To estimate the cost-effectiveness of the SMARTgirl program implemented through Flagship 
CoE compared to Flagship TA SMARTgirl programs and Non-Flagship EW programs. 
5. Evaluation Design
Given the absence of baseline data for the control and intervention groups before the inception of the 
program, a Static Group Comparison or Posttest-Only with Nonequivalent Groups (a quasi-experimental 
design), was the most appropriate to address the evaluation questions. In this design, the intervention 
group was measured and compared to a control group. The SMARTgirl CoE implementation sites under the 
USAID HIV Flagship project and Global Fund supported sites through KHANA (Flagship-TA and Non-Flagship
sites) were treated as intervention group. The control group included the locations without EW 
intervention programs. See figure 1.
Given the non-randomization of the subjects to the intervention or control group, there were potential 
threats to internal validity with this type of design. Differences in outcomes such as uptake of HIV tests, 
condom use, and STI screening, might be attributed to the differences in the characteristics of the 
entertainment workers in the control group and intervention. Therefore, econometric methods were 
applied to control confounding variables and isolate the effect of the HIV program in EW. In order words, 
the impact of the SMARTgirl program for EW were identified as the differences in the above described 
outcomes between intervention and control groups. 
The levels of exposure to the SMARTgirl program might have had different effects on outcomes of interest. 
For example, the probability of HIV test uptake among EW exposed to only outreach workers might be 
lower than those EW reached by outreach workers, social media and via printed education materials. For 
this scenario, a within-group comparison, e.g. CoE versus Flagship TA, was made and other extraneous 
variables were controlled by econometric methods. Also, a program exposure index was constructed and 
categorized into different levels of exposure. Then comparison between those levels was made. 
11
Figure 1 Posttest-only with nonequivalent groups
Cost allocation was included in this comparative study in order to estimate cost-effectiveness, as 
determined by the unit costs of the innovation program (CoE) relative to the unit costs of the program 
implemented in non-CoE locations. The unit costs included the costs per HIV test and cost per new HIV case 
identified. Cost allocation exercise included the process of identifying, aggregating, and assigning costs to 
activities. This study design was retrospective in nature, leveraging data from past financial records, 
budgets, invoices, inventories, contracts, etc. 
6. Method of Evaluation 
6.1. Method of Data Collection from EW
A cross sectional survey using structured interviews was conducted with EW in the coverage of Flagship CoE 
under the USAID HIV Flagship project, Flagship TA areas under Global Fund, Non-Flagship Global Fund 
areas, and areas with no known EW intervention program. Retrospective data regarding behavior and 
perception of EW about the service delivery was collected. 
6.2. Reference Period of Costing Data Collection
The costing data were collected from the reference period of 01 October 2015 to 30 September 2016. This 
period was a reference period for the comparison of costs per unit between Flagship CoE, Flagship TA, and 
Non-Flagship areas. Program data regarding the number of HIV tests and HIV positive yield from the period 
above was used to compute unit cost using cost allocation. 
7. Sampling Procedures
7.1. Sample Size 
Key related variables from the evaluation of SMARTgirl family planning and HIV integration project in 2015 
were available to perform comparisons between CoE and non-CoE areas. These variables were used to 
provide basic understanding regarding sample size requirement for this evaluation. Percentage distribution 
of these variables for CoE and non-CoE areas is illustrated in Table 4. 
Post-intervention data 
collection point Sites with no EW program
Post-intervention data 
collection point
Non-Flagship Sites
SMARTgirl Program
CoE sites & TA sites
Post-intervention data 
collection point
Non-random assignment of 
EWs to groups
Intervention 
group
Control group
12
Table 4 Percentage distribution of key related variables for sample size computation
No. Variable Non-CoE CoE
1 Had HIV test in the past 12 months 83% 85% 
2 Had STI screening in the past 12 months 53% 55% 
3 Always use condom with boyfriend 36% 33% 
4 Always use condom with clients 33% 24% 
The results of computation of sample size using the power twoproportions command in STATA are 
demonstrated in Table 5, using Pearson's chi-squared test.
Table 5 Estimated sample sizes for a two-sample proportions test 
No. Variable P1 P2 δ α 1 – β Est. N
1 Had HIV test in the past 12 months 0.8300 0.8500 0.0200 0.0500 0.8000 10548
2 Had STI screening in the past 12 
months 
0.5300 0.5500 0.0200 0.0500 0.8000 19496
3 Always use condom with boyfriend 0.3600 0.3300 -0.0300 0.0500 0.8000 7882
4 Always use condom with clients 0.3300 0.2400 -0.0900 0.0500 0.8000 788
p2: proportion for corresponding variable for CoE; p1: proportion of corresponding variable for Non-CoE; N: estimated 
total sample size; n= estimated sample size per group; alpha (α): significance level; β: type II error probability; 1 – β: 
power; delta (δ): effect size; Null hypothesis (Ho): p2 = p1 versus Alternative hypothesis (Ha): p2! = p1 using Pearson's 
chi-squared test
The computation of sample size corresponding to the difference of key variables between CoE and non-CoE 
provided some options for the determination of appropriate sample size for this evaluation. Given the 
timeframe and anticipated complexities of recruiting subjects for this study as well as the available 
resources, the estimated sample size for this evaluation was 788. 
This estimate was based on the formula for simple random sampling. Since respondent-driven sampling 
strategy was used to recruit sample for this evaluation, this sample size needed to be adjusted for variance. 
In theory, the distribution of estimates from respondent-driven sampling is more variable than the 
estimates from simple random sampling. It was adjusted by multiplying with a design effect. For the 
context of this study, the design effect was set at 1.5. The sample size based on the above estimation was 
1,182. However, the final sample size was increased to 1,300 EW in order that statistical assumptions in 
statistical analysis techniques could be ensured. The above sample size was allocated to 300 EW for 
Flagship CoE areas, 300 EW for Flagship TA areas, 300 EW for Non-Flagship areas, and 400 for areas without 
any intervention program for EW (See Table 6). This allocation was to ensure that number of observations 
would be sufficient for comparison between these three locations, and the application of advanced 
statistical model or econometric methods to measure the program impact. 
13
Table 6 Actual sample size
Province IP OD GF Flagship Total Intervention Sample 
Size
# EW # EW # EW
Phnom Penh CWPD Chaktomuk 4,685 1,150 5,835 Flagship CoE 100
Kampong 
Cham
PSOD Kampong 
Cham
- 640 640 Flagship CoE 100
Siem Reap CWPD Siem Reap 2,379 2,379 Flagship CoE 100
Banteay 
Meanchey 
PFD Poi Pet 1,102 - 1,102 Flagship TA 100
Battambang CWPD Battambang 1,739 - 1,739 Flagship TA 100
Pursat PFD Sampov Meas 373 - 373 Flagship TA 100
Kampong 
Thom
CWPD Kampong 
Thom
500 - 500 Non-Flagship 100
Preah 
Sihanouk
KHEMARA Preah Sihanouk 477 477 Non-Flagship 100
Pailin CWPD Pailin 367 367 Non-Flagship 100
Battambang - Moung - - 150 No program 66
Kep - Kep - - 192 No program 126
Kampot - Kampong Trach - - 50 No program 25
Kampong 
Thom
- Baray - - 100 No program 70
- Stoung - - 75 No program 57
Preah Vihear - Sra Em - - 76 No program 56
Total 9,243 4,169 14,055 1300 
The intent of this sample size computation was to ensure sufficient sample size for the comparison of 
outcomes between groups, but not to generalize to the whole population. The statistical analysis was not 
limited to the comparison of proportions between groups. More advanced statistical modeling was used to 
control some potential confounders. But there was no specific formula to calculate sample size for each 
statistical model. Based on rules of thumb, sample size requirement for each statistical modeling was a 
function of the predictors in the model. The more predictor variables in the model, the larger sample size 
was required to allow for variation. The number of observations for each predictor variable was required to 
have at least 10 to 15 cases in order that variance across variables in the model could be ensured. These 
rules, however, oversimplified the issues of sample size requirement. In fact, the sample size required 
depended on the effect size (how well the predictors predicted the outcome), how much statistical power 
(the probability that it would reject a false null hypothesis) we wanted to detect these effects, and what we 
were testing (the significance of the coefficients or the significance of the model overall). Theoretically, the 
minimum sample size required to achieve a high level of statistical power (.8, with large effect size (R2 =.26) 
expected, and with up to 20 predictor variables in the model, then a sample size of 1043 would suffice 
(Field, 2013) . Therefore, the overall sample size of 1300 estimated above would suffice for both
comparison of proportions and statistical/econometric modeling. 
14
7.2. Sampling Strategy 
To better reach hidden, hard-to-reach, and unreached EW, respondent driven sampling (RDS) was the most 
appropriate strategy to select the sample for this evaluation study. This sampling strategy was also 
appropriate in the context where there was no sampling frame, and where the definite number of 
population had not been known. 
RDS initiated recruitment with purposively (non-randomly) selected study participants, known as seeds, but 
it would result in a reasonably randomly drawn sample. The initial sample would not be biased by the 
purposive selection of seeds if the equilibrium (an indication that the final sample is not biased) was 
reached. Equilibrium was the point at which the sample characteristics no longer change no matter how 
many more individuals enter into the sample. Studies using RDS have demonstrated that four to six waves 
were usually needed to reach equilibrium and the tendency for in-group affiliation could be reduced as 
well. Moreover, selecting a diverse set of initial seeds could speed the approach to equilibrium. The 
increased number of waves allowed the target sample size to be attained and ensured a broad array of 
participants to have the opportunity to recruit their peers (Grazina, 2008). 
The initial seeds were selected from the existing networks of previous studies, and by field research 
managers based on initial visits. Outreach workers facilitated the recruitment process for the initial seeds 
when necessary. Each of these seeds recruited two EW, then each of the two EW continued recruiting two 
more EW. Sampling ended when the target sample size of 1300 was reached (See Table 7). 
Table 7 Results of recruitment
Province OD
Sample 
Size
Initial 
Seed
#Wave
Ineligib
le
Rejecte
d 
coupon 
f 
Rejecte
d 
coupon 
f 
Refusal
# of 
Dead 
seeds
Phnom Penh Chaktomuk 100 1 9 16 1 33 0 34
Kampong Cham Kampong Cham 100 1 12 7 0 1 0 0
Siem Reap Siem Reap 100 2 7-11 1 0 0 0 0
Banteay 
Meanchey PoiPet 100 3 1-20 20 0 13 0 15 
Battambang Battambong 100 2 7 6 1 0 0 29
Pursat Sampov Meas 100 1 15 18 0 0 1 0
Kampong Thom Kampong Thom 100 3 2-14 15 1 11 0 15
Preah Sihanouk Preah Sihanouk 100 2 4-9 0 0 0 0 0
Pailin Pailin 100 4 2-9 15 0 8 0 24
Battambang Moung 66 1 10 8 0 0 0 31
Kep Kep 126 4 2-20 29 13 53 0 66
Kampot Kampong Trach 25 1 8 9 0 5 0 9
Kampong Thom Baray 70 1 12 12 4 19 1 25
Stung 57 2 3-8 0 0 0 0 0
Preah Vihear SraEm 56 7 1-5 20 0 16 0 25
Total 1300 35 176 20 159 2 273
Though the recruitment chain planned for this study was started with initial seeds of 1 or 2 EW 
corresponding to the sample size allocated to each site, the initial seeds were varied in the field (Tables 8 
and 9) depending on the speed of recruitment, survival rate, and characteristics of the study site. For 
example, for Sra Em OD in Preah Vihear province, only one initial seed was required to complete the 
recruitment chain for 40 EW, but in real fieldwork, 7 initial seeds were used for the recruitment due to 
extremely small network sizes and their working places were very far from each other. The Social network 
sizes are shown in figure 2.
15
Table 8 Types of initial seeds 
Province Seed 
Number EW Type
Chaktomuk Seed 1 Karaoke worker
Kampong Cham Seed 1 Karaoke worker
Siem Reap Seed 1 Restaurant
Seed 2 Massage parlor worker
Poi Pet
Seed 1 Massage parlor worker
Seed 2 Restaurant
Seed 3 Restaurant
Battambang Seed 1 Freelance sex worker (park, street, phone…)
Seed 2 Karaoke worker
Sampov Meas Seed 1 Karaoke worker
Kampong Thom
Seed 1 Massage parlor worker
Seed 2 Karaoke worker
Seed 3 Freelance sex worker (park, street, phone…) 
Preah Sihanouk Seed 1 Restaurant 
Seed 2 Residential sex worker 
Pailin
Seed 1 Restaurant
Seed 2 Residential sex worker
Seed 3 Karaoke worker
Seed 4 Café/Restaurant
Moung Seed 1 Karaoke worker
Kep
Seed 1 Karaoke worker down stair and guest house up stair 
Seed 2 Karaoke worker
Seed 3 Freelance base in beach
Seed 4 Karaoke worker
Kampong Trach Seed 1 Karaoke worker
Baray Seed 1 Karaoke worker
Stoung Seed 1 Karaoke worker
Seed 2 Karaoke worker
Sra Em
Seed 1 Beer garden worker
Seed 2 Karaoke worker
Seed 3 Karaoke worker
Seed 4 Massage parlor worker
Seed 5 Karaoke worker
Seed 6 Restaurant
Seed 7 Karaoke worker
16
Table 9 RDS recruitment numbers and probabilities
Type of EW
Freela
nce 
sex 
Reside
ntial
sex 
worke
Massa
ge 
parlor
Beer 
garden
KTV
Cafe/ 
restau
rant
Other
Number of participants recruited
Freelance sex worker 12 2 3 0 15 1 5
Residential sex worker 1 35 0 0 9 0 4
Massage parlor 2 1 103 3 16 2 2
Beer garden 0 0 3 30 23 7 3
KTV 10 3 8 21 673 19 37
Cafe/restaurant 3 0 5 12 11 77 3
Other 8 1 1 0 34 7 41
Total 36 42 123 66 781 113 95
Recruitment Probabilities
Freelance sex worker 0.316 0.053 0.079 0 0.395 0.026 0.132
Residential sex worker 0.02 0.714 0 0 0.184 0 0.082
Massage parlor 0.016 0.008 0.798 0.023 0.124 0.016 0.016
Beer garden 0 0 0.045 0.455 0.348 0.106 0.045
KTV 0.013 0.004 0.01 0.027 0.873 0.025 0.048
Cafe/restaurant 0.027 0 0.045 0.108 0.099 0.694 0.027
Other 0.087 0.011 0.011 0 0.37 0.076 0.446
Figure 2 Average social network size by type of EW
4.4
5.8
4.8
7.4
6.2
5.4
5.0
Freelance sex
worker
Venue-based sex
worker
Massager parlor Beer garden ktv Cafe/restaurant Other
17
7.3. Recruitment Process
Outreach workers from NGOs working directly with EW were employed to help facilitate the recruitment
process. Initial seeds were given 2 coupons each to recruit other candidates from the target populations by 
giving them a coupon with a unique serial number used for attributing the referral to the seed. The 
recruited candidates, in turn, became seeds after participating in the study. The new seeds were given 
coupons that they would use to refer their peers. To motivate seeds and new recruits to participate in the 
study, an incentive mechanism was used. Seeds were eligible for additional financial compensation if 
individuals they referred were eligible and accepted to participate in the study. Eligible new recruits 
received a financial compensation if they were qualified and accepted to participate in the study. 
Participants that did not meet the inclusion criteria could not receive any incentive. See figure 3.
Figure 3 RDS recruitment process
7.4. Eligibility 
EW were assessed for eligibility by the field research team. The assessment ensured that the survey 
participants met eligibility criteria and gave consent. Eligible participants received an explanation of the 
study’s purpose and the nature of the questions to be asked. The field research team reviewed the consent 
form with the participant. If the participant acknowledged full understanding of her participation in the 
study and agreed to participate, she would be enrolled. Below were the inclusion criteria for participants in 
this evaluation:
• Exchanged sex for money or other items or services in the last 3 months;
• Female aged 18 or older;
• Khmer speaker;
• Able to adequately grant informed consent;
• Working in the study OD coverage;
• First time enrolled in the study; and
• A valid referral coupon from a previous study participant (except seeds).
Additional candidate(s) 
recruited
Field researcher 
screened candidate
If not eligible, 
completed Ineligibility Form 
If potential 
participant refused to 
participate, completed 
Refusal Form
Explained the study & obtain 
consent 
• Initiate Check List Form 
Conducted interview
Explained RDS 
recruitment process
Gave out 2 recruitment coupons
Paid primary 
incentive
Completed Financial Form
Paid secondary 
incentive
If rejected
Used Coupon Rejecter 
Questionnaire
18
152 participants of the total 1456 (10.4%) recruited were not eligible because they had not had sex with a 
male partner in exchange for money or goods in the previous 3 months. 10 EW recruited were under age 
18 years. These EW were not included in the study. 
7.5. Place and Time of Interviews
Generally, the locations of interviews were decided with consultation with respondents in compliance with 
standard requirements of appropriate interview locations. The places of interviews varied from site to site 
corresponding to the characteristics of their living space and working venue. For example, most EW in 
Phnom Penh preferred to have a meeting at their rented home or room for the interview. But 35% of EW in 
Siem Reap met with field researchers at a pagoda for the interviews because their rented room/house and 
working place were not convenient for discussion. 
The time for conducting interviews with EW working at Karaoke and beer garden was commonly from 10 
AM to 4 PM. The schedule needed to be at some time after they woke up and at some point before they 
went to work in late afternoon. For EW working at massage parlor and freelance EW, the interviews were 
done at any time they were available. 
Before the interview started, field researchers had spent around 10-15 minutes to introduce themselves, 
objectives of the study, get informed consent, warm up, build rapport, and encourage the EW to participate 
in the study. Overall, about 40 to 60 minutes were spent for each interview session. In addition to this, the 
recruitment process was consumed about 10-15 minutes in order to ensure that the recruiters understood 
the recruitment process and to reduce the probability of failure in recruiting their qualified peers. 
7. Evaluation Team 
The evaluation was carried out by a team possessing academic backgrounds in health and social science, 
capacities, skills and experience in research and impact evaluation design, management, and analysis. The 
team consisted of the following people: 
• Dr. Christian Pitter, MD MPH, Chief of Party, USAID HIV Innovate and Evaluate Project, 
University Research Co. LLC.
• KHUN Sithon, Ph.D. (Demography), M.A. (Population & Reproductive Health Research), B.A. 
(Sociology), Director of Research, USAID HIV Innovate and Evaluate Project, University Research 
Co. LLC.
• Mrs. Oeng Sothary, B.A. (Sociology), Research Project Officer, USAID HIV Innovate and Evaluate 
Project, University Research Co. LLC.
Data collection was carried out by a pool of field researchers possessing bachelor degrees in social sciences, 
and were well equipped with knowledge, skills, and extensive experience in structured interviewing and 
data collection techniques with EW. They had been engaged in research and evaluation with USAID HIV 
Innovate & Evaluate Project in addition to their previous fieldwork experience with other organizations. 
These field researchers were a central part to this evaluation. Moreover, the quality of data collected was 
managed by a qualified data analyst with experience and skills in data management using tablet computers. 
Data collection was conducted by Mao Sosengphyrun, Chhor Lyda, Chhuoy Socheat, Loeurng Samoeun, 
Dieb Sreyroth, Tap Bopea, Phorn Somaly, Srun Piroth, Em Phal Nida, Hor Danet, Khen Sophal, Iv Kham 
Prasith, Eam Socheata, Ham Leakhena, Inn Sieklim, Pech Chanra, Kaing Sonai, Douk Satya, Ouch Chanriith, 
San Kimtin, and Ke Sreypech.
This external evaluation was carried out by the USAID HIV Innovate and Evaluate project, a project that is 
independent of the intervention project and the national HIV program in Cambodia. All members of the 
evaluation team were independent. All field researchers were trained to be independent and followed 
standard processes of independent evaluation. 
19
8. Data Collection 
8.1. Data Collection Team
Since cost allocation was integrated in this program evaluation, two types of data collection team were 
formulated: a social research team and a cost allocation team. 
The social research team was led by the HIEP Director of Research and coordinated by two research 
operation managers, Mr. Song Koeun and Mr. Pho Yaty, and a research project officer. Five field research 
teams were formed to oversee and ensure quality of fieldwork for data collection in field. Each team was 
composed of a field research manager and 2-3 female field researchers. Overall, there were 6 field 
researcher managers, 15 field researchers and 2 research operation managers carrying out data collection 
for this evaluation. The social research team collected data from EW. 
A well-constructed conceptual and methodological costing analysis required a great deal of technical 
knowledge and skill in accounting and budgeting systems. The cost allocation team composed of 2 external 
consultants with accounting background to collect data under the coordination and support from the 
finance and administration director. The team was equipped with a standardized method for data 
collection to ensure uniformity in assigning costs. 
8.2. Training for Data Collection
The social data collection team was provided with a six-day training, from 21 to 28 December 2016, 
focusing on the study background, evaluation protocol including methodology and sampling strategies, 
professional interviewing techniques and fieldwork strategies, questionnaire and tablet survey application, 
consent forms, data collection, ethical requirements in independent evaluation, and the intervention 
program. Technical personnel implementing partners were invited to provide orientation about the 
implementation of the SMARTgirl program. The objective of the training was to equip the data collection 
team with knowledge in intervention project, evaluation methodology, and concept and skill in quantitative 
interviewing techniques and data collection management. Mock interviews were carried out among the 
research team. 
The cost allocation team was trained by the HIEP Administration and Finance Director in a three-day 
training course. A pilot test of this cost allocation was carried out. Refinement of the tools and process was 
made based on the results of the pilot. 
8.3. Fieldwork Management
After being recruited by the seeds, the participants were contacted by field researchers to make 
appointments for interviews. The interviews were done at anyplace agreed by the participants as 
convenient for conversation, while ensuring safety for field researchers. To further ensure researchers’ 
safety, female field researchers were accompanied by a male field research manager to interview with each 
participant when necessary but he was not allowed to stay close to hear the interviews. Generally the male 
field researcher stayed far from the place of interview but within a distance that he could see the female 
field researcher. In some cases, field researchers informed the field research manager about locations of 
the appointments and the schedule of daily interviews. In addition, the tablet application enabled the 
monitoring of the timing and location of interviews. Both options were used to ensure that interview 
actually happened in the field and to facilitate the efficiency and effectiveness of the interviews. 
Monitoring data collection was central to ensuring data quality. Two research operation mangers 
monitored data collection activities on a daily basis. A facilitation model was used in order to build a strong 
research team with a common goal of providing good quality data. The key role of the senior research team 
was to facilitate the jobs of field researchers who carried out data collection activities. Daily monitoring 
allowed to oversee the following issues: 
• The completion of interview quota;
• The number of participants completed the interview, but refused to become recruiters;
• The number of individuals who refused to participate in the interview;
• The number of participants who were found to be ineligible for the interview; and 
20
• Pattern of refusal (reasons why peers did not want to be interviewed or were not willing to recruit 
their peers). 
Based on the monitoring information, the research management team assessed how recruiters were 
administering study information and the consent form, how the field team and recruiters checked 
eligibility, especially to ensure that it was being determined correctly, and how the field researchers 
provided recruiter training. Moreover, a fieldwork operation manual for data collection was designed, and 
this guided the processes of data collection. The social media platform WhatsApp was intensively used by 
the evaluation team, creating a platform of interactive communication among the field researchers across 
the study areas and the management team at the central office which allowed the provision of 
backstopping support from the management team at central office. 
8.4. Data Collection Schedule
The data collection schedule was divided into two rounds in order that the quality of data could be 
ensured. The first round of data collection was from 9 to 26 January 2017, and the second round was 
from 30 January to 17 February. Lessons learned from the first round were used to improve the quality 
of fieldwork for data collection in the second round. 
Table 10 Data collection schedule 
No OD
Date
#Participants Started Ended # Days
First round of data collection
1 Chaktomuk 9 Jan 2017 25 Jan 2017 17 100
2 Baray 9 Jan 2017 18 Jan 2017 10 70
3 Kampong Cham 9 Jan 2017 21 Jan 2017 13 100
4 Moung 9 Jan 2017 18 Jan 2017 10 66
5 Kampong Trach 9 Jan 2017 15 Jan 2017 7 25
6 Sampov meas 9 Jan 2017 26 Jan 2017 18 100
7 Kep 16 Jan 2017 26 Jan 2017 11 58
8 Kampong Thom 19 Jan 2017 26 Jan 2017 8 57
9 Battambong 19 Jan 2017 26 Jan 2017 8 66
10 Siem Reap 22 Jan 2017 26 Jan 2017 5 41
Total 683
Second round of data collection
11 Kep 30 Jan 2017 13 Feb 207 15 68
12 Kampong Thom 30 Jan 2017 5 Feb 2017 7 43
13 Battambong 30 Jan 2017 3 Feb 2017 5 34
14 Siem Reap 30 Jan 2017 9 Feb 2017 11 59
15 Preah Sihanouk 30 Jan 2017 15 Feb 2017 17 100
16 Pailin 30 Jan 2017 12 Feb 2017 14 100
17 Poi Pet 4 Feb 2017 15 Feb 2017 12 100
18 Stoung 10 Feb 2017 17 Feb 2017 8 57
19 Sra Em 5 Feb 2017 14 Feb 2017 9 56
Total 683
8.5. Multiplicity
Duplication of respondents is a concern when using RDS. To avoid using complicated tracking systems, such 
as biometrics, that could lead to mistrust among this hard-to-reach population, the mobile phone number, 
working venue, age, duration living in current location, and other related individual characteristics were 
used to prevent people attempting to defraud the survey. Research team members were required to work 
21
together closely, especially to carry out daily verification. Duplicate questionnaires were excluded from the 
sample. 
8.6. Instruments
Two types of instruments were designed to collect data from EW and costing data. To collect data from EW, 
a tablet assisted personal interviewing application, SurveyToGo for Android was purchased and used in this 
study. The questionnaire and screening forms were designed in SurveyToGo Studio. 
The following four methods were used to evaluate draft survey questions developed by research team in 
order that content standards, cognitive standards, and usability of the questions were ensured and quality 
of measurement (validity and reliability) were enhanced. The content standards were related to whether 
the questions asked about the right things. Cognitive standards referred to the issues regarding whether 
respondents understood the questions consistently, had the information required to answer questions, and 
were willing and able to formulate answers to the questions; and, usability standards were in place to 
ensure the interviewer and respondents were able to complete the questionnaire easily as they were 
intended. 
• Technical team reviews: the subject matter team reviewed the questions to assess whether 
their content was appropriate for measuring the intended concepts.
• Semi-structured interviews with EW population were organized to explore what they knew 
about the issues that the questionnaire would cover, how they thought about the issues, and 
what terms they used in talking about them. 
• Cognitive interviews: the interviewers administered draft questions in individual interviews, 
probed to learn how the respondents understood the questions, attempted to learn how they 
formulated their answers. 
• Pilot tests: each field researchers conducted at least 3-5 interviews with EW population using 
RDS strategy. The interviews were recorded with permission from respondents. The recordings 
were reviewed by the interviewers and research team in order to identify questions that were 
difficult to read as worded or hard for respondents to answer. Then debriefing with 
interviewers was held to gain their insights into the problems they had in asking the questions 
or those the respondents had in answering. Data from field pretests were analyzed to identify 
signs of trouble. 
Since field researchers are central to this evaluation, necessary efforts and investments were 
made in order to ensure effective data collection with the ultimate goal of producing high 
quality of data. Pilot test interviews with EW were organized twice in order to ensure quality of 
interviews. The data collection tool was refined based on the results of the pilot tests. 
Additional tools had been designed in order to support the process and management of data collection. 
These tools include: 
• Screening questionnaire: used to select EW to take part in the study based on the inclusion criteria. 
• Consent form: explained the purpose, requirements, risks, and benefits of the study and asks the 
participant to acknowledge informed consent.
• Check list form: recorded progress of participants from the first contact for interview until primary 
incentive payment
• Field incident form: reported unexpected occurrences in the field.
• Coupon tracking form: monitored the flow of participants, distribution of coupons and completion 
of study requirements by participants,
• Non-eligibility form: summarized why participants are ineligible at screening.
• Refusal form: summarized why participants refuse to participate at screening.
• Financial reporting form: tracked the payment of primary and secondary incentives.
• Inclusion criteria card and coupon: Guided recruits’ selection of their peers to participate in the 
study.
• HEART ranking tool: facilitated respondents’ replies to scalar questions.
22
To collect costing data, a financial matrix was developed to capture expenditure data from financial 
records. The costs incurred by the supporting and coordination activities carried out by staff from the 
implementing partners, Flagship, and Global Fund project through KHANA were included and collated. 
8.7. Incentives 
Participants (including seeds) received an incentive for completing the interview (primary incentive) and 
another incentive (secondary incentive) for recruiting their peers to participate in the study. Each 
participant recruited up to two peers. Recruiters received an incentive as long as her recruit presented a 
coupon, fulfilled the eligibility criteria, and enrolled in the study. 
Based on findings from previous studies, the primary incentive of US$ 5.00 ($2.50 for transport and $2.50 
for communication) and secondary incentive of US$ 2.50 for each recruit were considered as appropriate. 
These incentives were set low enough to be non-coercive but high enough to cover the costs of 
participation plus transportation. 
Primary incentive: the following conditions were required in order that a participant could claim the 
primary incentive: 
• Have a coupon (except for seeds);
• Fulfill the study eligibility criteria;
• Provide informed consent; and
• Complete the interview process. 
Secondary incentive: a participant received an incentive for each individual (no more than two) she 
recruited. The recruit must fulfill the eligibility criteria and study requirements. A participant who 
distributed a coupon to her peers were contacted again (second contact) by the field research team to get 
the incentive and asked to find their recruits and encouraged them to enter the study. The second contact 
for secondary incentive was a good opportunity for field researcher to ask participant about the peers who 
refused her offer of a coupon, and for exploring the reasons of refusals.
8.8. Sources of Costing Data
Historical data from accounting system and other records from Flagship project and Global Fund through 
KHANA, as well as implementing partners were collected. The costing data were provided by organizations 
based on the matrix for costing data input developed by the evaluation team. 
9. Data Management
9.1. Data Entry and Data Cleaning
Deploying the tablet survey software, SurveyToGo, provided the platform for data centralization and 
allowed the evaluation team for real-time data processing to monitor quotas and interviewers’ locations to 
ensure accurate and timely data. Data were synchronized to the server immediately after interviews were 
completed in the field. This tablet application allowed for efficient skip patterns and reduce the time 
required for data entry and cleaning. Field researchers were required to upload completed questionnaire 
after data has been edited. Data editing were done immediately after completing each interview. The data 
analyst in the office was responsible for data management, including: 
• Synchronization of data from field researchers.
• Response validation using advanced logic rules in SurveyToGo for Android. The validation rules 
control whether the answer is valid or not. 
• Checking data and send feedback to field research manager, and
• Tracking enumerator progress and efficiency. 
In addition, the SurveyToGo application provided solutions to the following types of data errors: 
• Domain errors: each question had a domain (or range) of valid answers. An answer outside this 
domain was considered an error.
23
• Routing errors (skip pattern errors): the questionnaire contained routing instructions. A routing 
error occurs when an interviewer or respondent fails to follow a routing instruction, and a 
wrong path is taken through the questionnaire. As a result, the wrong questions are answered, 
leaving applicable questions unanswered and inapplicable items with entries. 
9.2. Weighting Adjustments
After data editing, the clean file required further preparation prior to data analysis. The selection of 
respondents with probability directly proportional to network size required weighting adjustment 
procedure in order to correct for unequal selection probabilities. Thus, based on the determined sample 
size and sample design, weighting adjustments were constructed and applied in the analysis. The logic of 
using weighting adjustments was to minimize the mean square errors of the estimates (difference between 
the value of the sample estimates and the true values), reducing biases in the estimates due to sampling 
and the biases of estimates from data missing due to nonresponse. 
Since all participants did not have the same probability of selection, the RDS population proportion 
estimates (PPEs) was applied. This procedure weighted each sample element by the inverse of its 
probability of selection so units with a small chance of being selected had more weight. In other words, 
groups with larger average network sizes were assigned lower weights, while groups with the smaller 
average network sizes were assigned higher weights. RDSAT application was employed to compute RDS 
weights for econometric modeling. 
10. Operational Definition of Key Variables
Key index variables were operationally defined below: 
• Program Exposure
Ten variables related to program activities were aggregated into a single measure, ‘index’, based on linear 
combinations, using Principle Component Analysis (PCA) in STATA 14. Then participants were ranked by 
program exposure score from PCA and classified in terciles, with the first 2 parts (lowest to medium scores) 
categorized as ‘some exposure’ and the last part with high score grouped as ‘high exposure’. Participants 
without exposure to any of the above activities/services were classified as ‘no exposure’’. The program 
exposure variable was coded 0 for no exposure, 1 for some exposure, and 2 for high exposure. Thus the 
comparison of outcomes was made between these categories:
1. Ever seen logo of SMARTgirl
2. Ever been approached by OW of SMARTgirl in the past 6 months
3. Individual meeting with OW in the past 6 months
4. Group meeting with OW in the past 6 months 
5. Received a copy of the SMARTgirl service directory
6. Visited Drop-in-Center in the past 6 months
7. Visited SMARTgirl Club in the past 6 months
8. Visited SMARTgirl Khmer website 
9. Visited SMARTgirl Facebook 
10. Called Voice4U (1295) 
• Risk Index
Principle Component Analysis (PCA) scores were classified into: low, medium, and high, based on the 
following 5 variables: 
1. Average number of sexual partners per week.
2. Condom use in last sex with live-in-partner. 
3. Condom use in last sex with boyfriend.
4. Condom use in last sex with client.
5. Drunk before having sex in the past 12 months
• Stigma and Discrimination Index
24
Principle Component Analysis (PCA) scores were classified into: low, medium, and high, based on the 
following 14 variables: 
1. Do you feel that health service providers usually treat you with respect?
2. Do you feel that health service providers ever discriminate against people like you? 
3. In the past 12 months, have you experienced any discriminatory behaviour by health service 
providers?
4. Felt discriminated against by service provider.
5. Felt verbally harassed during the visit to the health facility.
6. Believed that being an EW/a SMARTgirl, would be socially excluded. 
7. By presenting a health referral slip, would be identified as an EW/a SMARTgirl. 
8. Felt was not treated with respect by OW/NGO staff. 
9. Felt got incomplete STI diagnosis at the health facility because of EW status.
10. Felt was given the same drug by service provider because they knew status as EW.
11. Felt ashamed when repeatedly infected with STIs by partner(s).
12. Thought would be blamed by a service provider for being infected with HIV.
13. Thought would be blamed by a service provider for being infected with an STI.
14. Though would be blamed service provider for requesting for abortion. 
• HIV/STI Prevention Knowledge Index
Principle Component Analysis (PCA) scores were classified into: low and high, based on 5 variables: 
1. Can a male condom be re-used? 
2. Do you think a condom can prevent you from being infected with HIV? 
3. Do you think a condom can prevent you from being infected with an STI? 
4. Do you usually check the expiry date on condom before using it? 
5. Do you usually make sure there are no holes in the packaging before opening your condom? 
• Program Contact
This variable was measured by the combination analysis of EW reached with the following characteristics: 
1. Contacted by an NGO outreach worker for sexual, HIV or birth spacing education;
2. Participated in any NGO activity for sexual, HIV or birth spacing education in the past 12 months;
3. Received any sexual, HIV or birth spacing information from an NGO in the past 12 months,
4. Used any sexual, HIV or birth spacing services in the past 12 months; and 
5. Used any sexual, HIV or birth spacing services from any NGO in the past 12 months. 
EW that were not exposed to any of these variable were categorized as “unreached”, and as “reached” if 
exposed to any of these variables. 
• Exposure to Printed Education Materials Index
Principle Component Analysis (PCA) scores were categorized into some exposure and high exposure. Any 
EW that did not expose to any of these variables were classified as “no exposure”. The PCA score was based 
on the following 20 variables: 
1. Ever seen “Two better than one”
2. Ever seen “Your health your choice (Srey Ra)” 
3. Ever seen “Your decision” 
4. Ever seen “Thida and Leakhena” 
5. Ever seen “Smart choice” 
6. Ever seen “Secret bag” 
7. Ever seen “For my future” 
8. Ever seen “Value of life” 
9. Ever seen “Your choice”
10. Ever seen “Road of life”
11. Ever seen “Good habit”
12. Ever seen “Counselling card” 
13. Ever seen “Alcohol use”
14. Ever seen “Risk screening tools”
15. Ever seen “Service package guideline (SMARTgirl)” 
25
16. Ever seen “SMARTgirl fan” 
17. Ever seen “1295 sticker” 
18. Ever seen “Birth spacing network”
19. Ever seen “condom use and contraceptives”
20. Ever seen “Blood drop”
11. Data Analysis Framework 
To address multiple evaluation questions, a range of statistical analysis techniques were employed using 
STATA 14. The data analysis framework included the following comparisons and techniques. 
• Outcomes under the intervention groups with the outcomes under the non-intervention group: 
comparing the outcomes in SMARTgirl program – Flagship CoE, Flagship TA, and Non-Flagship, 
with the outcomes in locations with no EW intervention program. The key outcomes included 
the uptake of HIV tests, condom use, STI screening and treatment, and referrals to health 
services. 
• Outcomes between intervention groups: the analysis also compared outcomes in Flagship CoE
areas to areas with no EW intervention program, and Flagship TA areas with the outcomes in 
locations with no EW intervention program.
• Outcomes within the intervention groups: comparing the outcomes in Flagship CoE areas with 
Flagship TA areas. 
• Comparing the outcomes of participants with different levels of exposure to SMARTgirl 
program. 
Since the program variables above were categorical variables measured at the nominal level, proportion 
comparisons were made. Other confounding factors were not controlled. This analysis was to provide a 
general description of the outcome characteristics. 
Theoretically we could not estimate the average treatment effect (impact of the program) by simply taking 
the difference between the sample proportions for the intervention and control groups, because there 
were covariates that were related to the control groups and intervention groups. Therefore, a more robust 
estimator, Inverse-Probability-Weighted Regression Adjustment (IPWRA) was used to estimate the impact 
of the SMARTgirl program. In this modeling approach, covariates were identified and after conditioning on 
those covariates, any remaining influences on the intervention groups were not related to the control 
groups. The IPWRA estimator model was applied to both intervention groups and controls groups. Based 
on the above planned comparisons, each participant could be from SMARTgirl program under CoE areas, 
SMARTgirl program under non-CoE areas, or non-interventions; or with high exposed to SMARTgirl 
program, some exposed to the program, or not exposed to the program. For these scenarios, multivalued 
treatment effects were used so that these comparisons could be made. 
In addition to these techniques, descriptive statistics were used to describe the characteristics of the 
participants, utilization of services, and their exposures with and perspectives regarding the program 
activities and services. Regression techniques were applied to examine : The effects of exposure to SBC 
material, ICT and condom distribution, knowledge and behavior regarding HIV prevention and utilization of 
health services, the effects of SMARTgirl in reducing stigma and discrimination, and the barriers to services. 
Furthermore, comparisons of these effects were performed across the Flagship CoE areas, Flagship TA 
areas, and Non-Flagship areas. 
For the cost-effectiveness analysis, the first step in the analysis was to identify the resources that were 
used in the implementation of the program. Resources were classified corresponding to the inputs of 
activities for the program such as management, administration, and program components. Resources used 
in the last fiscal year were estimated by Flagship through reviewing administrative and financial 
documents. Comparison was made between the unit costs i.e. costs per HIV test, and cost per new case 
detection, under Flagship CoE, Flagship TA, and Non-Flagship. The Total Costs induced: listing of all costs of 
SMARTgirl program such as staff, transport, utilities, incentives, training, and other supplies, etc. The costs 
of testing device were not included. These costs were presented in actual amount and percentage of total 
26
costs. The Unit Costs: a unit cost (average) captures the relationship between total costs and the related 
volume of new HIV cases identified and number of HIV tests. This cost shed light on the intensity of 
resources used. 
12. Ethical Considerations
The protocol, data collection tools, and informed consent form were reviewed and approved by the 
Cambodian National Ethic Committee for Health Research (NECHR) on 30 December 2016 prior to starting 
of the study. The protocol was followed, with no exceptions.
The information sheet and consent form were translated from English to Khmer by research project staff 
who had good knowledge of the study area. Field researchers gave a copy of the informed consent form to 
every participant to read preceding the interview and asked participants if they had any questions. In cases 
of low-literacy, the information sheet and consent forms were read aloud by the field researchers to the 
participant during the consent process so that the participant could provide their consent with their 
signature.
The interviews were organized in a safe, private, and accessible location. Interviews were taken 
approximately one hour. The questionnaire was administered face-to-face with no other person in the 
setting than the field researcher and the study participant. The research team safeguarded these 
protections for participants:
o Participation was completely voluntary; 
o Subjects were free to withdraw at any time; 
o Informed consent was signed in a private setting;
o Confidentiality was guaranteed on all documents and tools used; 
o No names were used in written documentation of the study; and
o Field researchers were trained in discussing sensitive issues and protecting respondents’ 
confidentiality and human rights.
13. Limitations and Challenges
Carrying out studies with EW, especially using RDS, presented challenges. The main challenge was that the 
minimum sample size for EW from non-intervention areas could not be reached because the actual size of 
the population was too small to meet the desired sample size, and that some seeds were unable to recruit 
their peers to participate in the study within the defined timeframe. For example, in Moung OD, 100 EW 
had been required for the interviews, but the actual number of EW recruited to participate in the study 
were only 66 and even though the field researchers stayed longer in the field, the number of EW were 
exhausted. Based on mapping, there were only 6 KTVs (Karaoke bars) including a KTV’s family style, and 
few restaurants in this OD. 
Unfortunately at the time of data collection, many EW were in panic situation because the government had 
implemented a large scale anti-drug campaign and some EW were arrested. This situation brought 
challenges for the research team in reaching target participants and in getting responses to questions 
about overlapping risk including illicit drug use (Bourmant, 2017). Finally the research team decided to skip 
these overlapping risk questions. 
The research team increased sample size in other locations under the non-program area in order to ensure 
sufficient total sample size for the analysis. 
With the issue of spillover effects, particularly with the nature of high mobility among the EW, this might 
contaminate the effect of exposure to program among respondents from the non-program area, Non￾Flagship, and Flagship TA. Additionally, the cumulative effects of exposure to interventions before the 
Flagship could not be accounted for this evaluation design. Therefore, the true effects of the program could 
not be ensured, especially the comparisons between the Flagship CoE, Flagship TA, Non-Flagship, and no 
27
program areas, as well as between the levels of program exposure. The outcome differences might be 
accounted by other unobserved factors that were not in the model. 
Financial matters are seen as internal affair for implementers, and disclosure to outsiders was not easy. In 
this cost allocation assignment, the allocation of the costs were made by the Flagship and implementing 
partners. Consequently, this has the risk of distorting program costs. 
The analysis of the impact of the SMARTgirl programs on HIV confirmation testing, identifying new cases, 
reducing LTFU for HIV testing confirmation, and ART enrollment, links to ART and adherence support, viral 
suppression, and retention could not be performed as intended in the protocol. The total number of EW 
reported to be HIV positive was only 7, which is an insufficient number for meaningful analysis. 
28
14. Results 
14.1. Descriptive Analysis
Program and Participant Characteristics 
As shown in table 11, of the 1300 respondents, 300 were from the geographic areas of the Flagship CoE 
areas, 300 were from the geographic areas of Flagship TA areas, 300 were from the geographic areas of 
Non-Flagship areas, and 400 were from geographic areas with no existing EW programs. Across locations, 
50% of EW had no exposure to any EW program. About 85% of EW in "No Program" areas reported no 
exposure, while about one third (33-36%) of EW in the other three program geographic areas reported no 
exposure. By contrast, the majority (52-55%) of EW in the three types of program geographic areas 
reported a high level of exposure to the program. The differences in exposure levels across the different 
geographic areas was statistically significant, p = 0.000.
Table 11 Program exposure by geographic area
Program exposure
Geographic Area
Flagship 
CoE
Flagship TA 
(GF)
Non-Flagship 
(GF)
No 
Program
Total Chi-Square 
Test
No exposure 33.0% 36.3% 35.7% 84.5% 50.2%
289.241, 
df=6, p=.000
Some exposure 14.7% 10.0% 9.3% 7.3% 10.1%
High exposure 52.3% 53.7% 55.0% 8.3% 39.7%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Table 12 describes the sociodemographic characteristics of EW by geographic area and program exposure. 
While no significant differences on age was found by geographic area, a significantly larger proportion of 
EW aged 25-30 years (which accounted for 39% of all EW in the sample) were highly exposed to the 
program (44%). The differences in exposure by age group were statistically significant, p = 0.011. EW in the 
CoE areas had higher levels of education, with 45% having seven years or more of education, compared to 
29-36% of EW from other areas, p = 0.000. 
More EW had their main occupation at a KTV (62%) than other locations, including only 10% having their 
main occupation at massage parlors, 9% at a cafe/restaurant. There were significant differences in main 
occupation location by geographic area and by program exposure, p = 0.000 for both comparisons. For 
example, a much larger proportion of EW in no program locations worked at a KTV when compared to 
program implementation locations (84% versus 48-54%); while a larger proportion of EW in CoE geographic 
areas (13%) worked in beer gardens than the geographic locations (2-4%). At the same time, a larger 
proportion of EW in Non-Flagship areas (12%) reported their main occupation as a residential sex worker 
than the other program and non-program geographic areas (0-2%); and a larger proportion of EW in 
Flagship TA intervention areas gave their main occupation as a freelance sex worker compared to other 
areas (8% versus 0.3-4%). The distribution of EW with high exposure to the program largely mirrored the 
distribution of main occupations.
EW from CoE areas and Flagship TA areas had higher incomes than those in Non-Flagship or no program 
areas. The proportion of reported income greater than $300/month was 56%, 52%, 43%, and 29% in CoE, 
Flagship TA, Non-Flagship and no-program areas respectively. These differences were statistically 
significant, p = 0.000. More than half of all EW reported being divorced or separated (53%), while 21% had 
a non-cohabitating partner/boyfriend, and 10% reported being married.
The analysis showed that EW were fairly mobile, with 52% of all EW reporting less than 12 months at their 
current work place for their main occupation, and 43% of all EW reporting less than 12 months in their 
current place of living. EW in no program areas were most mobile and EW in CoE areas were least mobile 
29
where 79% of EW in CoE areas lived in their current locations for 12 months or more , while 34% of EW in 
the non-program areas reported living in their current locations for 12 months or more, p=0.000.
Table 12 Sociodemographic characteristics by geographic area and program exposure 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure High 
exposure
Total
Chi-Square 
Test
Age
18-24 years 33.7% 35.0% 35.0% 40.0% 36.2%
5.114, df=6, 
p=.529
40.1% 37.4% 31.0% 36.2%
3.117, df=4, 
p=.011
25-30 years 42.3% 38.0% 38.7% 36.0% 38.5% 35.1% 35.1% 43.8% 38.5%
31 years and above 24.0% 27.0% 26.3% 24.0% 25.2% 24.8% 27.5% 25.2% 25.2%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Education
Under 1 year 3.3% 11.3% 12.7% 12.8% 10.2%
30.872, 
df=6, p=.0009.8% 12.2% 10.3% 10.2%
.193, df=4, 
p=.526
1-6 years 52.0% 52.7% 58.3% 53.5% 54.1% 56.0% 54.2% 51.6% 54.1%
7 years and above 44.7% 36.0% 29.0% 33.8% 35.7% 34.2% 33.6% 38.2% 35.7%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Main Occupation 
Freelance sex worker .3% 7.7% 4.0% 1.0% 3.1%
435.748, df=18, p=.000
3.4% 3.8% 2.5% 3.1%
65.561, df=12, p=.000
Residential sex worker 0.0% .7% 12.0% 1.5% 3.4% .9% 4.6% 6.2% 3.4%
Massager parlor 9.0% 26.7% 5.0% 1.3% 9.8% 7.7% 10.7% 12.2% 9.8%
Beer garden 12.7% 4.3% 3.3% 2.0% 5.3% 4.6% 11.5% 4.7% 5.3%
KTV 54.0% 48.0% 52.7% 84.3% 61.6% 67.1% 54.2% 56.6% 61.6%
Cafe/restaurant 20.7% 8.3% 10.3% .5% 9.2% 6.7% 9.2% 12.4% 9.2%
Other 3.3% 4.3% 12.7% 9.5% 7.6% 9.6% 6.1% 5.4% 7.6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Income
Under $150 6.7% 11.0% 11.3% 13.8% 10.9%
72.208, df=12, 
p=.000
12.4% 13.7% 8.3% 10.9%
30.861,df=8, 
p=.000
$150 - $250 25.3% 24.3% 32.7% 43.0% 32.2% 37.2% 26.0% 27.5% 32.2%
$251 - $300 12.3% 13.0% 13.0% 14.0% 13.2% 13.6% 13.0% 12.6% 13.2%
$301 - $500 31.3% 32.7% 29.0% 19.8% 27.5% 23.3% 30.5% 32.2% 27.5%
$501 and above 24.3% 19.0% 14.0% 9.5% 16.2% 13.5% 16.8% 19.4% 16.2%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Marital status
Single 9.0% 6.7% 7.0% 7.5% 7.5%
44.171, df=18, p=.001
7.4% 9.2% 7.4% 7.5%
20.937, df=12, p=.051
Have partner/boyfriend 
and living together 6.0% 4.0% 9.7% 7.3% 6.8% 7.5% 5.3% 6.2% 6.8%
Have partner/boyfriend 
but not living together 17.0% 18.7% 25.3% 22.0% 20.8% 21.9% 21.4% 19.4% 20.8%
Married(having husband) 16.0% 9.7% 11.0% 6.3% 10.4% 7.8% 19.1% 11.4% 10.4%
Divorced/separated 51.0% 59.3% 45.0% 54.0% 52.5% 53.3% 43.5% 53.7% 52.5%
Widowed 1.0% 1.7% 1.7% 3.0% 1.9% 2.1% 1.5% 1.7% 1.9%
Other 0.0% 0.0% .3% 0.0% .1% 0.0% 0.0% .2% .1%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
30
Table 12 Sociodemographic characteristics by geographic area and program exposure (Continued) 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure High 
exposure
Total
Chi-Square 
Test
Duration living in current 
location
Less than 12 months 21.3% 43.0% 34.0% 66.3% 43.1%
155.522,
df=3,p=.0
56.0% 35.9% 28.5% 43.1%
92.367, 
df=2,
p=.000
12 months or more 78.7% 57.0% 66.0% 33.8% 56.9% 44.0% 64.1% 71.5% 56.9%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Duration working in 
current workplace
Less than 12 months 33.7% 56.3% 44.0% 69.0% 52.2%
96.674, 
df=3,p=.0
64.6% 48.1% 37.4% 52.2%
86.558, 
df=2, 
p=.000
12 months or more 66.3% 43.7% 56.0% 31.0% 47.8% 35.4% 51.9% 62.6% 47.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Risk, Discrimination and Stigma, Program Contact, and Sexual Activities 
Table 13 shows that based on the risk index, there were statistically significant differences (p = 0.001) in 
HIV risk based on geographic area. Higher proportions of EW in CoE (46%) and in no program (43%) areas 
were at low risk, compared to Flagship TA (35%) and Non-Flagship (37%) areas. No statistically significant 
differences were found with regard to the relationship between risk index and program exposure.
The data also showed that there were statistically significant differences in reported stigma, with 49% of 
EW in CoE areas reporting low stigma, compared to 25-39% of EW from other areas. 45% of EW from no 
program areas reported high stigma, compared to 20-34% of EW from program areas. Additionally, 47% of 
highly program-exposed EW reported low levels of stigma, while conversely, a large proportion (44%) of 
EW with no program exposure reported high levels of stigma. These differences were statistically 
significant, p = 0.000.
Table 13 Risk, discrimination and stigma, and program contact by geographic area and program 
exposure
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Risk index
Low risk 46.3% 35.3% 37.0% 43.0% 40.6%
22.410, 
df=6, 
001
43.5% 38.9% 37.4% 40.6%
.652, df=4, 
p=.325
Medium risk 26.0% 42.7% 34.3% 31.0% 33.3% 31.9% 33.6% 35.1% 33.3%
High risk 27.7% 22.0% 28.7% 26.0% 26.1% 24.7% 27.5% 27.5% 26.1%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Stigmatization 
Low 49.3% 32.3% 39.3% 24.5% 35.5%
66.361, 
df=6, 
000
25.7% 37.4% 47.3% 35.5%
6.818, df=4, 
p=.000
Medium 30.3% 33.3% 31.3% 30.3% 31.2% 30.5% 29.8% 32.6% 31.2%
High 20.3% 34.3% 29.3% 45.3% 33.3% 43.8% 32.8% 20.2% 33.3%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Program contact 
Unreached 14.3% 15.7% 13.3% 46.8% 24.4%
157.189,
df=3,p=.0
00
41.7% 26.0% 2.1% 24.4%
244.371, 
df=2, 
p=.000
Reached 85.7% 84.3% 86.7% 53.3% 75.6% 58.3% 74.0% 97.9% 75.6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
31
Table 14 shows that larger proportions of EW in program areas (65-75%) had been engaged in sex work for 
12 months or more, compared to only 47% of EW from non-program areas. These differences were 
statistically significant, p = 0.000. 
The overwhelming majority (92.5%) of all EW had fewer than seven sexual partners per week. Important 
and statistically significant differences (p = 0.000) were seen with regard to the average number of sexual 
partners per week. Much larger proportions of EW in Flagship TA and Non-Flagship areas (15% and 13%, 
respectively) reported seven or more sexual partners per week than did EW in CoE (0.3%) or non-program 
(3%) locations. 
The majority of EW primarily found clients from a bar/nightclub/KTV/massage parlor (59%), by phone 
(12%), or a restaurant (10%). Notably, a larger proportion of EW in CoE areas found clients through 
restaurants than EW in other areas, 23% versus 1-9% while a larger proportion of EW in Non-Flagship areas 
found clients by phone (18%) compared to other areas (7-12%). The differences in the main source of 
clients by geographic area was statistically significant, p = 0.000. Large proportions of EW (25%-39%) stated 
that in addition to their main source of clients, they additionally found clients by phone.
Table 14 Sexual activities by geographic area and program exposure
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Duration in sex work
Under 12 months 25.3% 35.0% 26.7% 53.5% 36.5%
78.787,df=
3p=.000
49.9% 31.3% 20.9% 36.5%
106.217,
df=2,p=.00
0
12 months or more 74.7% 65.0% 73.3% 46.5% 63.5% 50.1% 68.7% 79.1% 63.5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300 
Disclosed status as sex worker 
to friend or family
No 60.7% 44.3% 51.3% 45.8% 50.2%
20.598,df=
3p=.000
49.5% 47.3% 51.7% 50.2%
1.065, 
df=2, 
578
Yes 39.3% 55.7% 48.7% 54.3% 49.8% 50.5% 52.7% 48.3% 49.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300 
Usual number of sexual 
partners per week
0 17.0% 12.0% 14.7% 24.5% 17.6%
99.280,df=9, 
p=.000
23.1% 16.0% 11.0% 17.6%
38.602, df=6, 
p=000
1-2 69.7% 54.3% 53.0% 54.8% 57.7% 55.4% 58.8% 60.3% 57.7%
3-6 13.0% 18.3% 19.3% 17.8% 17.2% 16.2% 19.1% 17.8% 17.2%
7 or more .3% 15.3% 13.0% 3.0% 7.5% 5.2% 6.1% 10.9% 7.5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Main source of clients
Meka (sex broker) 1.7% 1.0% .3% 1.5% 1.2%
459.166, df=33, p=.000
1.2% 2.3% .8% 1.2%
70.320, df=22, p=.000
Restaurant 23.3% 7.7% 9.3% 1.3% 9.7% 6.6% 10.7% 13.4% 9.7%
Beer garden 6.3% 2.7% .3% 1.3% 2.5% 2.3% 5.3% 2.1% 2.5%
Street/public park .3% 7.0% .7% 0.0% 1.8% 1.7% 2.3% 1.9% 1.8%
Introduced by friends 2.0% 2.0% .7% 1.5% 1.5% 2.3% 2.3% .4% 1.5%
Facebook .7% .3% 0.0% .3% .3% .6% 0.0% 0.0% .3%
Apartment/Condo/Brothel 0.0% .3% 13.0% 1.3% 3.5% 1.4% 3.1% 6.2% 3.5%
Bar/nightclub/karaoke/massa
ge pallor 53.0% 47.0% 53.7% 75.3% 58.6% 63.4% 54.2% 53.7% 58.6% 
Street-based massage/coin 
massage
4.7% 15.7% .3% .3% 4.8% 3.1% 4.6% 7.2% 4.8% 
Guest house/hotel .3% 4.3% .3% 1.5% 1.6% 1.7% 2.3% 1.4% 1.6%
Phone call 7.3% 9.0% 18.3% 12.0% 11.7% 12.4% 11.5% 10.9% 11.7%
Other .3% 3.0% 3.0% 4.0% 2.7% 3.4% 1.5% 2.1% 2.7%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
32
Strategic Behavioral Communication 
Table 15 describes strategic behavioral communication among EW. In CoE areas, 49% of EW recall having 
ever seen the SMARTgirl logo, compared to only 37% of EW in Flagship TA areas, 30% of EW in Non￾Flagship areas and only 10% of EW in no program areas. The vast majority (81%) of EW reported finding the 
SMARTgirl logo attractive.
Approximately half (51-56%) of EW in the three program area types reported that they were contacted by 
an outreach worker in the previous 12 months. Smaller proportions of EW reported being contacted by 
outreach workers in the previous three months, with 34% of EW in CoE areas, 37% of EW in Flagship TA 
areas, and 43% of EW in Non-Flagship areas. Among EW in three program area types, 21-25% of EW 
reported having had an individual meeting with an outreach worker in the previous six months, while 37-
49% of EW reported a group meeting with an outreach worker in the previous six months. Only 26% of EW 
in the CoE areas reported ever receiving a copy of the SMARTgirl guide, and only one-third of these (32%) 
reported ever having used it.
The majority of EW in program areas reported that their main source of information about HIV/AIDS and 
STI services was outreach workers and volunteers (52-65%), while only 22% of EW in no program areas 
reported the same. Twice the proportion of EW (20%) in non-program areas reported that their main 
source of information about HIV/AIDS and STI services was friends or colleagues than EW in program areas
(10%). Similarly, a much larger proportion of EW in non-program areas reported that their main source of 
information about HIV/AIDS and STI services was television than EW in program areas, 18% versus 5-7%
respectively, the differences between groups was statistically significant, p = 0.000. Overall, a plurality of 
EW (37%) reported that TV was the best communication channel to provide information about sexual, HIV, 
and reproductive health matters, 22% stated that Facebook was the best communication channel while 
18% reported outreach activities as the best communication channel, and 14% reported radio as the best 
communication channel.
Table 15 Strategic behavioral communication
Variable
Geographic Area
Flagship 
CoE 
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Ever seen SMARTgirl logo
No 51.0% 63.3% 70.0% 90.5% 70.4%
138.904 
df=3 
p=.000
Yes 49.0% 36.7% 30.0% 9.5% 29.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness of SMARTgirl logo
Not attractive 21.0% 18.0% 20.7% 16.8% 18.9%
2.836 
df=3 
p=.418
Attractive 79.0% 82.0% 79.3% 83.3% 81.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Contact with OW in past 12 months
No 48.7% 48.0% 44.0% 94.0% 61.4%
260.903 
df=3 
p=.000
Yes 51.3% 52.0% 56.0% 6.0% 38.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Contact with OW in past 6 months
No 54.3% 52.3% 46.3% 95.3% 64.6%
241.70
5 
df=3
P 000
Yes 45.7% 47.7% 53.7% 4.8% 35.4%
Total 100% 100% 100% 100% 100%
33
Table 15 Strategic behavioral communication (Continued) 
Variable
Geographic Area
Flagship 
CoE 
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Contact with OW in past 3 months
No 66.3% 63.3% 56.7% 97.8% 73.1%
186.239
df=3
P=.000
Yes 33.7% 36.7% 43.3% 2.3% 26.9%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Individual meeting with OW in the past 
6 months
No 75.3% 79.0% 78.7% 97.3% 83.7%
79.621
df=3
P=.000
Yes 24.7% 21.0% 21.3% 2.8% 16.3%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Group meeting with OW in the past 6 
months
No 62.7% 54.7% 51.3% 95.8% 68.4%
209.543
df=3
P=.000
Yes 37.3% 45.3% 48.7% 4.3% 31.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Received a copy of SMARTgirl guide
No 74.0% 79.0% 83.7% 93.8% 83.5%
54.466
df=3
P=.000
Yes 26.0% 21.0% 16.3% 6.3% 16.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Ever used SMARTgirl guide (of those 
that received a copy)
No 67.9% 68.3% 71.4% 56.0% 67.4%
1.873
df=3
P=.599
Yes 32.1% 31.7% 28.6% 44.0% 32.6%
Total 100% 100% 100% 100% 100%
78 63 49 25 215
Main source information about 
HIV/AIDS and STI services
Do not know 5.0% 11.7% 8.0% 10.3% 8.8%
300.382, df=36, P=.000
NGO outreach workers/volunteers 65.3% 52.3% 58.0% 22.3% 47.4%
Friends/colleague 10.0% 10.0% 9.7% 20.3% 13.1%
Magazine/newspaper 1.0% .3% 1.0% .3% .6%
Special events .7% 8.0% .7% 0.0% 2.2%
Education material of SMARTgirl 0.0% 2.0% 0.0% 0.0% .5%
Sexual partner 1.0% .7% 2.3% 4.3% 2.2%
Voice4U .3% 0.0% 0.0% 0.0% .1%
Other Websites 0.0% .3% 0.0% 0.0% .1%
Other Facebooks 1.0% 2.0% 2.0% 3.3% 2.2%
TV 5.3% 7.3% 7.3% 17.8% 10.1%
Radio 1.7% 2.0% 2.7% 3.5% 2.5%
Other 8.7% 3.3% 8.3% 18.3% 10.3%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
34
Table 15 Strategic behavioral communication (Continued) 
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Best communication channel 
Facebook 20.0% 21.7% 27.0% 20.8% 22.2%
193.277, df=30, p=.000
Messenger 0.0% .3% .7% .8% .5%
Line .3% .3% .7% 1.0% .6%
WeChat 0.0% .3% 0.0% 0.0% .1%
Website .3% .3% 0.0% 0.0% .2%
Voice4U .7% .3% 2.0% .5% .8%
Outreach activities 35.7% 2.7% 23.7% 10.8% 17.6%
TV 28.0% 42.0% 34.7% 43.0% 37.4%
Radio 10.7% 24.3% 9.3% 13.5% 14.4%
Magazines/Newspaper 2.0% 3.0% 1.3% 3.5% 2.5%
Other 2.3% 4.7% .7% 6.3% 3.7%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Figure 4 shows that overall, of the 20 printed materials, only 4 of these materials were recognized by more 
than half of EW, and the majority (12 of the 20) printed materials were reported ever seen by one third or 
less EW in CoE areas. Among EW that had seen the materials, however, the vast majority found them to be 
“attractive”. More detail is available in Annex I, table 36. 
Figure 4 Percentage of EW ever saw printed education materials (Flagship CoE)
24.3%
40.0%
45.0%
20.7%
39.7%
19.3%
12.0%
26.7%
40.3%
27.7%
28.3%
54.3%
30.7%
22.3%
22.0%
61.7%
26.7%
61.3%
55.7%
32.3%
Ever seen “Two better than one”
Ever seen “Your health your choice (Srey Ra)” 
Ever seen “Your decision” 
Ever seen “Thida and Leakhena” 
Ever seen “Smart choice” 
Ever seen “Secret bag” 
Ever seen “For my future” 
Ever seen “Value of life” 
Ever seen “Your choice”
Ever seen “Road of life”
Ever seen “Good habit” 
Ever seen “Counselling card” 
Ever seen “Alcohol use”
Ever seen “Risk screening tools”
Ever seen “Service package guideline (SMARTgirl)” 
Ever seen “SMARTgirl fan” 
Ever seen “1295 sticker” 
Ever seen “Birth spacing network”
Ever seen “condom use and contraceptives”
Ever seen “Blood drop” 
35
SMARTgirl Club/Drop-in-Centre 
Table 16 shows that small proportions of EW in program areas (1-10%) had ever heard about a drop-in 
center for entertainment workers, with less than 3% having visited in the previous six months. Among the 
2.3% EW in Non-Flagship areas who knew about the drop-in centers, they had the highest rate of having 
visited one in the previous six months. Larger proportions of EW in program areas (31-61%) reported ever 
heard about the SMARTgirl club, 61% of EW in CoE areas reported ever having heard of the SMARTgirl club. 
20% of EW in CoE areas reported having visited the SMARTgirl club in the previous six months, as did 8% of 
EW in Non-Flagship areas and 7% of EW in Flagship TA areas. The differences between groups was 
statistically significant, p = 0.000.
Table 16 SMARTgirl Club/Drop-in-Centre
Variable
Geographic Area
Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square 
Test
Heard about drop-in-center for 
entertainment workers 
No 95.7% 99.0% 89.7% 98.5% 95.9%
44.133, 
df=3, 
p=.000
Yes 4.3% 1.0% 10.3% 1.5% 4.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Heard about SMARTgirl club for 
entertainment workers
No 39.3% 65.7% 69.0% 88.8% 67.5%
191.479, 
df=3, 
p=.000
Yes 60.7% 34.3% 31.0% 11.3% 32.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Visited drop-in-center in the 
past 6 months 
No 99.7% 100.0% 97.7% 100.0% 99.4%
19.201, 
df=3, 
p=.000
Yes 0.3% 0.0% 2.3% 0.0% 0.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Visited SMARTgirl club in the 
past 6 months 
No 79.7% 92.7% 92.0% 99.3% 91.5%
85.875, 
df=3, 
p=.000
Yes 20.3% 7.3% 8.0% 0.8% 8.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Usual monthly frequency of 
visits to drop-in-center (among 
those that visited)
 
Never visited 85.7% 66.7% 56.3% 100.0% 67.9%
Fisher' Exact 
Test= 0.896 
Less than once 0.0% 0.0% 6.3% 0.0% 3.6%
Once 14.3% 0.0% 18.8% 0.0% 14.3%
More than once 0.0% 33.3% 18.8% 0.0% 14.3%
Total 100% 100% 100% 100% 100%
7 3 16 2 28
Usual monthly frequency of 
visits to SMARTgirl club (among 
those that visited)
 
Never visited 49.0% 50.0% 34.7% 50.0% 45.9%
31.504, df=9, 
p=.000
Less than once 8.2% 17.4% 44.9% 14.3% 19.3%
Once 19.4% 17.4% 8.2% 7.1% 15.5%
More than once 23.5% 15.2% 12.2% 28.6% 19.3%
Total 100% 100% 100% 100% 100%
98 46 49 14 207
36
Social Media and Communication Technologies
As shown in table 17, overall, only 9% of all EW had ever heard of the SMARTgirl Khmer website (16% in 
CoE areas), 14% of EW had ever heard of the SMARTgirl Facebook page (22% in CoE areas), and only 15% of 
EW had ever heard of Voice4U (27% in CoE areas). Only 1.2% of all EW had ever visited the SMARTgirl 
Khmer website (3% in CoE areas), only 3% of EW had ever visited the SMARTgirl Facebook page (7% in CoE 
areas), and only 3% of EW had ever called Voice4U (10% in CoE areas). Even smaller proportions had ever 
utilized these communication channels in the past six months in Flagship, Flagship TA and Non-Flagship 
areas (0.6%, 2.2%, and 1.8 %) respectively. All these reported differences between groups were statistically 
significant, with p values ranging from 0.000 to 0.015.
Table 17 Social media and communication technologies 
Variable Geographic Area
Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test
Ever heard about SMARTgirl Khmer Website
No 84.0% 91.7% 92.0% 94.8% 90.9%
25.143, df=3, 
p=.000 
Yes 16.0% 8.3% 8.0% 5.3% 9.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Ever visited SMARTgirl Khmer 
Website 
No 97.3% 98.0% 100.0% 99.5% 98.8% 12.044, 
Yes 2.7% 2.0% 0.0% .5% 1.2% df=3, 
Total 100% 100% 100% 100% 100% p=.007
300 300 300 400 1300
Visited SMARTgirl Khmer Website in Past 6 Months
No 99.0% 98.3% 100.0% 100.0% 99.4% 10.481, 
Yes 1.0% 1.7% 0.0% 0.0% .6% df=3, 
Total 100% 100% 100% 100% 100% p=.015
300 300 300 400 1300
Ever heard about SMARTgirl Facebook Page
No 78.0% 88.7% 85.3% 90.3% 85.9%
23.715, df=3, 
p=.000 
Yes 22.0% 11.3% 14.7% 9.8% 14.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Ever visited SMARTgirl Facebook Page
No 93.3% 97.3% 97.7% 98.3% 96.8% 15.212, 
Yes 6.7% 2.7% 2.3% 1.8% 3.2% df=3, p=.002
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Visited SMARTgirl Facebook Page in Past 6 Months
No 95.3% 97.7% 98.0% 99.5% 97.8% 13.743, 
Yes 4.7% 2.3% 2.0% .5% 2.2% df=3, 
Total 100% 100% 100% 100% 100% p=.003
300 300 300 400 1300
Ever ticked "Like" on the SMARTgirl Facebook page
No 96.7% 98.0% 98.7% 99.5% 98.3% 8.738, 
Yes 3.3% 2.0% 1.3% 0.5% 1.7% df=3, 
Total 100% 100% 100% 100% 100% p=.034
300 300 300 400 1300
Ever heard about Voice4U
No 73.0% 88.3% 84.0% 93.3% 85.3%
58.989, df=3, 
p=.000 
Yes 27.0% 11.7% 16.0% 6.8% 14.7%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Ever called Voice4u (1295)
No 90.3% 98.7% 97.7% 99.3% 96.7%
50.658, df=3, 
p=.000 
Yes 9.7% 1.3% 2.3% .8% 3.3%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Called Voice4U (1295) in Past 6 Months
No 94.7% 99.0% 98.7% 99.8% 98.2%
27.377, df=3, 
p=.000 
Yes 5.3% 1.0% 1.3% .3% 1.8%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
37
Referrals to Health and Social Services 
Table 18 shows that relatively small proportions (13-18%) of EW in program geographic areas reported 
having been referred in the previous 12 months for STI screening/treatment by an outreach worker. EW 
with high program exposure were more likely (26%) to have been referred for STI screening/treatment by 
an outreach worker than those with no program exposure (2%). This difference was statistically significant, 
p = 0.000. Very small proportions (1-7%) of EW in program geographic areas reported having been referred 
in the previous 12 months for family planning by an outreach worker. EW with high program exposure were 
more likely (9%) to have been referred for STI screening/treatment by an outreach worker than those with 
low or no program exposure (1%). This difference was statistically significant, p = 0.000. Less than 1% of EW 
were referred to legal or psychosocial services, with no statistically significant differences noted by 
geographic area.
Table 18 Referral to health and social services 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square Test
Referred by SMARTgirl/ NGO 
outreach worker to STI 
screening and treatment in 
last 12 months
 
No 82.0% 86.7% 82.7% 98.3% 88.2%
59.536, df=3, 
p=.000
98.5% 94.7% 73.6% 88.2%
176.859, df=2, 
p=.000
Yes 18.0% 13.3% 17.3% 1.8% 11.8% 1.5% 5.3% 26.4% 11.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by SMARTgirl/ NGO 
outreach worker to HIV 
confirmatory test in last 12 
months
 
No 97.3% 88.0% 98.7% 99.5% 96.2%
72.294, df=3, 
p=.000
98.6% 98.5% 92.4% 96.2%
31.885, df=3, 
p=.000
Yes 2.7% 12.0% 1.3% .5% 3.8% 1.4% 1.5% 7.6% 3.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by SMARTgirl/ NGO 
outreach worker to Pre￾ART/ART services in last 12 
months
 
No 100% 100% 99.0% 99.5% 99.6%
5.421, df=3, 
p=.143
99.7% 99.2% 99.6% 99.6%
.595, df=3, 
p=.743
Yes 0.0% 0.0% 1.0% .5% .4% .3% .8% .4% .4%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by SMARTgirl/ NGO 
outreach worker to TB 
diagnostic in last 12 months
 
No 99.3% 98.3% 99.7% 100% 99.4%
8.301, df=3, 
p=.040
99.7% 99.2% 99.0% 99.4%
2.122, df=2, 
p=.346
Yes .7% 1.7% .3% 0.0% .6% .3% .8% 1.0% .6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
38
Table 18 Referral to health and social services (Continued) 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square Test
Referred by SMARTgirl/ NGO 
outreach worker to 
methadone maintenance 
therapy in last 12 months
No 99.7% 99.3% 100% 99.3% 99.5% 
2.498, df=3, 
p=.476
99.7% 100.0
% 99.2% 99.5% 
2.055, df=2, 
p=.358
Yes .3% .7% 0.0% .8% .5% .3% 0.0% .8% .5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Referred by SMARTgirl/ NGO 
outreach worker to Family 
planning in last 12 months
No 93.7% 92.7% 98.7% 98.3% 96.0%
23.763, df=3, 
p=.000
99.2% 99.2% 91.1% 96.0%
53.820, df=2, 
p=.000
Yes 6.3% 7.3% 1.3% 1.8% 4.0% .8% .8% 8.9% 4.0%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by SMARTgirl/ NGO 
outreach worker to legal 
service in last 12 months
No 99.3% 98.3% 99.7% 99.8% 99.3%
5.846, df=3, 
p=.119
99.8% 99.2% 98.6% 99.3%
6.083, df=2, 
p=.048
Yes .7% 1.7% .3% .3% .7% .2% .8% 1.4% .7%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Referred by SMARTgirl/ NGO 
outreach worker to 
Psychosocial service in last 12 
months
No 99.7% 98.7% 99.7% 99.8% 99.5%
4.632, df=3, 
p=.201
99.8% 100% 98.8% 99.5%
6.275, df=2, 
p=.043
Yes .3% 1.3% .3% .3% .5% .2% 0.0% 1.2% .5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
39
Condoms
Table 19 shows that nearly three-fourths of all EW reported using a condom the last time they had sex 
(74%), with no statistically significant difference based on location or level of program exposure (p = 0.145 
and p = 0.634, respectively). Only 7% of all EW reported using a condom the last time having sex with their 
husbands with 12% EW in the CoE areas reported the same than EW in the other geographic areas (4-7%); 
This difference was statistically significant (p = 0.015). No statistically significant difference was noted with 
regard to condom use with husbands based on program exposure (p = 0.525). Among those EW that did 
not use a condom at the last sex with their husband, the majority (70%) gave trusting their partner as the 
primary reason, and 12% reported the desire to have a baby as the second most common reason- and the 
reasons varied by geographic area and this difference was statistically significant, p=0.000.
Table 17 also shows that the overwhelming majority (97%) of EW reported using a condom the last time 
they had sex with a client. There were no statistically significant differences noted based on location or on 
level of exposure to the program. Of the small proportion (3%) that reported not having used a condom at 
their last sexual encounter with a client, the largest proportion said the reason was that their partner 
refused (38%), and 11% reported that no condom was available, while 8% said that the reason they did not 
use a condom was that they were drunk at the time.
Among all EW, 94% reported knowing where they could get a condom, with larger proportions of EW in 
Flagship TA and Non-Flagship areas (99% for both) compared with 89% in CoE areas and 90% in non￾program areas. These differences were statistically significant, p = 0.000. 55% of EW reported that the most 
common places they normally got condoms were from their workplace (guesthouse/B/massage 
parlor/KTV/spa/sauna/beer garden), 59% reported from a store/gas station/vendor/pharmacy, and, 29%
from a SMARTgirl outreach worker, and 29% reported to normally getting a condom from a client. The vast 
majority of EW (86%) reported that condoms were always available when they needed them and that they 
knew how to use a condom correctly (73%). 81% of all EW reported initiating condom use, with a larger 
proportion of highly program exposed EW (87%) reporting this event than the unexposed EW (76%). Highly 
program exposed EW were also more likely than unexposed EW to check the expiration date on condom 
before use (48% versus 29%, respectively) and to always ensure that there are no holes in the packaging 
before use (47% versus 31%, respectively). These differences were statistically significant, p = 0.000. Peer￾peer sales was the least common usual source of condoms for EW (only 0.1%).
Table 19 Condoms
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Condom use at last sex 
with any partner 
No 24.3% 29.7% 29.0% 23.3% 26.3% 
5.400, df=3, 
p=.145
26.0% 29.8% 25.8% 26.3% 
.911, df=2, 
p=.634
Yes 75.7% 70.3% 71.0% 76.8% 73.7% 74.0% 70.2% 74.2% 73.7% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Condom use at last sex 
with husband 
No 88.0% 93.5% 95.8% 93.5% 92.8% 
10.510, df=3, 
p=.015
91.8% 93.3% 93.9% 92.8% 
1.290, df=2, 
p=.525
Yes 12.0% 6.5% 4.2% 6.5% 7.2% 8.2% 6.7% 6.1% 7.2% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 208 214 215 293 930 466 90 374 930 
40
Table 19 Condoms (Continued)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Main reason for not 
using a condom at last 
sex with husband
 
No condom available 1.6% 3.0% 0.0% 1.1% 1.4% 
47.370, df=15, p=.000
1.2% 1.2% 1.7% 1.4% 
6.009, df=10, p=.815
Partner refused 10.9% 6.5% 6.8% 4.4% 6.8% 7.0% 9.5% 6.0% 6.8%
Trusted partner 77.0% 75.0% 62.6% 67.9% 70.2% 69.6% 66.7% 71.8% 70.2% 
Demonstrated fidelity 3.3% 3.5% 9.2% 9.1% 6.6% 8.2% 4.8% 5.1% 6.6%
Want having a baby 3.8% 10.5% 17.5% 13.9% 11.8% 11.0% 14.3% 12.3% 11.8% 
Other 3.3% 1.5% 3.9% 3.6% 3.1% 3.0% 3.6% 3.1% 3.1% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 183 200 206 274 863 428 84 351 863 
Condom use at last sex 
with live-in partner 
No 57.3% 68.7% 62.3% 49.2% 58.7% 
8.758, df=3, 
p=.033
58.5% 63.0% 57.7% 58.7% 
.495, df=2, 
Yes 42.7% 31.3% 37.7% 50.8% 41.3% 41.5% 37.0% 42.3% 41.3% 
p=.781
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 143 83 138 120 484 234 54 196 484 
Main reason for not 
using a condom at last 
sex with live-in partner
No condom available 1.2% 0.0% 3.5% 1.7% 1.8% 
16.024, df=15, p=.380
.7% 0.0% 3.5% 1.8% 
16.237, df=10, p=.093
Partner refused 7.3% 15.8% 8.1% 3.4% 8.5% 6.6% 14.7% 8.8% 8.5% 
Trusted partner 63.4% 61.4% 55.8% 69.5% 62.0% 65.7% 58.8% 58.4% 62.0% 
Showed fidelity 19.5% 14.0% 19.8% 18.6% 18.3% 21.2% 11.8% 16.8% 18.3% 
Want having a baby 2.4% 7.0% 4.7% 5.1% 4.6% 2.2% 2.9% 8.0% 4.6% 
Other 6.1% 1.8% 8.1% 1.7% 4.9% 3.6% 11.8% 4.4% 4.9% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 82 57 86 59 284 137 34 113 284 
Condom use at last sex 
with boyfriend 
No 44.7% 53.2% 48.5% 45.3% 48.0% 
3.851, df=3, 
p=.278
48.9% 46.1% 47.4% 48.0% 
.305, df=2, 
Yes 55.3% 46.8% 51.5% 54.7% 52.0% 51.1% 53.9% 52.6% 52.0% 
p=.859
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 170 220 198 258 846 405 89 352 846 
41
Table 19 Condoms (Continued)
Variable
Geographic Area Program Exposure
Flagship CoE
Flagship TA 
(GF)
Non￾Flagship 
(GF)
No Program
Total
Chi-Square 
Test
No exposure
Some 
exposure
High 
exposure
Total
Chi-Square 
Test
Main reason for not 
using a condom at last 
sex with boyfriend
 
No condom available 5.3% .9% 1.0% 1.7% 2.0%
12.421, df=15, p=.647
2.0% 0.0% 2.4% 2.0%
10.338, df=10, p=.411
Partner refused 9.2% 9.4% 11.5% 10.3% 10.1% 8.6% 7.3% 12.6% 10.1%
Trusted partner 59.2% 63.2% 64.6% 59.0% 61.6% 63.6% 65.9% 58.1% 61.6%
Showed fidelity 13.2% 14.5% 12.5% 18.8% 15.0% 16.2% 9.8% 15.0% 15.0%
Want having a baby 5.3% 7.7% 4.2% 7.7% 6.4% 6.6% 4.9% 6.6% 6.4%
Other 7.9% 4.3% 6.3% 2.6% 4.9% 3.0% 12.2% 5.4% 4.9%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 76 117 96 117 406 198 41 167 406 
Condom use at last sex 
with client 
No 4.4% 2.8% 4.0% 2.0% 3.2% 
3.795, df=3, 
p=.284
2.5% 1.7% 4.5% 3.2% 
4.355, df=2, 
p=.113
Yes 95.6% 97.2% 96.0% 98.0% 96.8% 97.5% 98.3% 95.5% 96.8% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 272 252 275 358 1157 570 118 469 1157 
Main reason for not 
using a condom at the 
last sex with client
No condom available 16.7% 0.0% 9.1% 14.3% 10.8% 
20.359, df=21, p=.499
7.1% 0.0% 14.3% 10.8% 
8.220, df=14, p=.878
Partner refused 58.3% 14.3% 45.5% 14.3% 37.8% 21.4% 50.0% 47.6% 37.8% 
Trusted partner 8.3% 42.9% 27.3% 42.9% 27.0% 42.9% 50.0% 14.3% 27.0% 
Drunk 8.3% 14.3% 0.0% 14.3% 8.1% 14.3% 0.0% 4.8% 8.1% 
Condom reduces sexual 
pleasure 8.3% 14.3% 0.0% 0.0% 5.4% 7.1% 0.0% 4.8% 5.4% 
Was forced by partner 
not to a use condom 0.0% 14.3% 9.1% 0.0% 5.4% 7.1% 0.0% 4.8% 5.4% 
Want having a baby 0.0% 0.0% 9.1% 0.0% 2.7% 0.0% 0.0% 4.8% 2.7% 
Other 0.0% 0.0% 0.0% 14.3% 2.7% 0.0% 0.0% 4.8% 2.7% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
12 7 11 7 37 14 2 21 37 
Knowledge of condom 
sources 
No 10.7% .7% 1.3% 10.0% 6.0% 
49.645, df=3, 
p=.000
8.1% 5.3% 3.5% 6.0% 
11.057, df=2, 
p=.004
Yes 89.3% 99.3% 98.7% 90.0% 94.0% 91.9% 94.7% 96.5% 94.0% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
Condom availability 
when needed
No 2.7% 7.3% 2.3% 4.5% 4.2% 
25.900, df=6, 
p=.000
5.2% 2.3% 3.5% 4.2% 
25.596, df=4, 
p=.000
Yes, always 89.7% 79.3% 91.3% 83.0% 85.6% 81.0% 87.8% 90.9% 85.6% 
Yes, sometimes 7.7% 13.3% 6.3% 12.5% 10.2% 13.8% 9.9% 5.6% 10.2% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
42
Table 19 Condoms (Continued)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Knows how to use a 
condom correctly 
No 28.7% 27.3% 19.3% 31.5% 27.1% 
13.468, df=3, 
p=.004
35.1% 26.0% 17.2% 27.1% 
46.453, df=2, 
p=.000
Yes 71.3% 72.7% 80.7% 68.5% 72.9% 64.9% 74.0% 82.8% 72.9% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Ever received training 
or a demonstration on 
how to use condom 
correctly
 
No 37.7% 50.0% 29.0% 61.3% 45.8% 
82.708, df=3, 
p=.000
61.7% 39.7% 27.1% 45.8% 
141.053, df=2, 
p=.000
Yes 62.3% 50.0% 71.0% 38.8% 54.2% 38.3% 60.3% 72.9% 54.2% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Usually initiator of 
condom use 
Client 10.7% 10.7% 7.0% 16.8% 11.7% 
27.320, df=6, p=.000
15.0% 11.5% 7.6% 11.7% 
23.511, df=4, p=.000
Sexual 
partner/sweetheart 5.3% 8.3% 5.7% 10.3% 7.6% 9.3% 6.9% 5.6% 7.6% 
Myself 84.0% 81.0% 87.3% 73.0% 80.7% 75.7% 81.7% 86.8% 80.7% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Usually checks the expiry 
date on condom before 
use
 
No 54.3% 60.7% 58.7% 74.5% 63.0% 
35.478, df=3, 
p=.000
71.5% 63.4% 52.1% 63.0% 
46.471, df=2, 
p=.000
Yes 45.7% 39.3% 41.3% 25.5% 37.0% 28.5% 36.6% 47.9% 37.0% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Inspects condom 
packaging for holes 
before use
Yes, always 36.0% 37.7% 48.7% 31.5% 37.9% 
30.798, df=6, 
p=.000
31.4% 36.6% 46.5% 37.9% 
37.999, df=4, 
p=.000
Yes, sometimes 22.7% 18.3% 18.3% 17.0% 18.9% 17.8% 22.9% 19.4% 18.9% 
Never 41.3% 44.0% 33.0% 51.5% 43.2% 50.8% 40.5% 34.1% 43.2% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
43
Table 19 Condoms (Continued)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Source of last condom 
obtained 
SMARTgirl OW 9.0% 17.0% 14.7% .8% 9.6% 
129.695, df=30, p=.000 
.8% 2.3% 22.7% 9.6% 
218.663, df=20, p=.000
SMARTgirl club 3.0% .7% 1.0% 0.0% 1.1% 0.0% 0.0% 2.7% 1.1% 
Street-based sale 2.0% .7% 0.0% 1.5% 1.1% 1.2% 1.5% .8% 1.1% 
NGO/outreach 
worker/DIC 1.7% .7% .3% 1.0% .9% 1.4% .8% .4% .9% 
Store/gas 
station/vendor/ 
pharmacy
26.0% 23.3% 21.3% 18.0% 21.8% 21.3% 23.7% 22.1% 21.8% 
Guesthouse/brothel/m
assage parlor/karaoke/ 
spas/saunas/beer
31.3% 23.7% 26.0% 42.8% 31.8% 34.0% 41.2% 26.7% 31.8% 
Client 14.0% 15.7% 17.0% 20.3% 17.0% 21.6% 12.2% 12.4% 17.0% 
Sexual 
partner/sweetheart 9.3% 14.7% 12.3% 10.0% 11.5% 13.8% 12.2% 8.3% 11.5% 
Friend 1.0% 1.7% 2.3% 1.3% 1.5% 1.4% 3.1% 1.4% 1.5% 
Family health 
clinic/health center 1.3% 1.0% 3.0% 2.8% 2.1% 2.5% 3.1% 1.4% 2.1% 
Other 1.3% 1.0% 2.0% 1.8% 1.5% 2.1% 0.0% 1.2% 1.5% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
 300 300 300 400 1300 653 131 516 1300 
Payment for last condom 
obtained
Bought 34.7% 34.3% 32.3% 42.5% 36.5% 
9.506, df=3, 
p=.023
38.6% 38.9% 33.1% 36.5% 
4.081, df=2, 
p=.130
Free 65.3% 65.7% 67.7% 57.5% 63.5% 61.4% 61.1% 66.9% 63.5% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300 
Consistent condom use
Not always 61.3% 49.3% 57.0% 56.8% 56.2% 
9.082, df=3, 
p=.028
59.7% 66.4% 49.0% 56.2% 
19.613, df=2, 
p=.000
Always 38.7% 50.7% 43.0% 43.3% 43.8% 40.3% 33.6% 51.0% 43.8% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
44
STI Screening and Treatment
Table 20 shows that while about half of EW in program areas (44-50%) were screened for an STI in the 
previous 6 months, a smaller proportion of EW in non-program areas (35%) were screened. This difference 
was statistically significant (p = 0.000). Most EW received their last STI screening or treatment at a public 
health facility (46%), while 20% received screening at a private clinic (20%), 13% at an NGO clinic and 12% a 
at drop-in center the difference in geographic area and program exposure was significant (p=0.000). A 
minority of EW in program areas (15-18%) were referred for STI screening by an outreach worker in the 
previous 6 months. 
Table 20 STI screening and treatment 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square Test
How many STI screenings are 
required each year? 
I don't know 48.3% 58.0% 56.3% 57.5% 55.2%
18.994, df=15, p=.214
62.6% 57.3% 45.3% 55.2%
65.066, df=10, p=.000
Once each year 2.0% 2.0% 1.7% 3.5% 2.4% 3.1% 3.1% 1.4% 2.4%
Twice each year 14.3% 10.0% 10.7% 10.5% 11.3% 11.2% 8.4% 12.2% 11.3%
Three times each year 15.7% 10.3% 13.3% 14.5% 13.5% 11.3% 18.3% 15.1% 13.5%
Four times each year 16.7% 16.3% 14.7% 10.8% 14.3% 8.9% 9.9% 22.3% 14.3%
More than four times each 
year
3.0% 3.3% 3.3% 3.3% 3.2% 2.9% 3.1% 3.7% 3.2% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Ever suspected yourself having 
any STI? 
No 67.0% 63.3% 60.3% 69.5% 65.4%
7.279, df=3, 
p=.064
67.7% 67.2% 62.0% 65.4%
4.304, df=2, 
p=.116
Yes 33.0% 36.7% 39.7% 30.5% 34.6% 32.3% 32.8% 38.0% 34.6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
STI screening in the past 12 
months 
No 40.7% 45.0% 42.3% 56.8% 47.0%
23.200, df=3, 
p=.000
58.8% 57.3% 29.5% 47.0%
105.810, df=2, 
p=.000
Yes 59.3% 55.0% 57.7% 43.3% 53.0% 41.2% 42.7% 70.5% 53.0%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
STI screening in the past 6 
months 
No 50.0% 56.0% 50.3% 65.0% 56.1%
21.447, 
df=3, 
67.8% 69.5% 37.8% 56.1%
116.275, 
df=2, 
Yes 50.0% 44.0% 49.7% 35.0% 43.9% 32.2% 30.5% 62.2% 43.9%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
STI screening in the past 3 
months 
No 66.7% 75.7% 61.3% 78.3% 71.1%
29.778, df=3, 
p=.000
79.8% 82.4% 57.2% 71.1%
80.863, df=2, 
p=.000
Yes 33.3% 24.3% 38.7% 21.8% 28.9% 20.2% 17.6% 42.8% 28.9%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
45
Table 20 STI screening and treatment (Continued) 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square Test
Location of last STI 
screening/treatment? 
Public health facilities 29.4% 40.9% 55.5% 57.0% 46.2%
175.299, df=15, p=.000
49.6% 47.5% 42.5% 46.2%
129.649, df=10, p=.000
Private clinics 18.6% 17.7% 13.6% 27.3% 19.6% 28.6% 19.2% 11.1% 19.6%
NGO clinics 22.6% 9.3% 9.1% 10.9% 12.9% 13.5% 21.2% 10.4% 12.9%
SMARTgirl clubs 8.1% 2.8% 3.6% 0.0% 3.5% 0.0% 2.0% 7.2% 3.5%
Other NGO facilities 
(club/drop-in-center) 10.0% 27.4% 8.6% 3.5% 12.0% 7.0% 8.1% 17.6% 12.0% 
NGO outreach worker in 
community 11.3% 1.9% 9.5% 1.2% 5.8% 1.3% 2.0% 11.1% 5.8% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
221 215 220 256 912 399 99 414 912
Referred by OW for STI 
screening/treatment in the 
past 12 months 
 
No 77.0% 79.7% 77.3% 95.5% 83.4%
62.127, df=3, 
p=.000
95.3% 92.4% 66.1% 83.4%
185.472, df=2, 
p=.000
Yes 23.0% 20.3% 22.7% 4.5% 16.6% 4.7% 7.6% 33.9% 16.6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by OW for STI 
screening and treatment in 
the past 6 months 
 
No 81.7% 85.0% 81.7% 97.0% 87.2%
52.015, df=3, 
p=.000
97.4% 94.7% 72.3% 87.2%
169.646, df=2, 
p=.000
Yes 18.3% 15.0% 18.3% 3.0% 12.8% 2.6% 5.3% 27.7% 12.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Referred by OW for STI 
screening and treatment in 
the past 3 months 
 
No 91.3% 92.7% 86.0% 98.8% 92.7%
42.324, df=3, 
p=.000
99.1% 96.9% 83.5% 92.7%
106.839, df=2, 
p=.000
Yes 8.7% 7.3% 14.0% 1.3% 7.3% .9% 3.1% 16.5% 7.3%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
Last referral location for STI 
screening and treatment 
Public health facilities 51.6% 40.7% 72.3% 76.1% 58.7%
93.837, df=12, p=.000
52.1% 40.0% 63.4% 58.7%
32.618, df=8, p=.000
Private clinics 1.1% 0.0% .9% 0.0% .6% 1.4% 0.0% .4% .6%
NGO clinics 35.8% 19.8% 13.4% 23.9% 22.7% 28.2% 51.4% 16.8% 22.7%
SMARTgirl clubs 9.5% 3.3% 8.0% 0.0% 6.1% 0.0% 5.7% 8.0% 6.1%
Other NGO facilities 
(club/drop-in-center) 2.1% 36.3% 5.4% 0.0% 11.9% 18.3% 2.9% 11.3% 11.9% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
95 91 112 46 344 71 35 238 344
46
HIV Testing (Any Type)
Table 21 shows that more than half of EW in program areas (60-67%) received HTC in the previous 6 
months, while a smaller proportion of EW in non-program areas (46%) received HTC in the previous 6 
months. This difference was statistically significant (p = 0.000). Most EW last received HTC at a public 
health facility (33%), while 18% received HTC in a private clinic, 27% in a community setting by an outreach 
worker and, 8% received HTC at an NGO clinic (8%). A minority of EW in program areas (10-16%) were 
referred for HTC by an outreach worker in the previous 6 months. These differences were statistically 
significant (p=0.000) by geographic area and by program exposure type.
Table 21 HIV testing and counseling (HTC)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square 
Test
How many HIV tests are 
required per year? 
I don't know 35.3% 44.0% 40.7% 44.5% 41.4%
25.376, df=15, p=.045
48.4% 42.0% 32.4% 41.4%
73.895, df=10, p=.000
Once a year 1.3% 3.0% 2.7% 3.3% 2.6% 3.7% 2.3% 1.4% 2.6%
Twice a year 20.7% 18.7% 15.0% 17.5% 17.9% 17.3% 16.8% 19.0% 17.9%
Three times a year 17.3% 13.3% 17.7% 19.0% 17.0% 16.4% 23.7% 16.1% 17.0%
Four times a year 23.3% 18.3% 21.7% 13.0% 18.6% 11.8% 13.7% 28.5% 18.6% 
More than four a year 2.0% 2.7% 2.3% 2.8% 2.5% 2.5% 1.5% 2.7% 2.5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Ever suspected self of 
having contracting HIV? 
No 68.0% 58.3% 64.0% 68.5% 65.0%
9.333, df=3, 
p=.025
68.0% 63.4% 61.6% 65.0%
5.307, df=2, 
p=.070
Yes 32.0% 41.7% 36.0% 31.5% 35.0% 32.0% 36.6% 38.4% 35.0%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
 300 300 300 400 1300 653 131 516 1300 
HIV test in the past 12 
months 
No 23.7% 26.0% 22.7% 40.5% 29.2%
36.862, df=3, 
p=.000
40.7% 40.5% 11.6% 29.2%
127.245, 
df=2, p=.000
Yes 76.3% 74.0% 77.3% 59.5% 70.8% 59.3% 59.5% 88.4% 70.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300 
HIV test in the past 6 
months 
No 38.7% 40.3% 33.3% 53.8% 42.5%
33.419, df=3, 
p=.000
55.7% 60.3% 21.1% 42.5%
160.375, 
df=2, p=.000
Yes 61.3% 59.7% 66.7% 46.3% 57.5% 44.3% 39.7% 78.9% 57.5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
HIV test in the past 3 
months 
No 62.0% 62.7% 51.3% 73.3% 63.2%
35.738, df=3, 
p=.000
73.8% 76.3% 46.3% 63.2%
104.521, 
df=2, p=.000
Yes 38.0% 37.3% 48.7% 26.8% 36.8% 26.2% 23.7% 53.7% 36.8%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
47
Table 21 HIV testing and counseling (HTC) (Continued)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square 
Test
Location of last HIV test 
Never tested 5.7% 6.0% 4.0% 13.8% 7.8%
167.769, df=18, p=.000
12.9% 6.9% 1.7% 7.8%
323.749, df=12, p=.000
Public health facility 20.0% 32.7% 33.7% 43.8% 33.4% 42.3% 37.4% 21.1% 33.4%
Private clinic 17.0% 18.0% 16.0% 20.3% 18.0% 22.8% 20.6% 11.2% 18.0%
NGO clinic 14.3% 4.7% 8.0% 6.0% 8.1% 7.7% 12.2% 7.6% 8.1%
SMARTgirl club 9.3% 2.3% 4.3% 0.0% 3.7% 0.0% 1.5% 8.9% 3.7%
Other NGO facility 
(club/drop-in-center) .7% 1.0% 1.7% 2.8% 1.6% 2.5% .8% .8% 1.6% 
NGO outreach worker in 
community 33.0% 35.3% 32.3% 13.5% 27.4% 11.9% 20.6% 48.6% 27.4% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
Referred by OW for HIV 
test in the past 6 months 
No 89.3% 84.0% 86.0% 98.0% 90.0%
45.926, df=3, 
p=.000
96.6% 95.4% 80.2% 90.0%
90.875, df=2, 
p=.000
Yes 10.7% 16.0% 14.0% 2.0% 10.0% 3.4% 4.6% 19.8% 10.0% 
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
300 300 300 400 1300 653 131 516 1300
48
HIV Testing Using Finger Prick (CBHTC)
Table 22 shows that approximately half of EW in program areas (48-56%) received HTC using a finger prick 
(Community-Based HIV Testing and Counselling, or CBHTC), and a smaller proportion of EW (34%) in non￾program areas received CBHTC. This difference was statistically significant (p = 0.000). Only 0.7% of EW 
reported being HIV positive. 
Table 22 HIV testing using finger prick (CBHTC)
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High exposure
Total
Chi-Square Test
HIV test using finger prick 
in the past 12 months 
No 35.3% 37.3% 33.7% 54.5% 41.3%
42.309, df=3, 
p=.000
56.4% 60.3% 17.4% 41.3%
201.712, 
df=2, p=.000
Yes 64.7% 62.7% 66.3% 45.5% 58.7% 43.6% 39.7% 82.6% 58.7%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
HIV test using finger prick 
in the past 6 months 
No 48.7% 51.7% 43.7% 65.8% 53.5%
39.006, df=3, 
p=.000
68.3% 77.1% 28.7% 53.5%
214.552, 
df=2, p=.000
Yes 51.3% 48.3% 56.3% 34.3% 46.5% 31.7% 22.9% 71.3% 46.5%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
HIV test using finger prick 
in the past 3 months 
No 70.3% 70.0% 59.3% 82.0% 71.3%
43.766, df=3, 
p=.000
82.8% 86.3% 52.9% 71.3%
142.214, df=2, 
p=.000
Yes 29.7% 30.0% 40.7% 18.0% 28.7% 17.2% 13.7% 47.1% 28.7%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
300 300 300 400 1300 653 131 516 1300
Tested for HIV using finger 
prick at drop-in-center in 
the last 3 months
No 98.8% 100% 98.8% 100% 99.4% 
6.178, df=3, 
p=.103
100% 100% 98.8% 99.4% 
6.965, df=2, 
p=.031
Yes 1.2% 0.0% 1.2% 0.0% .6% 0.0% 0.0% 1.3% .6%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
253 245 258 278 1034 455 99 480 1034
Tested for HIV using finger 
prick at SMARTgirl Club in 
the last 3 months
No 90.5% 98.0% 94.6% 100% 95.8%
33.877, df=3, 
p=.000
100% 100% 91.0% 95.8%
51.783, df=2, 
p=.000
Yes 9.5% 2.0% 5.4% 0.0% 4.2% 0.0% 0.0% 9.0% 4.2%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 
253 245 258 278 1034 455 99 480 1034
HIV status
Negative 100% 99.2% 97.7% 99.3% 99.0%
18.084, df=6, 
p=.006
98.2% 100% 99.6% 99.0%
6.217, df=4, 
p=.183
Positive 0.0% 0.0% 2.3% .4% .7% 1.1% 0.0% .4% .7%
Refuse to answer 0.0% .8% 0.0% .4% .3% .7% 0.0% 0.0% .3%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100%
253 245 258 278 1034 455 99 480 1034
49
HIV/STI Prevention Knowledge 
Figure 5 shows that overall, most (55%) of EW demonstrated a low level of HIV/STI prevention knowledge, 
compared to a minority (45%) that demonstrated a high level of HIV/STI prevention knowledge. This 
pattern was evident across all geographic areas, except for Non-Flagship geographic areas, where the 
majority (54%) of EW had a high level of HIV/STI prevention knowledge. In addition, EW in non-program 
areas had the highest proportion with a low level of HIV/STI prevention knowledge (63%). These 
differences were statistically significant, p=0.000.
Figure 5 HIV/STI prevention knowledge by geographic area 
Figure 6 shows that overall, higher program exposure was associated with higher levels of HIV/STI 
prevention knowledge. EW with no program exposure had the largest proportion with a low level of 
HIV/STI prevention knowledge (63%) and the lowest proportion (37%) with a high level of HIV/STI 
prevention knowledge. Conversely, EW with high program exposure had the highest (55%) level of HIV/STI 
prevention knowledge compared to other levels of program exposure. These differences were statistically 
significant, p=0.000.
Figure 6 HIV/STI prevention knowledge by level of program exposure 
52.7% 56.7%
45.7%
62.8%
55.1%
47.3% 43.3%
54.3%
37.3%
44.9%
Flagship CoE Flagship TA Non-Flagship No program Total
N=1300, Chi-Square=21.266, df=3, p=.000 
Low High
63.4%
53.4%
45.0%
55.1%
36.6%
46.6%
55.0%
44.9%
No exposure Some exposure High exposure Total
N=1300, Chi-Square=39.764, df=2, p=.000 
Low High
50
14.2. Program Impact 
The analysis employed a treatment effects analysis to identify the causal effects of geographic area and 
program exposure on the outcome variables of STI screening and treatment, HIV testing, and stigma and 
discrimination. The treatment-effects analysis employed an inverse probability weighted regression 
adjustment (IPWRA) model. This "doubly robust" approach, controlled for confounders with this 
observational data set in identifying causal effects on the outcomes of interest.
STI Screening and Treatment 
Figures 7 and 8 show the impact of geographic area and program exposure on the probability of an EW 
receiving STI screening/treatment in the previous six months, controlling for confounders as described in 
Annex I, tables 37 and 38. EW in Non-Flagship geographic areas had a 47.3% probability of receiving STI 
screening/treatment in the previous six months, while EW in CoE geographic areas had a 46.9% probability, 
EW in Flagship TA geographic areas had a 43.5% probability, and EW in no program areas had a 37% 
probability of receiving STI screening/treatment in the previous six months. Thus, EW in all program areas 
were more likely to have received STI screening/treatment in the previous six months than EW located in 
areas without a program.
Program exposure had a notable impact on the probability of an EW receiving STI screening/treatment in 
the previous six months. While EW with no exposure had only a 33.2% probability of STI 
screening/treatment in the previous six months, highly exposed EW had a 61.5% probability. This translates 
into a 28.3% increased probability that highly exposed EW would receive STI screening/treatment in the 
previous six months when compared to their unexposed counterparts, regardless of geographic location or 
other potential confounders.
Figure 7 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting STI screening and treatment in the previous 6 
months 
***p<.001
Figure 8 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting STI screening and treatment in the previous 6 months
***p<.001
33.2%
26.5%
61.5%
No exposure*** Some exposure*** High exposure***
46.9% 43.5% 47.3%
37.0%
Flagship CoE*** Flagship TA (GF)*** Non-Flagship (GF)*** No Program***
51
Table 23 shows the results of the binary logistic regression model to explain factors associated with EW 
receiving STI screening/treatment in the previous six months. Age, education, income, mobility and 
stigma/discrimination showed no statistically significant impacts. However, suspicion of ever having an STI 
was associated with STI screening (odds ratio 1.55, p = 0.001), the risk index showed no significant 
association. Having had an individual or group meeting with an outreach worker in the previous six months 
doubled the likelihood of STI screening/treatment in the previous six months (odds ratio 2.11, p = 0.000 
and odds ratio 2.06, p = 0.000, respectively). High exposure to outreach education printed materials also 
increased the likelihood of STI screening (odds ratio 1.46, p = 0.040).
Table 23 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on STI screening and treatment in the past 6 months 
Variable Odds ratio Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Age
18-24 years – reference 
25-30 years 1.15 0.1747 0.029 0.0319
31 years and above 1.22 0.2223 0.041 0.0386
Education
Under 1 year – reference 
1-6 years 0.96 0.1983 -0.009 0.0435
7 years and above 1.06 0.2372 0.012 0.0472
Income 
Under $150 – reference 
$150 - $250 1.25 0.2599 0.047 0.0430
$250 - $300 1.21 0.3003 0.040 0.0517
$300 - $500 1.12 0.2441 0.023 0.0451
$500 and above 1.12 0.2701 0.023 0.0501
Marital status 
Single - reference
Having partner/boyfriend 1.05 0.1900 0.010 0.1900
Married(having husband) 1.16 0.3467 0.032 0.0631
Divorced/separated/widowed 0.97 0.2400 -0.006 0.0517
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.13 0.2227 0.027 0.0416
Duration working in current workplace
Less than 12 months – reference
12 months and above 1.21 0.2620 0.040 0.0464
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.86 0.1441 -0.033 0.0350
Disclosed status as sex worker (0=no, 1=yes) 0.76* 0.0941 -0.058* 0.0259
Ever suspected having any STI (0=no, 1=yes) 1.55** 0.2031 0.092** 0.0270
Risk index
Low risk – reference 
Medium risk 1.34* 0.1905 0.062* 0.0302
High risk 1.03 0.1608 0.006 0.0325
Stigma and discrimination index 
Low – reference 
Medium 0.98 0.1513 -0.003 0.0318
High 1.22 0.1894 0.042 0.0323
Individual meeting with OW in the past 6 months (0=no, 1=yes) 2.11*** 0.4120 0.157*** 0.0401
Group meeting with OW in the past 6 months (0=no, 1=yes) 2.06*** 0.3126 0.152*** 0.0307
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 1.50 0.3820 0.085 0.0534
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 0.99 0.1629 -0.002 0.0349
High exposure 1.46* 0.2681 0.082* 0.0407
Received a copy of the SMARTgirl service directory guide (0=no, 1=yes) 1.31 0.2495 0.057 0.0399
Visited SMARTgirl Khmer website (0=no, 1=yes) 2.23 1.4946 0.169 0.1408
Visited SMARTgirl Facebook (0=no, 1=yes) 1.44 0.5933 0.076 0.0867
Called Voice4U (1295) (0=no, 1=yes) 1.23 0.4806 0.043 0.0822
N=1300, Wald chi-square=163.92, df=29, Sig. Level=0.000, Pseudo R-square=0.1121
52
HIV Testing (Any Type)
Figures 9 and 10 show the impact of geographic area and program exposure on the probability of an EW 
receiving HTC in the previous six months, controlling for confounders as described in Annex I, tables 39 and 
40. EW in all program geographic areas were more likely to have received HTC in the previous six months 
than EW located in areas without a program. At 67.2% probability, EW in Non-Flagship geographic areas 
had the highest probability of receiving HTC in the previous six months, while EW in CoE geographic areas 
had a 58.5% probability, EW in Flagship TA geographic areas had a 61% probability, and EW in no program 
areas only had a 48.1% probability of receiving HTC in the previous six months. 
Program exposure had a notable impact on the probability of an EW receiving HTC in the previous six 
months. While EW with no exposure had only a 44% probability of HTC in the previous six months, highly 
exposed EW had a 79.4% probability.
Figure 9 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV testing (any type) in the previous 6 
months 
***p<0.001
Figure 10 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV testing (any type) in the previous 6 months 
***p<0.001
58.5% 61.0%
67.2%
48.1%
Flagship CoE*** Flagship TA (GF)*** Non-Flagship (GF)*** No Program***
44.0%
37.7%
79.4%
No exposure*** Some exposure*** High exposure***
53
Table 24 shows the results of the binary logistic regression model to elucidate the factors associated with 
EW receiving HTC in the previous six months. Age, education, income, mobility and stigma/discrimination 
showed no statistically significant effects. Similarly, exposure to outreach education printed materials and 
receipt of the SMARTgirl service directory guide showed no impact while having had an individual or group 
meeting with an outreach worker in the previous six months greatly increased the likelihood of getting an 
HTC in the previous six months (odds ratio 5.56, p = 0.000 and odds ratio 3.34, p = 0.000, respectively). 
Having visited the SMARTgirl Facebook page also showed an impact on receiving HTC (odds ratio 4.13, p = 
0.018).
Table 24 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on any type of HIV testing in the past 6 months 
Variable Odds 
ratio 
Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Age
18-24 years – reference 
25-30 years 1.14 0.1759 0.027 0.0310
31 years and above 1.05 0.1926 0.010 0.0371
Education
Under 1 year – reference 
1-6 years 0.85 0.1838 -0.033 0.0435
7 years and above 0.99 0.2302 -0.001 0.0466
Income 
Under $150 – reference 
$150 - $250 1.24 0.2548 0.044 0.0414
$250 - $300 0.99 0.2475 -0.001 0.0505
$300 - $500 1.18 0.2543 0.034 0.0435
$500 and above 1.19 0.2900 0.035 0.0494
Marital status 
Single - reference
Having partner/boyfriend 0.96 -0.1700 -0.009 -0.1700
Married(having husband) 0.80 0.2492 -0.045 0.0624
Divorced/separated/widowed 0.78 0.1942 -0.051 0.0500
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.28 0.2563 0.049 0.0410
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.95 0.2082 -0.011 0.0443
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.81 0.1360 -0.041 0.0331
Disclosed status as sex worker (0=no, 1=yes) 0.86 0.1090 -0.030 0.0254
Ever suspected contracting HIV (0=no, 1=yes) 0.78 0.1040 -0.050 0.0267
Risk index
Low risk – reference 
Medium risk 1.36* 0.1979 0.061* 0.0294
High risk 1.13 0.1785 0.025 0.0321
Stigma and discrimination index 
Low – reference 
Medium 1.25 0.1984 0.045 0.0317
High 1.39* 0.2184 0.066* 0.0311
Individual meeting with OW in the past 6 months (0=no, 1=yes) 5.56*** 1.5735 0.346*** 0.0544
Group meeting with OW in the past 6 months (0=no, 1=yes) 3.34*** 0.5686 0.243*** 0.0315
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 1.68 0.6314 0.104 0.0756
54
Table 24 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on any type of HIV testing in the past 6 months (Continued)
Variable Odds 
ratio 
Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 0.92 0.1422 -0.018 0.0318
High exposure 1.24 0.2291 0.044 0.0380
Received a copy of the SMARTgirl service directory guide (0=no, 
1=yes)
0.81 0.1772 -0.041 0.0439
Visited SMARTgirl Khmer website (0=no, 1=yes) 0.42 0.3039 -0.173 0.1438
Visited SMARTgirl Facebook (0=no, 1=yes) 4.13* 2.4897 0.286* 0.1196
Called Voice4U (1295) (0=no, 1=yes) 1.75 0.8656 0.112 0.1002
Constant 0.69 0.2792
N 1300
Wald chi-square 163.47
Degree of freedom 29
Sig. Level 0.000
Pseudo R-square 0.1454
HIV Finger Prick Test (CBHTC)
Figures 11 and 12 show the impact of geographic area and program exposure on the probability of an EW 
receiving CBHTC in the previous six months, controlling for confounders as described in Annex I, tables 41 
and 42. EW in Non-Flagship geographic areas had a 56.6%% probability of receiving CBHTC in the previous 
six months, while EW in CoE geographic areas had a 48.7% probability, EW in Flagship TA geographic areas 
had a 52.3% probability, and EW in no program areas only had a 36.8% probability of receiving CBHTC in 
the previous six months. 
Program exposure had a significant impact on the probability of an EW receiving CBHTC in the previous six 
months. While EW with no exposure had only a 32.5% probability of CBHTC in the previous six months, 
highly exposed EW had a 72.3% probability.
Figure 11 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV finger prick testing in the past 6 months 
***p<0.001
48.7%
52.3%
56.6%
36.8%
Flagship CoE*** Flagship TA (GF)*** Non-Flagship (GF)*** No Program***
55
Figure 12 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV finger prick testing in the past 6 months 
***p<0.001
Table 25 shows the results of the binary logistic regression model explaining the factors associated with EW 
receiving CBHTC in the previous six months. Age, education, income, mobility and stigma/discrimination 
showed no statistically significant impacts. Exposure to related social media, outreach education printed 
materials and receipt of the SMARTgirl service directory guide also showed no impact. Similar to what was 
seen for overall HTC access, having had an individual or group meeting with an outreach worker in the 
previous six months greatly increased the likelihood of receiving HTC in the previous six months (odds ratio 
5.56, p = 0.000 and odds ratio 3.99, p = 0.000, respectively).
Table 25 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on HIV finger prick testing in the past 6 months 
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Age
18-24 years – reference 
25-30 years 1.01 0.1574 0.001 0.0303
31 years and above 0.87 0.1661 -0.027 0.0366
Education
Under 1 year – reference 
1-6 years 0.90 0.2021 -0.019 0.0432
7 years and above 1.06 0.2534 0.012 0.0463
Income 
Under $150 – reference 
$150 - $250 1.16 0.2401 0.029 0.0398
$250 - $300 1.00 0.2513 0.000 0.0482
$300 - $500 0.94 0.2042 -0.011 0.0414
$500 and above 0.93 0.2344 -0.014 0.0481
Marital status 
Single - reference
Having partner/boyfriend 0.95 -0.2200 -0.011 -0.2200 
Married(having husband) 0.83 0.2609 -0.036 0.0612
Divorced/separated/widowed 0.74 0.1845 -0.059 0.0491
32.5%
20.5%
72.3%
No exposure*** Some exposure*** High exposure***
56
Table 25 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on HIV finger prick testing in the past 6 months (Continued)
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.27 0.2559 0.046 0.0393
Duration working in current workplace
Less than 12 months – reference
12 months and above 1.03 0.2274 0.005 0.0428
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.90 0.1548 -0.019 0.0328
Disclosed status as sex worker (0=no, 1=yes) 0.91 0.1183 -0.017 0.0249
Ever suspected contracting HIV (0=no, 1=yes) 0.77 0.1039 -0.051 0.0259
Risk index
Low risk – reference 
Medium risk 1.36* 0.2029 0.060* 0.0288
High risk 1.23 0.1995 0.040 0.0311
Stigma and discrimination index 
Low – reference 
Medium 1.18 0.1902 0.031 0.0304
High 1.34 0.2188 0.056 0.0306
Individual meeting with OW in the past 6 months (0=no, 1=yes) 5.56*** 1.4074 0.330*** 0.0456
Group meeting with OW in the past 6 months (0=no, 1=yes) 3.99*** 0.6497 0.267*** 0.0277
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 1.76 0.6044 0.109 0.0657
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 0.89 0.1472 -0.022 0.0319
High exposure 1.06 0.2022 0.012 0.0374
Received a copy of the SMARTgirl service directory guide (0=no, 
1=yes) 0.93 0.1928 -0.014 0.0400 
Visited SMARTgirl Khmer website (0=no, 1=yes) 0.33 0.2108 -0.216 0.1234
Visited SMARTgirl Facebook (0=no, 1=yes) 2.77* 1.3518 0.196* 0.0932
Called Voice4U (1295) (0=no, 1=yes) 1.45 0.5657 0.072 0.0752
Constant 0.43* 0.1747
N 1300
Wald chi-square 213.90
Degree of freedom 29
Sig. Level 0.000
Pseudo R-square 0.1788
57
Stigma and Discrimination
Figures 13 and 14 show the impact of geographic area and program exposure on the probability of EW 
experiencing stigma and discrimination in the previous 12 months, controlling for confounders as described 
in Annex I, tables 43 and 44. EW in no program geographic areas had the highest probability of reporting 
high stigma and discrimination at 61.7%, followed by EW in Flagship TA geographic areas at 51.8%, EW in 
Non-Flagship areas at 42.5%, and EW in CoE geographic areas at 42.5%. 
Program exposure had a significant impact on the probability of an EW experiencing high stigma and 
discrimination. While EW with no exposure had a 59.6% probability of experiencing high stigma and 
discrimination, highly exposed EW had only a 37.8% probability of this, and EW with some exposure had a 
52.2% probability.
Figure 13 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of being in high stigma and discrimination 
***p<0.001
Figure 14 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of being in high stigma and discrimination 
***p<0.001
41.9%
50.8%
42.5%
61.7%
Flagship CoE*** Flagship TA (GF)*** Non-Flagship (GF)*** No Program***
59.6%
52.2%
37.8%
No exposure*** Some exposure*** High exposure***
58
Table 26 shows the results of the binary logistic regression model highlighting the factors associated with 
EW experiencing stigma and discrimination in the previous 12 months as measured by the stigma and 
discrimination index. EW aged 25-30 years of age were less likely to report high stigma and discrimination 
than their younger counterparts (odds ratio 0.666, p = 0.006), as did EW with higher monthly incomes. For 
example, EW with incomes $300 – $500/month were about half as likely to report high stigma and 
discrimination as EW with incomes of less than $150/month (odds ratio 0.564, p = 0.008).
Having had an individual or group meeting with an outreach worker in the past six months greatly reduced 
the likelihood of reporting high stigma and discrimination (odds ratio 0.644, p = 0.021 and odds ratio 0.618, 
p = 0.001, respectively).
Table 26 Binary Logistic Regression Model: The effect of exposure to different activities of 
SMARTGirl program on high stigma and discrimination 
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Age
18-24 years – reference 
25-30 years 0.666** 0.0989 -0.090** 0.0327
31 years and above 0.971 0.1709 -0.007 0.0389
Education
Under 1 year – reference 
1-6 years 1.197 0.2411 0.040 0.0451
7 years and above 0.628* 0.1354 -0.104* 0.0482
Income 
Under $150 – reference 
$150 - $250 0.868 0.1823 -0.031 0.0461
$250 - $300 0.607* 0.1514 -0.111* 0.0551
$300 - $500 0.564** 0.1220 -0.127** 0.0477
$500 and above 0.659 0.1577 -0.093 0.0530
Marital status 
Single - reference
Having partner/boyfriend 1.0721 0.290 0.0153 0.290
Married(having husband) 0.892 0.2556 -0.025 0.0630
Divorced/separated/widowed 0.914 0.2096 -0.020 0.0504
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.638* 0.1168 -0.100* 0.0409
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.962 0.1991 -0.009 0.0457
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.234 0.2074 0.046 0.0362
Disclosed status as sex worker (0=no, 1=yes) 1.014 0.1229 0.003 0.0266
Ever suspected contracting HIV (0=no, 1=yes) 1.253 0.1623 0.050 0.0283
Risk index
Low risk – reference 
Medium risk 0.965 0.1361 -0.008 0.0310
High risk 1.004 0.1500 0.001 0.0329
Individual meeting with OW in the past 6 months (0=no, 1=yes) 0.644* 0.1224 -0.097* 0.0415
Group meeting with OW in the past 6 months (0=no, 1=yes) 0.618* 0.0927 -0.106** 0.0324
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 0.616 0.1695 -0.106 0.0601
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 0.869 0.1349 -0.031 0.0343
High exposure 0.862 0.1523 -0.033 0.0393
Received a copy of the SMARTgirl service directory guide (0=no, 1=yes) 0.634* 0.1242 -0.100* 0.0427
Visited SMARTgirl Khmer website (0=no, 1=yes) 1.729 1.1669 0.120 0.1482
Visited SMARTgirl Facebook (0=no, 1=yes) 0.385* 0.1868 -0.210* 0.1059
Called Voice4U (1295) (0=no, 1=yes) 1.928 0.7610 0.144 0.0865
Constant 2.920** 1.0251
N=1300, Wald chi-square=127.99, df=27, Sig. Level=0.000, Pseudo R-square=0.0920
59
Condom Use 
Table 27 shows the results of the binary logistic regression model of the factors associated with EW 
reporting use of a condom at their last sex with a client. Overall, geographic areas with programs appeared 
to have a lower likelihood of EW condom use at the last sex with a client, EW in Flagship TA areas showed a 
statistically significant lower likelihood of condom use (odds ratio 0.474, p = 0.002) when compared to the 
non-program areas. 
Interestingly, EW with monthly incomes of $500 or more had more than twice the likelihood of using 
condoms at last sex with a client (odds ratio 2.218, p = 0.025). Compared to single women, EW that had a 
partner/boyfriend or that were divorced/separated/widowed were less likely to use a condom (odds ratio 
0.262, p = 0.004 and odds ratio 0.404, p = 0.048, respectively). EW that had been engaged in sex work for 
12 months or more and EW that had disclosed their status as sex worker were nearly twice as likely to have 
reported condom use (odds ratio 1.864, p = 0.008 and odds ratio 1.784, p = 0.001, respectively).
Table 27 Binary Logistic Regression Model: The effect of geographic area on condom use at last sex 
with client
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Geographic area
No program – reference
Flagship CoE 0.869 0.22485 -0.014 0.02503
Flagship TA (GF) 0.474** 0.11311 -0.088** 0.02806
Non-Flagship (GF) 0.720 0.19341 -0.034 0.02801
Age
18-24 years – reference 
25-30 years 0.938 0.19500 -0.007 0.02409
31 years and above 1.299 0.32189 0.027 0.02568
Education
Under 1 year – reference 
1-6 years 0.551 0.18896 -0.056* 0.02757
7 years and above 0.522 0.18622 -0.062* 0.02989
Income 
Under $150 – reference 
$150 - $250 1.228 0.35759 0.025 0.03682
$250 - $300 0.684 0.21298 -0.055 0.04412
$300 - $500 1.549 0.46921 0.050 0.03669
$500 and above 2.218* 0.78737 0.081* 0.03787
Marital status 
Single - reference
Having partner/boyfriend 0.262** 0.12018 -0.134*** 0.03298
Married (having husband) 2.325 1.58276 0.035 0.02942
Divorced/separated/widowed 0.404* 0.18538 -0.077** 0.02922
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.066 0.26121 0.007 0.02734
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.731 0.20532 -0.035 0.03178
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.864** 0.43719 0.073* 0.02871
Disclosed status as sex worker (0=no, 1=yes) 1.784** 0.31009 0.064** 0.01916
Ever suspected having STI (0=no, 1=yes) 1.006 0.18582 0.001 0.02052
Ever suspected contracting HIV (0=no, 1=yes) 0.951 0.17743 -0.006 0.02073
Stigma and discrimination index 
Low – reference 
Medium 0.930 0.19492 -0.008 0.02343
High 0.990 0.21194 -0.001 0.02347
N=1300, Wald chi-square=76.20, df=22, Sig. Level=0.000, Pseudo R-square=0.0831
60
Table 28 shows that the level of program exposure had no impact on EW use of a condom at their last sex 
with a client. 
Table 28 Binary Logistic Regression Model: The effect of program exposure on condom use at last 
sex with client
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Program exposure
No exposure – reference
Some exposure 1.057 0.34060 0.006 0.03484
High exposure 0.960 0.18628 -0.005 0.02197
Age
18-24 years – reference 
25-30 years 0.947 0.19534 -0.006 0.02423
31 years and above 1.319 0.32677 0.029 0.02590
Education
Under 1 year – reference 
1-6 years 0.537 0.18648 -0.058 0.02758
7 years and above 0.508 0.18165 -0.065* 0.02957
Income 
Under $150 – reference 
$150 - $250 1.281 0.37411 0.030 0.03678
$250 - $300 0.685 0.21510 -0.055 0.04476
$300 - $500 1.438 0.43456 0.042 0.03718
$500 and above 2.046* 0.72561 0.075 0.03860
Marital status 
Single - reference
Having partner/boyfriend 0.269** 0.12365 -0.129*** 0.03239
Married (having husband) 2.225 1.50892 0.033 0.02897
Divorced/separated/widowed 0.386* 0.17599 -0.081** 0.02869
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.942 0.23019 -0.007 0.02735
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.813 0.23039 -0.023 0.03223
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.776* 0.41943 0.068* 0.02906
Disclosed status as sex worker (0=no, 1=yes) 1.762** 0.30430 0.064** 0.01931
Ever suspected having STI (0=no, 1=yes) 1.004 0.18469 0.000 0.02067
Ever suspected contracting HIV (0=no, 1=yes) 0.910 0.16745 -0.011 0.02069
Stigma and discrimination index 
Low – reference 
Medium 0.927 0.19139 -0.009 0.02337
High 0.989 0.21269 -0.001 0.02383
N=1300, Wald chi-square=67.43, df=21, Sig. Level=0.000, Pseudo R-square=0.0723
Table 29 shows the impact of different program components on EW use of a condom at their last sex with a 
client. No statistically significant impacts were seen with regard to individual or group meetings with 
outreach workers, visits to SMARTgirl club nor receipt of the SMARTgirl service directory. Having received a 
training or demonstration on condom use increased the likelihood of condom use (odds ratio 1.741, p = 
0.012) as did exposure to outreach education printed materials (for EW with "some exposure" odds ratio 
1.728, p = 0.027). There were strong and statistically significant relationships based on where EW got their 
last condom supply. Compared to EW that got condoms from a SMARTgirl outreach worker, those that got 
their last condoms from a "Guesthouse/brothel/massage parlor/Karaoke/Spa/Sauna/Beer garden" or 
bought them from "Store/gas station/vendor/ pharmacy" were three times more likely to use a condom 
61
with their last client (odds ratio 3.566, p = 0.000 and odds ratio 3.105, p = 0.002, respectively), and those 
that got condoms from their client were six times more likely to use a condom with their last client (odds 
ratio 6.269, p = 0.000).
Table 29 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl
program on condom use at last sex with client 
Variable Odds ratio Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Age
18-24 years – reference 
25-30 years 0.767 0.17888 -0.025 0.02173
31 years and above 1.109 0.33917 0.009 0.02624
Education
Under 1 year – reference 
1-6 years 0.536 0.22516 -0.049 0.02886
7 years and above 0.474 0.20336 -0.061* 0.03052
Income 
Under $150 – reference 
$150 - $250 1.473 0.48119 0.039 0.03425
$250 - $300 0.732 0.25638 -0.037 0.04071
$300 - $500 1.589 0.53094 0.046 0.03461
$500 and above 2.675* 1.08951 0.085* 0.03635
Marital status 
Single - reference
Having partner/boyfriend 0.400 0.18976 -0.079* 0.03459
Married(having husband) 3.533 2.63325 0.056 0.03354
Divorced/separated/widowed 0.474 0.22366 -0.061 0.03264
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.050 0.27993 0.005 0.02469
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.779 0.25018 -0.023 0.02981
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.376 0.37175 0.030 0.02617
Disclosed status as sex worker (0=no, 1=yes) 1.908** 0.37229 0.060** 0.01754
Ever suspected having STI (0=no, 1=yes) 0.976 0.20845 -0.002 0.01972
Ever suspected contracting HIV (0=no, 1=yes) 0.829 0.16766 -0.017 0.01871
Individual meeting with OW in the past 6 months (0=no, 1=yes) 1.317 0.40968 0.025 0.02863
Group meeting with OW in the past 6 months (0=no, 1=yes) 0.830 0.22893 -0.017 0.02546
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 0.840 0.35063 -0.016 0.03850
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 1.728* 0.42700 0.052* 0.02427
High exposure 1.422 0.41356 0.035 0.02921
Received a copy of the SMARTgirl service directory guide (0=no, 1=yes) 1.161 0.37041 0.014 0.02943
Visited SMARTgirl Khmer website (0=no, 1=yes) 3.690 4.05760 0.120 0.10159
Visited SMARTgirl Facebook (0=no, 1=yes) 0.289* 0.15498 -0.115* 0.04917
Called Voice4U (1295) (0=no, 1=yes) 2.183 1.39881 0.072 0.05908
Received training/demonstration on condom use (0=never, 1=ever) 1.741* 0.38373 0.051* 0.02019
Place to obtain condom in last time
SMARTgirl OW – reference 
SMARTgirl club 1.046 0.81224 0.007 0.11376
Street-based sale 5.530 6.18640 0.160* 0.06593
NGO/outreach worker/DIC (not SMARTgirl) 0.185* 0.15256 -0.323 0.16638
Store/gas station/vendor/ pharmacy 3.105** 1.15856 0.125** 0.04556
Guesthouse/brothel/massage parlor/Karaoke/Spa/Sauna/Beer garden 3.566*** 1.22031 0.135** 0.04338
Client 6.269*** 2.75583 0.166*** 0.04406
Sexual partner/sweetheart 0.460* 0.15959 -0.135* 0.05838
Friend 2.334 2.06021 0.102 0.08763
Family health clinic/health center 1.151 0.71471 0.020 0.08781
Other 0.094*** 0.05439 -0.458*** 0.10265
N=1300, Wald chi-square=180.39, df=37, Sig. Level=0.000, Pseudo R-square=0.2299
62
Referrals for STI Screening and Treatment
Table 30 shows that, based on the results of the binary logistic regression model, EW in the three 
geographic areas served by the programs were overwhelmingly more likely to have a referral for STI
screening/treatment in the previous six months than those in no program areas, with odds ratios ranging 
from 7.580 to 9.704. These differences were statistically significant, p = 0.000.
Table 30 Binary Logistic Regression Model: The effect of geographic area on referral for STI 
screening and treatment in the last 6 months
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Geographic area
No program – reference
Flagship CoE 8.962*** 3.76244 0.133*** 0.02140
Flagship TA (GF) 7.580*** 3.28160 0.113*** 0.02206
Non-Flagship (GF) 9.704*** 4.18297 0.143*** 0.02323
Age
18-24 years – reference 
25-30 years 1.215 0.28299 0.019 0.02180
31 years and above 0.973 0.28019 -0.002 0.02548
Education
Under 1 year – reference 
1-6 years 0.481* 0.13718 -0.080* 0.03542
7 years and above 0.527* 0.16331 -0.072 0.03770
Income 
Under $150 – reference 
$150 - $250 1.259 0.40847 0.023 0.03125
$250 - $300 0.648 0.26925 -0.035 0.03359
$300 - $500 1.057 0.34991 0.005 0.03095
$500 and above 0.928 0.34495 -0.007 0.03380
Marital status 
Single - reference
Having partner/boyfriend 1.221 0.49382 0.016 0.03158
Married (having husband) 1.866 0.79435 0.058 0.03731
Divorced/separated/widowed 1.432 0.56063 0.031 0.03068
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.606 0.17499 -0.050 0.03007
Duration working in current workplace
Less than 12 months – reference
12 months and above 2.128* 0.62639 0.069** 0.02617
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.652 0.45964 0.044 0.02225
Disclosed status as sex worker (0=no, 1=yes) 0.948 0.17309 -0.005 0.01708
Ever suspected having STI (0=no, 1=yes) 1.486* 0.28838 0.037* 0.01811
Risk index
Low risk – reference 
Medium risk 1.433 0.31228 0.034 0.02054
High risk 1.228 0.28516 0.018 0.02097
Stigma and discrimination index 
Low – reference 
Medium 0.686 0.14616 -0.039 0.02173
High 0.431** 0.10571 -0.075*** 0.02069
Constant 0.016*** 0.01140
N 1300
Wald chi-square 100.32
Degree of freedom 23
Sig. Level 0.000
Pseudo R-square 0.1393
63
Table 31 shows that, based on the results of the binary logistic regression model, exposure to the program 
strongly impacted referral for STI screening/treatment in the last six months. The odds ratio for EW with a 
medium level of program exposure receiving a referral for STI screening/treatment was 3.036 (p = 0.029) 
and the odds ratio for EW with a high level of exposure was 18.885 (p = 0.000).
Table 31 Binary Logistic Regression Model: The effect of program exposure on referral for STI 
screening and treatment in the last 6 months
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Geographic area
No exposure – reference
Medium exposure 3.036* 1.54331 0.033 0.01927
High exposure 18.885*** 6.83641 0.220*** 0.02126
Age
18-24 years – reference 
25-30 years 1.119 0.26605 0.010 0.02041
31 years and above 0.977 0.29231 -0.002 0.02494
Education
Under 1 year – reference 
1-6 years 0.597 0.17600 -0.049 0.02984
7 years and above 0.593 0.18584 -0.049 0.03119
Income 
Under $150 – reference 
$150 - $250 1.179 0.38870 0.015 0.02989
$250 - $300 0.593 0.25932 -0.041 0.03363
$300 - $500 0.922 0.32205 -0.007 0.03066
$500 and above 0.841 0.32214 -0.015 0.03286
Marital status 
Single - reference
Having partner/boyfriend 1.193 0.50335 0.013 0.03114
Married (having husband) 2.167 0.95707 0.069 0.03673
Divorced/separated/widowed 1.395 0.55830 0.026 0.02954
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.574 0.21841 -0.050 0.03499
Duration working in current workplace
Less than 12 months – reference
12 months and above 2.260* 0.83637 0.068* 0.02952
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.248 0.35689 0.018 0.02310
Disclosed status as sex worker (0=no, 1=yes) 0.932 0.17693 -0.006 0.01626
Ever suspected having STI (0=no, 1=yes) 1.472 0.29802 0.033 0.01726
Risk index
Low risk – reference 
Medium risk 1.277 0.29004 0.021 0.01962
High risk 1.107 0.26801 0.008 0.02023
Stigma and discrimination index 
Low – reference 
Medium 0.740 0.16353 -0.027 0.01944
High 0.619 0.15970 -0.041 0.02107
Constant 0.016*** 0.01007
N 1300
Wald chi-square 140.92
Degree of freedom 22
Sig. Level 0.000
Pseudo R-square 0.2336
64
Table 32 shows that of the SMARTgirl program elements, having individual or group meeting with outreach 
workers in the previous six months strongly predicted receiving a referral for STI screening/treatment in the 
previous six months (odds ratio 3.288, p=0.000 and odds ratio 4.295, p=0.000). In addition, having visited a 
SMARTgirl club in the previous six months was strongly associated with receipt of a referral for STI
screening/treatment (odds ratio 2.087, p=0,007), as was having had a higher level of exposure to outreach 
education printed materials (odds ratio 6.634, p=0.001). 
Table 32 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl
program on referral for STI screening and treatment in the last 6 months
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Age
18-24 years – reference 
25-30 years 1.151 0.30210 0.011 0.01956
31 years and above 0.919 0.30032 -0.006 0.02328
Education
Under 1 year – reference 
1-6 years 0.512* 0.16496 -0.056 0.02922
7 years and above 0.479* 0.16723 -0.061* 0.03073
Income 
Under $150 – reference 
$150 - $250 1.279 0.47148 0.020 0.02856
$250 - $300 0.542 0.25977 -0.040 0.03152
$300 - $500 0.894 0.35961 -0.008 0.02994
$500 and above 0.870 0.37327 -0.010 0.03165
Marital status 
Single - reference
Having partner/boyfriend 1.196 0.51296 0.012 0.02858
Married(having husband) 2.004 0.95162 0.053 0.03546
Divorced/separated/widowed 1.316 0.54748 0.019 0.02772
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.498 0.21105 -0.054 0.03358
Duration working in current workplace
Less than 12 months – reference
12 months and above 2.151 0.85779 0.056 0.02867
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.351 0.41083 0.022 0.02118
Disclosed status as sex worker (0=no, 1=yes) 0.845 0.17949 -0.012 0.01569
Ever suspected having STI (0=no, 1=yes) 1.335 0.29513 0.021 0.01629
Risk index
Low risk – reference 
Medium risk 1.262 0.31170 0.017 0.01855
High risk 1.039 0.27253 0.003 0.01881
Stigma and discrimination index 
Low – reference 
Medium 0.909 0.21751 -0.007 0.01796
High 0.832 0.23458 -0.014 0.02061
Individual meeting with OW in the past 6 months (0=no, 1=yes) 3.288*** 0.76296 0.088*** 0.01643
Group meeting with OW in the past 6 months (0=no, 1=yes) 4.295*** 1.06238 0.108*** 0.01746
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 2.082** 0.56550 0.054** 0.01961
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 2.834 1.59159 0.052* 0.02427
High exposure 6.634** 3.67000 0.035 0.02921
65
Table 32 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on referral for STI screening and treatment in the last 6 months (Continued)
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust 
Std. Err.
Received a copy of the SMARTgirl service directory guide (0=no, 
1=yes)
1.046 0.26281 0.014 0.02943
Visited SMARTgirl Khmer website (0=no, 1=yes) 5.907** 3.95587 0.120 0.10159
Visited SMARTgirl Facebook (0=no, 1=yes) 1.364 0.73465 -0.115* 0.04917
Called Voice4U (1295) (0=no, 1=yes) 0.341* 0.16302 0.072 0.05908
Constant 0.010*** 0.00772
N 1300
Wald chi-square 210.14
Degree of freedom 29
Sig. Level 0.0000
Pseudo R-square 0.3183
HIV/STI Prevention Knowledge 
Table 33 shows that geographical location within the catchment of the Flagship Centre of Excellence 
measurably impacted knowledge of HIV/STI prevention among EW (odds ratio 1.736, p = 0.001). Older EW 
were also more likely to have a high level of prevention knowledge (ages 25-30: odds ratio 1.736, p = 0.001 
and ages >30: odds ratio 2.919, p = 0.000) as were EW engaged in sex work for 12 month or more (odds 
ratio 1.482, p=0.015).
Table 33 Binary Logistic Regression Model: The effect of geographic area on high prevention 
knowledge
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Geographic area
No program – reference
Flagship CoE 1.736** 0.29399 0.126** 0.03871
Flagship TA (GF) 1.059 0.17982 0.013 0.03822
Non-Flagship (GF) 1.177 0.21022 0.037 0.04053
Age
18-24 years – reference 
25-30 years 1.619** 0.23679 0.110** 0.03307
31 years and above 2.919*** 0.51114 0.250*** 0.03964
Education
Under 1 year – reference 
1-6 years 0.938 0.19322 -0.014 0.04626
7 years and above 1.247 0.27641 0.050 0.04982
Income 
Under $150 – reference 
$150 - $250 0.620* 0.13021 -0.109* 0.04779
$250 - $300 0.741 0.18441 -0.069 0.05688
$300 - $500 0.691 0.14961 -0.085 0.04944
$500 and above 0.782 0.19214 -0.056 0.05618
Marital status 
Single - reference
Having partner/boyfriend 1.239 0.31694 0.047 0.05583
Married (having husband) 1.717 0.50341 0.122 0.06531
Divorced/separated/widowed 1.516 0.37345 0.093 0.05389
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.404 0.26488 0.077 0.04296
66
Table 33 Binary Logistic Regression Model: The effect of geographic area on high prevention 
knowledge (Continued) 
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust Std. 
Err.
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.688 0.14110 -0.082 0.04379
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.482* 0.24095 0.090* 0.03690
Disclosed status as sex worker (0=no, 1=yes) 1.005 0.12028 0.001 0.02705
Ever suspected having STI (0=no, 1=yes) 1.189 0.15561 0.039 0.02950
Ever suspected contracting HIV (0=no, 1=yes) 0.861 0.11280 -0.034 0.02954
Stigma and discrimination index 
Low – reference 
Medium 0.869 0.12636 -0.032 0.03315
High 0.777 0.11523 -0.057 0.03366
Constant 0.349** 0.13434
N 1300
Wald chi-square 103.20
Degree of freedom 22
Sig. Level 0.000
Pseudo R-square 0.0648
Table 34 shows that high exposure to the program strongly impacted having high HIV/STI prevention 
knowledge among EW (odds ratio 1.886, p = 0.000).
Table 34 Binary Logistic Regression Model: The effect of program exposure on high prevention 
knowledge
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust 
Std. Err.
Program exposure
No exposure – reference
Some exposure 1.365 0.27257 0.071 0.04584
High exposure 1.886*** 0.25790 0.146*** 0.03135
Age
18-24 years – reference 
25-30 years 1.602** 0.23402 0.107** 0.03286
31 years and above 2.988*** 0.52580 0.254*** 0.03947
Education
Under 1 year – reference 
1-6 years 0.968 0.19634 -0.007 0.04520
7 years and above 1.237 0.26998 0.048 0.04866
Income 
Under $150 – reference 
$150 - $250 0.617* 0.12919 -0.109* 0.04729
$250 - $300 0.725 0.17979 -0.073 0.05623
$300 - $500 0.655* 0.14094 -0.096* 0.04868
$500 and above 0.732 0.17900 -0.071 0.05540
Marital status 
Single - reference
Having partner/boyfriend 1.294 0.32414 0.057 0.05432
Married (having husband) 1.687 0.48901 0.117 0.06407
Divorced/separated/widowed 1.476 0.35542 0.086 0.05230
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.383 0.25737 0.073 0.04210
67
Table 34 Binary Logistic Regression Model: The effect of program exposure on high prevention 
knowledge (Continued) 
Variable Odds ratio Robust Std. 
Err.
Predicted 
Probability 
Robust 
Std. Err.
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.678 0.13834 -0.085* 0.04309
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.376 0.22586 0.072 0.03712
Disclosed status as sex worker (0=no, 1=yes) 1.004 0.12006 0.001 0.02682
Ever suspected having STI (0=no, 1=yes) 1.196 0.15563 0.040 0.02910
Ever suspected contracting HIV (0=no, 1=yes) 0.832 0.10860 -0.041 0.02920
Stigma and discrimination index 
Low – reference 
Medium 0.891 0.13093 -0.026 0.03314
High 0.859 0.12937 -0.034 0.03397
Constant 0.330** 0.12515
N 1300
Wald chi-square 112.96
Degree of freedom 21
Sig. Level 0.000
Pseudo R-square 0.0702
Table 35 shows that of the SMARTgirl program elements, only one was found to have an impact on 
measurable high prevention knowledge among EW: having received a training or demonstration on 
condom use increased the likelihood of condom use (odds ratio 2.650, p = 0.000). Notably, no statistically 
significant impacts were seen in meeting with SMARTgirl outreach workers or exposure to printed 
materials. 
Table 35 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl
program on high prevention knowledge 
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust Std. 
Err.
Age
18-24 years – reference 
25-30 years 1.424* 0.22092 0.073* 0.03208
31 years and above 2.403*** 0.46097 0.184*** 0.04038
Education
Under 1 year – reference 
1-6 years 0.946 0.20563 -0.011 0.04390
7 years and above 1.173 0.27432 0.032 0.04733
Income 
Under $150 – reference 
$150 - $250 0.651* 0.13961 -0.088* 0.04380
$250 - $300 0.825 0.21143 -0.039 0.05256
$300 - $500 0.648 0.14553 -0.088 0.04575
$500 and above 0.809 0.20460 -0.043 0.05189
Marital status 
Single - reference
Having partner/boyfriend 1.212 0.30057 0.039 0.04933
Married(having husband) 1.566 0.45215 0.091 0.05836
Divorced/separated/widowed 1.419 0.32940 0.071 0.04631
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.268 0.24994 0.049 0.04040
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.692 0.15016 -0.073 0.04186
68
Table 35 Binary Logistic Regression Model: The effect of exposure to different activities of SMARTgirl 
program on high prevention knowledge (Continued) 
Variable Odds 
ratio 
Robust 
Std. Err.
Predicted 
Probability 
Robust Std. 
Err.
Duration engaged in sex work
Less than 12 months – reference
12 months and above 1.143 0.19878 0.027 0.03554
Disclosed status as sex worker (0=no, 1=yes) 0.917 0.11700 -0.018 0.02583
Ever suspected having STI (0=no, 1=yes) 1.113 0.15788 0.022 0.02875
Ever suspected contracting HIV (0=no, 1=yes) 0.805 0.11263 -0.044 0.02823
Individual meeting with OW in the past 6 months (0=no, 1=yes) 0.948 0.18327 -0.011 0.03917
Group meeting with OW in the past 6 months (0=no, 1=yes) 1.271 0.20492 0.049 0.03257
Visited SMARTgirl club in the past 6 months (0=no, 1=yes) 0.849 0.22732 -0.033 0.05422
Exposure to outreach education printed materials 
No exposure – reference 
Some exposure 0.850 0.14035 -0.033 0.03377
High exposure 1.329 0.25699 0.059 0.04073
Received a copy of the SMARTgirl service directory guide (0=no, 1=yes) 0.771 0.15309 -0.053 0.04022
Visited SMARTgirl Khmer website (0=no, 1=yes) 4.544 4.58048 0.307 0.20340
Visited SMARTgirl Facebook (0=no, 1=yes) 1.251 0.50031 0.045 0.08099
Called Voice4U (1295) (0=no, 1=yes) 1.933 0.80393 0.134 0.08389
Received training/demonstration on condom use (0=never, 1=ever) 2.650*** 0.36803 0.198*** 0.02613
Place to obtain condom in last time
SMARTgirl OW – reference 
SMARTgirl club 0.826 0.61757 -0.041 0.16065
Street-based sale 0.357 0.26429 -0.215 0.14379
NGO/outreach worker/DIC (not SMARTgirl) 0.188* 0.13110 -0.325** 0.11135
Store/gas station/vendor/ pharmacy 0.897 0.23734 -0.023 0.05673
Guesthouse/brothel/massage parlor/Karaoke/Spa/Sauna/Beer garden 0.713 0.17696 -0.073 0.05335
Client 0.383*** 0.10507 -0.201** 0.05778
Sexual partner/sweetheart 0.428** 0.12587 -0.179** 0.06193
Friend 0.860 0.45292 -0.032 0.11321
Family health clinic/health center 0.603 0.25485 -0.108 0.08979
Other 0.419 0.21984 -0.183 0.10647
Constant 0.468 0.20127
N 1300
Wald chi-square 201.13
Degree of freedom 37
Sig. Level 0.000
Pseudo R-square 0.1403
69
15. Cost Allocation
As shown in figure 15, there was wide variation in the total cost across program types, with the annual cost 
of the Flagship CoE areas ($601,617) approximately 5 times the annual cost of Flagship TA areas ($120,764), 
and approximately 8 times the annual cost of the Non-Flagship areas ($76,525). Description of cost 
allocation categories are found in Annex II. Omitting the higher level costs (TA and central office costs), the 
total cost in CoE areas would be approximately $259,064, compared to $88,953 in Flagship TA areas, and 
$51,663 in Non-Flagship areas.
Figure 15 Cost allocation 
Flagship COE Flagship TA Non-Flagship
IP field office - Other 4,849 813 667
IP field office - referral to services 4,496 1,974 2,197
IP field office - monitoring & evaluation 3,191 1,247 624
IP field office - communication 3,750 1,143 893
IP field office - transportation 1,273 1,154 330
IP field office - meeting (field staff and
outreach worker) 12,586 3,658 2,793
IP field office - capacity development 13,345 3,400 1,611
SMARTgirl Club/DIC 31,263 4,923
IP field office - administration 16,096 5,593 7,619
IP field office - program
management/support 129,908 39,987 24,442
IP field office - outreach staff 38,306 25,061 10,488
IP central office - administration 90,326 5,788 9,232
IP central office - program management 35,372 9,662 11,578
Global Fund - KHANA 4,053
TA 16,361
Flagship 216,855
0
100,000
200,000
300,000
400,000
500,000
600,000
Annual Cost
70
Figure 16 shows that the cost components were also markedly different across program types. Among the 
largest common cost categories, program management/support at a field office level accounted for 22% of 
CoE program costs, 33% of Flagship TA costs, and 32% of Non-Flagship costs. Another major component 
was outreach staff, which for CoE, Flagship TA, and Non-Flagship areas accounted for 6%, 21%, and 14% of 
the program cost, respectively.
Figure 16 Cost allocation of programs
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Flagship COE Flagship TA Non-Flagship
IP field office - Other
IP field office - referral to services
IP field office - monitoring & evaluation
IP field office - communication
IP field office - transportation
IP field office - meeting (field staff and
outreach worker)
IP field office - capacity development
SMARTgirl Club/DIC
IP field office - administration
IP field office - program
management/support
IP field office - outreach staff
IP central office - administration
IP central office - program management
Global Fund - KHANA
TA
Flagship
71
While total program costs were much higher for CoE than for Flagship TA and Non-Flagship sites, the 
number of HIV tests and the number of HIV cases detected at CoE were also significantly higher. See table 
36 and figure 17. Over the 12 month period, there were nearly 2.5 times the number of HIV tests of EW in 
Flagship CoE areas as in Flagship TA areas (14,870 versus 5,808), and approximately 5 times the number of 
HIV tests as in the Non-Flagship areas (2,550). Over the same period, Flagship CoE areas found more than 
three times the number of HIV cases as did the Flagship TA areas (39 versus 11), and more than 4 times the 
number of HIV cases as did the Non-Flagship areas. 
Despite the greater output of CoE in terms of the number of HIV tests performed and the number of HIV 
cases identified, the significantly higher program costs meant that the unit costs remained higher for CoE 
than Flagship TA and Non-Flagship areas. The cost per HIV test in CoE areas was approximately $40, 
compared to $21 at Flagship TA areas, and $30 at Non-Flagship areas. Similarly, the cost per HIV case 
detected in CoE areas was approximately $15,426, compared to $10,979 in Flagship TA areas, and $8,503 in 
Non-Flagship areas. While Flagship TA areas yielded the least expensive cost per test, Non-Flagship areas
yielded the lowest cost per HIV case detected. This inversion was largely because the HIV positivity rate was 
much higher in Non-Flagship locations (0.35%) than in Flagship TA locations (0.19%).
Omitting the higher level costs (TA and central office costs), the cost per HIV test in CoE areas would be 
approximately $17, compared to $15 in Flagship TA areas, and $20 in Non-Flagship areas; the cost per HIV 
case detected in CoE areas would be $6,643, $8087 in Flagship TA areas, and $5,740 in Non-Flagship areas.
Table 36 Annual unit costs 
Analysis Category Flagship COE Flagship TA Non-Flagship
Total Cost $601,617 $120,764 $76,525
HIV tests 14,870 5,808 2,550
HIV positive cases 39 11 9
Cost per HIV test $40.46 $20.79 $30.01
Cost per HIV case detected $15,426 $10,979 $8,503
Figure 17 Numbers of HIV tests and positive cases; unit costs per HIV test and per case detected
72
16. Discussion 
The data in this evaluation cover 1300 respondents in 4 geographic locations, comprising 300 from the 
geographic areas of the Flagship CoE sites, 300 from the geographic areas of Flagship TA sites, 300 from the 
geographic areas of Non-Flagship sites, and 400 from geographic areas with no existing EW programs. 
While it is not clear whether this was the result of programmatic targeting or the result of existing 
geographic and demographic realities, EW reached in CoE areas were older and more highly educated than 
the average for the group, and EW in Non-Flagship were younger and less well educated, which 
demonstrated some of the variabilities by program and geographic area. Additionally, there were larger 
populations in the catchment areas of Phnom Penh and Siem Reap where the CoE were located, which may 
have affected some of the findings, e.g. the number of HIV tests in the different catchment areas. There is 
no way to control for these potential effects. 
Employment 
KTV was the dominant primary employer of EW (62%) and comparatively small proportions of EW reported 
their main occupations to be either freelance sex worker (3.1%) or residential sex worker (3.4%). Given that 
the data from this evaluation were population-based (as opposed to time/location-based), this 
demonstrates the continued importance of targeting entertainment venues to reach sex workers.
Overall, about 57% of EW had incomes greater than $250 per month. With the monthly minimum wage in 
Cambodia for garment workers currently $140 per month (Daily, 2016), the significantly higher incomes 
available to EW show that there continues to be a significant economic incentive for women to participate 
in entertainment work. 
Mobility 
The high level of geographic mobility (43% of EW with less than one year living in current location) and 
employment mobility (52% of EW with less than one year at current work place) showed the challenges 
inherent in reaching EW. Particularly because program exposure was shown to positively impact uptake of 
STI screening, HCT, HIV prevention knowledge and, stigma and discrimination, consistent repeated program 
outreach to EW populations seems to have been valuable. 
Risk
57% of EW were at medium or high risk for HIV, as judged by the risk assessment index. None of the 
programs, however, appeared to successfully target high risk EW, with all three of the program areas 
predominantly reaching a low and medium risk EW (74%). Indeed, a larger proportion of EW with low risk 
(37%) than with high risk (28%) had a high level of exposure to the program, showing the challenges 
experienced in reaching highest risk EW.
Only 7.5% of EW had seven or more sexual partners per week, though 82% usually had at least one sexual 
partner per week, indicating ongoing risk exposure. The 2016 Integrated HIV Bio-Behavioral Surveillance 
(IBBS) indicated that average number of clients in the previous week among those who reported paid sex 
was 3.3 and 61% of EW reported sex in exchange for gift or money in the previous 12 months (FEWIBBS, 
2016) 
Program Exposure
Only about half (51%) of EW in CoE geographic areas were familiar with the SMARTgirl logo, and the same 
proportion reported contact with an outreach worker in the previous 12 months. Only 34% of EW in CoE 
areas had contact with an outreach worker in the previous three months, which was actually lower than 
the proportions of EW that had contact with an outreach worker in the previous three months in Flagship 
TA and Non-Flagship areas (37% and 43%, respectively). This, despite EW in these latter areas having more 
mobile EW populations, indicating a potentially lower level of program intensity in the CoE areas. 
The majority (52-65%) of EW in program areas reported that OW were their main source of information 
about HIV and STI services, though only 3-36% of EW said that OW were their preferred channel for this 
type of information. By contrast, there appeared to be strong demand for broadcast media (television and 
73
radio) which combined for 52% of responses by EW as their preferred channels for information about HIV 
and STI services. 
A solid one fifth of EW (22%) stated that Facebook was their preferred communication channel. This 
apparent demand for social media contrasts with the reported underutilization of the SMARTgirl website 
(only 3% utilization among EW in CoE areas) and the SMARTgirl Facebook page (only 7% utilization among 
EW in CoE areas). This underutilization is partially explained by low levels of knowledge about these 
resources with only 16% and 22% of EW reporting any knowledge of the website and Facebook page, 
respectively. 
Exposure to printed education materials was suboptimal, with only four of the 20 printed materials being 
well recognized by the EW interviewed. Furthermore, there was no measurable impact on prevention 
knowledge among EW with regard to exposure to printed educational materials.
There was low utilization of SMARTgirl clubs and drop-in centers. The majority of EW (61%) in CoE areas 
had heard about the SMARTgirl club but only 20% had visited the SMARTgirl club in the previous six 
months. Less than 3% of EW in Flagship TA and Non-Flagship areas had visited a drop-in center in the 
previous six months. Visits to the SMARTgirl club did not correlate well with prevention knowledge, 
condom use, stigma/discrimination, STI screening, or HTC uptake.
Utilization of referrals was also suboptimal, with less than 10% of EW in program areas having been 
referred for family planning. Indeed, small proportions of EW were referred for STI services in the previous 
12 months (13-18% in the three program geographic areas). Even for highly exposed EW, only 26% had 
been referred for STI screening/treatment in the previous 12 months, demonstrating the relative weakness 
of this program component.
Condoms and program/intervention exposure
Though only 74% of respondents reported using a condom at last sex with any partner, 97% reported using 
a condom at their last sex with a client. No differences were seen by location, so even EW in non-program 
areas had very high condom use rates with clients. This speaks well about the culture of condom use 
among Cambodian EW in their professional setting, as well as the wide availability of condoms (86% of EW 
reported condoms were always available when needed). The success of the CUP is seen in Cambodia with 
over 90% of EW reporting use of condom with the most recent paid clients (FEWIBBS, 2016). However, the 
most common reasons for not using a condom were that the EW trusted their partners and that not using a 
condom demonstrated fidelity to their partner. Among the small number of EW that did not use a condom 
at the last sex act with a client (37 of 1137 respondents), 38% said the main reason was that their client 
refused condom use. 
STI Screening & HTC
National guidelines call for STI screening and HTC at least every six months, but only 44% of EW reported 
STI screening in the past six months, with a significantly smaller proportion of EW in non-program 
geographic areas receiving STI screening than EW in the three program geographic areas. By contrast, 56% 
of EW had undergone HTC in the previous six months, again with a significantly lower proportion of EW in 
non-program geographic areas receiving this than EW in the three program areas. 
No meaningful analysis regarding the effectiveness of the EW programs with regard to key HIV status￾related outcomes (e.g. VCCT confirmation testing, identifying new cases, reducing LTFU for HIV testing 
confirmation, or ART enrollment and retention of HIV+ EW) could be made because only 7 EW reported 
being HIV+. In fact, no EW in the CoE or Flagship TA areas reported being HIV+. 
Program Impact
All Programs vs. No Program
Controlling for confounders, the EW programs had a measurable and positive impact on STI 
screening/treatment, HTC, and stigma/discrimination. EW in non-program areas were less likely than EW in 
74
program areas to utilize STI screening/treatment (37% versus 44-47%), less likely to receive HTC (48% 
versus 59-67%), and were more likely to report high stigma/discrimination (62% versus 42-51%). This shows 
powerful evidence of the impact of the EW program. Furthermore, compared to EW with high program 
exposure, EW with no program exposure were less likely to receive STI screening/treatment (33% versus 
62%), were less likely to receive HTC (44% versus 79%), and were more likely to experience high 
stigma/discrimination (60% versus 38%).
Comparisons among the Programs
Comparing program impact across geographic areas, it appears that the Non-Flagship areas performed best 
among the three, with EW in these geographic areas being significantly more likely to receive HTC (any type 
or community-based) and STI screening/treatment than the other two geographic program areas. CoE EW 
programs had the overall lowest impacts among the three (with the exception of STI screening/treatment 
where CoE performance was better than Flagship TA areas).
Cost
Program costs were much higher for CoE than for Flagship TA and Non-Flagship sites, yielding an overall 
higher cost per HIV test of approximately $40, compared to $21 in Flagship TA areas, and $30 in Non￾Flagship areas, as well as a higher cost per HIV case detected in CoE areas of $15,426, compared to $10,979 
in Flagship TA areas and $8,503 in Non-Flagship areas. 
17. Conclusion 
This evaluation provides valuable insights into the situation of EW across Cambodia, showing a great 
diversity of demographic, work, and socioeconomic conditions. Common issues were seen with regard to 
geographic mobility, gender-based and domestic violence, and condom use patterns. 
Uptake of HTC and STI screening/treatment fell below expectations among EW. There was underutilization 
of referrals for family planning, printed materials, SMARTgirl clubs and drop-in centers, as well as social 
media among EW across geographic location. 
However, it was clear that exposure to the EW programs boosted HTC and STI screening/treatment and 
was associated with decreased stigma and discrimination. The highest EW program impact was seen in 
Non-Flagship locations. Compared to CoE and Flagship TA areas, Non-Flagship areas also had the lowest 
cost per HIV infection detected.
75
18. References 
Bourmant, C. (2017). Is Cambodia’s war on drugs working? Phnom Penh Post
Retrieved from http://www.phnompenhpost.com/national/cambodias-war-drugs-working
CDC. (2010). CDC Fact Sheet:The Role of STD Prevention and Treatment in HIV Prevention.Retrieved from
https://www.cdc.gov/std/hiv/stds-and-hiv-fact-sheet-press.pdf
Daily, C. (2016). Government Raises Garment Wage to $153. Retrieved from 
https://www.cambodiadaily.com/news/garment-sector-minimum-wage-set-153-118642/
Female Entertainment WorkersIntegrated HIV Bio-Behavioral Surveillance, (FEWIBBS) (2016). [PowerPoint Slides]. 
Retrieved from Survellance unit, NCHADS
Feng. (2010). HIV/STD prevalence among MSM in Chengdu, China and associated risk factors for HIV infection. J Acquir 
Immune Defic Syndr., 53(suppl 1), S74–S80. 
Field, A. (2013). Discovering statistics uisng IBM SPSS Statistics (4th ed.).
Goel. (2012). Study of syndromic management approach in the management of sexually transmitted diseases in rural 
population. Indian J Sex Transm Dis., 33(2), 146–147. 
Heng, S., Morineau, G., Neal, J.J., Saphonn, V. & Fylkesnes, K. (2008). Sustained high prevanece of sexually transmiited 
infections among female sex workers in Cambodia: high turnover seriously challenges the 100% condom use 
programme. BMC Infectious Diseases, 8, 167. 
HIV Innovate and Evaluate Project (HIEP). (2015a). Evaluation of the SMARTgirl family planning integration program in 
Cambodia. 
HIV Innovate and Evaluate Project (HIEP). (2015b). Patterns of utilization and barriers for key populations to use HIV 
clinical services at health facilities in Cambodia. 
International Labour Organization (ILO). (2011). Cambodia - addressing HIV vulnerabilities of indirect sex workers 
during the financial crisis: Situational analysis, strategies and entry points for HIV/AIDS workplace education. 
Retrieved from, 
http://www.ilo.org/wcmsp5/groups/public/---asia/---ro-bangkok/documents/genericdocument/wcms_165487.pdf 
Kerr. (2016). Sex work and HIV incidence among people who inject drugs. AIDS, 30(4), 627-634
KHANA. (2016a). Coverage areas of GFATM-NFM and USAID Flagship for October 2015-December 2016: SMARTgirl
and non-SMARTgirl programs. Retrieved from,
http://www.khana.org.kh/publicationimages/publican_pdf/Annual_report_2016.pdf
KHANA. (2016b). KHANA Annual Report 2016: Leave No One Behaind: Stay Relevant And Sustained Impacts.
National Centre for HIV/AIDS, Dermatology and STDs (NCHADS). (2013). Standard Operating Procedures (SOP) for 
Boosted Continuum of Prevention to Care and Treatment for Most At Risk Populations in Cambodia: National 
Center for HIV/AID, Dermatology and STD. 
National Centre for HIV/AIDS, Dermatology and STDs (NCHADS). (2017). Standard Operational Procedures on Boosted￾Integrated Active Case Management (B-IACM)-March 2017
Plummer, F.A., Simonsen, J.N., Cameron, D.W., Ndinya-Achola, J.O., Kreiss, J.K., Gakinya, M.N., Waiyaki, P., Cheang, 
M., Piot, P., Ronald, A.R, et al. (1991). Cofactors in male-female sexual transmission of human 
immunodeficiency virus type 1. The Journal of Infectious Diseases, 163 (2), 233-239.
Thomas, JC. & Tucker MJ. (1996). The development and use of the concept of a sexually transmiited diseases core. The 
Journal of Infectious Diseases, 174 (2), S134-143.
Tran. (2005). HIV Infection and Risk Characteristics Among Female Sex Workers in Hanoi, Vietnam. J Acquir Immune 
Defic Syndr., 15(5), 581–586. 
USAID HIV Flagship Project. (2013). Annual progress report: November 14, 2012-September 30, 2013.
USAID HIV Flagship Project. (2014). Annual progress report: October 01, 2013-September 31, 2014.
Ward. (2010). The contribution of STIs to the sexual transmission of HIV. Curr Opin HIV AIDS., 5(4), 305–310
WHO. (2004). EXPERIENCES OF 100% CONDOM USE PROGRAMME IN SELECTED COUNTRIES OF ASIA. Retrieved 
30June, 2017, from http://www.wpro.who.int/publications/docs/100_condom_program_experience.pdf
World Health Organization (WHO). (2013). Joint review of the Cambodian national health sector response to HIV 
2013. Retrieved from, 
http://www.wpro.who.int/hiv/documents/docs/Joint_Programme_Review_Cambodia.pdf
76
19. Annexes 
19.1. Printed Education Materials 
Table 1 Exposure to printed education materials
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Ever seen “Two better than one”
No 75.7% 83.7% 82.3% 96.8% 85.5%
67.621
df=3 
p=.000
Yes 24.3% 16.3% 17.7% 3.3% 14.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 16.4% 18.4% 13.2% 0.0% 14.9%
2.998 df=3 
p=.000
Attractive 83.6% 81.6% 86.8% 100.0% 85.1%
Total 100% 100% 100% 100% 100%
73 49 53 13 188
Ever seen “Your health your choice (Srey 
Ra)” 
No 60.0% 67.3% 61.0% 81.0% 68.4%
46.920 df=3 
p=.000
Yes 40.0% 32.7% 39.0% 19.0% 31.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 10.8% 16.3% 13.7% 6.6% 12.2%
4.257 df=3 
p=.235
Attractive 89.2% 83.7% 86.3% 93.4% 87.8%
Total 100% 100% 100% 100% 100%
120 98 117 76 411
Ever seen “Your decision” 
No 55.0% 65.3% 62.0% 83.0% 67.6%
70.074 df=3 
p=.000
Yes 45.0% 34.7% 38.0% 17.0% 32.4%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 14.1% 15.4% 10.5% 22.1% 14.7%
4.594 df=3 
p=.204
Attractive 85.9% 84.6% 89.5% 77.9% 85.3%
Total 100% 100% 100% 100% 100%
135 104 114 68 421 
Ever seen “Thida and Leakhena” 
No 79.3% 86.7% 86.3% 94.0% 87.2%
33.378 df=3 
p=.000
Yes 20.7% 13.3% 13.7% 6.0% 12.8%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 17.7% 7.5% 12.2% 12.5% 13.2%
2.301 
df=3 
p=.512
Attractive 82.3% 92.5% 87.8% 87.5% 86.8%
Total 100% 100% 100% 100% 100%
62 40 41 24 167
Table 1 Exposure to printed education materials (Continued) 
77
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Ever seen “Smart choice” 
No 60.3% 74.3% 72.3% 93.3% 76.5%
109.595 
df=3 p=.000
Yes 39.7% 25.7% 27.7% 6.8% 23.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 12.6% 6.5% 8.4% 0.0% 8.8%
5.263 df=3 
p=.153
Attractive 87.4% 93.5% 91.6% 100.0% 91.2%
Total 100% 100% 100% 100% 100%
119 77 83 27 306
Ever seen “Secret bag” 
No 80.7% 90.7% 87.7% 95.3% 89.1%
38.867 
df=3 
p=.000
Yes 19.3% 9.3% 12.3% 4.8% 10.9%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 5.2% 25.0% 0.0% 10.5% 8.5%
14.239 df=3 
p=003
Attractive 94.8% 75.0% 100.0% 89.5% 91.5%
Total 100% 100% 100% 100% 100%
58 28 37 19 142
Ever seen “For my future” 
No 88.0% 93.7% 93.0% 94.8% 92.5%
12.428 df=3 
p=.006
Yes 12.0% 6.3% 7.0% 5.3% 7.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 2.8% 26.3% 14.3% 33.3% 16.5%
10.645 df=3 
p=.014
Attractive 97.2% 73.7% 85.7% 66.7% 83.5%
Total 100% 100% 100% 100% 100%
36 19 21 21 97
Ever seen “Value of life” 
No 73.3% 78.7% 77.0% 90.5% 80.7%
38.539 df=3 
p=.000
Yes 26.7% 21.3% 23.0% 9.5% 19.3%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 13.8% 17.2% 17.4% 10.5% 15.1%
1.231 df=3 
p=746
Attractive 86.3% 82.8% 82.6% 89.5% 84.9%
Total 100% 100% 100% 100% 100%
80 64 69 38 251
Ever seen “Your choice”
No 59.7% 66.7% 67.7% 85.3% 71.0%
62.518 df=3 
p=.000
Yes 40.3% 33.3% 32.3% 14.8% 29.0%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Table 1 Exposure to printed education materials (Continued) 
78
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Attractiveness (among those that “ever 
saw”)
Not attractive 14.9% 18.0% 26.8% 32.2% 21.5%
9.498 df=3 
p=.023
Attractive 85.1% 82.0% 73.2% 67.8% 78.5%
Total 100% 100% 100% 100% 100%
121 100 97 59 377
Ever seen “Road of life”
No 72.3% 80.7% 80.3% 93.0% 82.5%
53.603 df=3 
p=.000
Yes 27.7% 19.3% 19.7% 7.0% 17.5%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 8.4% 22.4% 20.3% 21.4% 16.7%
6.460 df=3 
p=.091
Attractive 91.6% 77.6% 79.7% 78.6% 83.3%
Total 100% 100% 100% 100% 100%
83 58 59 28 228
Ever seen “Good habit” 
No 71.7% 83.7% 78.3% 92.5% 82.4%
55.682 df=3 
p=.000
Yes 28.3% 16.3% 21.7% 7.5% 17.6%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 7.1% 18.4% 7.7% 13.3% 10.5%
5.108 df=3 
p=.164
Attractive 92.9% 81.6% 92.3% 86.7% 89.5%
Total 100% 100% 100% 100% 100%
85 49 65 30 229
Ever seen “Counselling card” 
No 45.7% 67.3% 57.3% 86.5% 65.9%
140.306 
df=3 p=.000
Yes 54.3% 32.7% 42.7% 13.5% 34.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 4.3% 6.1% 5.5% 11.1% 5.9%
3.466 df=3 
p=.325
Attractive 95.7% 93.9% 94.5% 88.9% 94.1%
Total 100% 100% 100% 100% 100%
163 98 128 54 443 
Ever seen “Alcohol use”
No 69.3% 81.3% 77.3% 92.5% 81.1%
63.741 df=3 
p=000
Yes 30.7% 18.7% 22.7% 7.5% 18.9%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 10.9% 17.9% 13.2% 13.3% 13.4%
1.467 df=3 
p=690
Attractive 89.1% 82.1% 86.8% 86.7% 86.6%
Total 100% 100% 100% 100% 100%
92 56 68 30 246
Table 1 Exposure to printed education materials (Continued) 
79
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Ever seen “Risk screening tools”
No 77.7% 90.3% 88.0% 97.3% 89.0%
68.020 df=3 
p=.000
Yes 22.3% 9.7% 12.0% 2.8% 11.0%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 10.4% 10.3% 2.8% 9.1% 8.4%
1.995 df=3
P=.573
Attractive 89.6% 89.7% 97.2% 90.9% 91.6%
Total 100% 100% 100% 100% 100%
67 29 36 11 143
Ever seen “Service package guideline 
(SMARTgirl)” 
No 78.0% 85.3% 83.3% 95.0% 86.2%
45.130 df=3 
p=.000
Yes 22.0% 14.7% 16.7% 5.0% 13.8%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 3.0% 15.9% 16.0% 10.0% 10.6%
6.871 df=3 
p=.076
Attractive 97.0% 84.1% 84.0% 90.0% 89.4%
Total 100% 100% 100% 100% 100%
66 44 50 20 180
Ever seen “SMARTgirl fan” 
No 38.3% 60.7% 53.3% 87.0% 61.9%
187.075
df=3 p=.000
Yes 61.7% 39.3% 46.7% 13.0% 38.1%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 5.4% 10.2% 8.6% 5.8% 7.5%
2.847 df=3 
p=.416
Attractive 94.6% 89.8% 91.4% 94.2% 92.5%
Total 100% 100% 100% 100% 100%
185 118 140 52 495
Ever seen “1295 sticker” 
No 73.3% 82.7% 87.0% 95.0% 85.3%
66.655 df=3 
p=.000
Yes 26.7% 17.3% 13.0% 5.0% 14.7%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 8.8% 7.7% 7.7% 5.0% 7.9%
.317
df=3 p=957
Attractive 91.3% 92.3% 92.3% 95.0% 92.1%
Total 100% 100% 100% 100% 100%
80 52 39 20 191
Ever seen “Birth spacing network”
No 38.7% 49.7% 42.0% 62.3% 49.2%
46.820 df=3 
p=.000
Yes 61.3% 50.3% 58.0% 37.8% 50.8%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Table 1 Exposure to printed education materials (Continued) 
80
Variable
Geographic Area
Flagship 
CoE
Flagship 
TA (GF)
Non￾Flagship 
(GF)
No 
Program
Total
Chi￾Square 
Test
Attractiveness (among those that “ever 
saw”)
Not attractive 10.9% 6.6% 10.3% 13.9% 10.5%
4.328 df=3 
p=.228
Attractive 89.1% 93.4% 89.7% 86.1% 89.5%
Total 100% 100% 100% 100% 100%
184 151 174 151 660
Ever seen “condom use and contraceptives”
No 44.3% 71.0% 57.7% 81.0% 64.8%
112.931 
df=3 p=.000
Yes 55.7% 29.0% 42.3% 19.0% 35.2%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300
Attractiveness (among those that “ever 
saw”)
Not attractive 7.2% 4.6% 7.9% 3.9% 6.3%
1.880 df=3
P=.598
Attractive 92.8% 95.4% 92.1% 96.1% 93.7%
Total 100% 100% 100% 100% 100%
167 87 127 76 457
Ever seen “Blood drop” 
No 67.7% 82.0% 78.3% 93.0% 81.2%
74.311 df=3 
p=.000
Yes 32.3% 18.0% 21.7% 7.0% 18.8%
Total 100% 100% 100% 100% 100%
300 300 300 400 1300 
Attractiveness (among those that “ever 
saw”)
Not attractive 7.2% 5.6% 3.1% 3.6% 5.3%
1.516 df=3 
p=.000
Attractive 92.8% 94.4% 96.9% 96.4% 94.7%
Total 100% 100% 100% 100% 100%
97 54 65 28 244
19.2. Sexual Activities
81
Table 2 Sexual Activities by geographic area and program exposure
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non-Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Other sources of clients 
No other sources 56.7% 49.0% 47.0% 52.5% 51.4% 6.541
p=.088 51.7% 45.8% 52.5% 51.4% 1.914
P=.384
Phone call 25.0% 36.3% 33.7% 39.0% 33.9% 16.042
P=.001 36.3% 35.1% 30.6% 33.9% 4.232
P=.121
Bar/nightclub/karaoke/
massage
4.7% 5.3% 8.7% 6.7% 6.4% 
4.739
P=.192 6.0% 6.1% 7.0% 6.4% 
0.505
P=.777 
Introduced by friends 7.7% 5.3% 9.7% 3.5% 6.3% 12.482
P=.006 5.5% 9.9% 6.4% 6.3% 3.603
P=.165
Meka (sex broker) 8.0% 2.3% 6.0% 1.6% 4.2% 24.684
P=.000 3.2% 7.6% 4.5% 4.2% 5.546
P=.062
Guess house/ hotel .3% 11.3% 5.0% 1.0% 4.2% 60.372
P=.000 3.7% 1.5% 5.4% 4.2% 4.745
P=.093
Restaurant 5.0% 3.0% 2.7% .5% 2.6% 13.903
P=.003 2.1% 1.5% 3.5% 2.6% 2.723
P=.256
Using intermediary (taxi, 
barman) 1.0% 1.7% 3.3% .5% 1.5% 
9.835
P=.020 1.2% 2.3% 1.7% 1.5% 
1.056
P=.590 
Facebook 1.0% 3.7% 1.0% .8% 1.5% 11.760
P=.008 1.1% 0.0% 2.5% 1.5% 6.263
P=.044
Other 1.3% 1.0% 1.3% 1.0% 1.2% 0.315
P=.957 1.2% 2.3% .8% 1.2% 2.161
P=.339
On the street/public 
park .0% 1.7% 2.0% .3% .9% 
10.395
P=.015 .6% .8% 1.4% .9% 
1.785
P=1.78
5
Beer garden 2.7% 0.0% .3% .3% .8% 18.635
P=.000 .5% 0.0% 1.4% .8% 4.169
P=.124
At party/ wedding party 0.0% 3.3% 0.0% 0.0% .8% 
33.592
P=.000 .8% .8% .8% .8% 
0.000
P=1.00
0
Messenger 0.0% 2.7% 0.0% 0.3% .7% 22.331
P=.000 .6% .8% .8% .7% 0.122
P=.941
Street-based 
massage/coin massage 1.0% 1.7% 1.7% 0.0% .7% 
10.481
P=.015 .3% 1.5% .8% .6% 
3.015
P=.222 
Line/Tango/Viber/
WhatsApp .3% 2.0% 0.0% 0.0% .5% 
15.991
P=.001 .2% .8% 1.0% .5% 
3.720
P=.156 
Brothel 0.0% .3% .3% 0.0% .2% 2.337
P=.505 0.0% 0.0% .4% .2% 3.043
P= .218
Bigo live 0.0% .3% 0.0% 0.0% .08% 3.336
P=.343 0.0% 0.0% .2% .08% 1.521
P=.468
Sport club/gym 0.0% 0.0% 0.0% .3% .08% 2.252
P=.522 .2% 0.0% 0.0% .08% 0.992
P=.609
Total 114.7
%
131.0
%
121.0
%
107.8
%
117.8
%
115.0
%
116.8
% 
121.5
%
117.8
%
344 393 363 431 1531 751 153 627 1531
82
19.3. Condom Use
Table 3 Condom use by geographic area and program exposure 
Variable
Geographic Area Program Exposure Flagship CoE Flagship TA (GF) Non- Flagship (GF) No Program Total Chi-Square Test No exposure Some exposure
High 
exposure
Total
Chi-Square 
Test
Usual source of 
condoms
SMARTgirl outreach 
worker 32.0% 43.3% 47.3% 3.5% 29.4% 204.855
, p=.000 3.2% 18.3% 65.3% 29.4% 544.182
, p=.000
SMARTgirl club 17.3% 3.0% 6.3% 0.8% 6.4% 87.167, 
p=.000 0.0% 2.3% 15.5% 6.4% 120.003
, p=.000
Peer seller 0.0% 0.3% 0.0% 0.0% 0.1% 3.336, 
p=.343 0.0% 0.0% 0.2% 0.1% 1.521, 
p=.468 
Street-based sales 9.3% 9.3% 0.3% 3.8% 5.5% 34.497, 
p=.000 5.5% 6.1% 5.4% 5.5% 0.094, 
p=.954 
FP Officer 0.3% 0.7% 0.7% 0.0% 0.4% 2.811, 
p=.422 0.0% 0.0% 1.0% 0.4% 7.626, 
p=.022 
NGO/outreach 
worker/DIC (not 
SMARTgirl)
11.3% 6.3% 10.0% 5.5% 8.1% 10.585, 
p=.014 8.1% 14.5% 6.4% 8.1% 9.255, 
p=.010 
Store/gas 
station/vendor/ 
pharmacy
 76.3% 63.7% 51.3% 47.3% 58.7% 69.872, 
p=.000 52.5% 64.1% 65.1% 58.7% 20.615, 
p=.000 
Guesthouse/brothel/m
assage parlor/karaoke/ 
spas/saunas/beer 
garden
51.0% 51.0% 50.7% 63.3% 54.7% 17.085, 
p=.001 58.4% 55.0% 50.0% 54.7% 8.107, 
p=.017 
Client 24.0% 33.0% 35.7% 23.0% 28.5% 19.477, 
p=.000 32.6% 22.9% 24.6% 28.5% 11.287, 
p=.004
Sexual 
partner/sweetheart 8.3% 20.0% 17.3% 11.0% 13.9% 22.829, 
p=.000 15.8% 13.7% 11.6% 13.9% 4.137, 
p=.126
Friend 4.3% 4.3% 6.0% 2.0% 4.0% 7.465, 
p=.058 3.8% 6.9% 3.5% 4.0% 3.212, 
p=.201
Never bought/received 0.7% 0.3% 0.7% 1.3% 0.8% 2.041, 
p=.564 1.4% 0.0% 0.2% 0.8% 6.427, 
p=.040
Family health 
clinic/health center 11.0% 12.3% 10.0% 8.8% 10.4% 2.542, 
p=.468 8.3% 15.3% 11.8% 10.4% 7.640, 
p=.022 
Other 1.7% 1.0% 2.7% 0.5% 1.4% 6.403, 
p=.094 1.7% 3.1% 0.6% 1.4% 5.540, 
p=.063 
Total 247.7
%
248.7
%
239.0
%
170.5
%
222.2
%
993.440
5, 
p=.000 
191.3
%
222.1
%
261.2
%
222.2
%
961.893
6, 
 743 746 717 682 2888 1249 291 1348 2888 p=.000 
19.4. Program Impact
STI Screening 
83
Table 4 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting STI screening and treatment in the past 6 
months 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Average probability of outcome by geographic area
Flagship CoE 0.469 0.0350 0.000 0.4002 0.5375
Flagship TA (GF) 0.435 0.0271 0.000 0.3823 0.4887
Non-Flagship (GF) 0.473 0.0292 0.000 0.4158 0.5304
No Program 0.370 0.0281 0.000 0.3148 0.4247
Comparison of probability of outcome between the 
geographic areas
No program – reference
Flagship CoE 0.099 0.0448 0.027 0.0114 0.1869
Flagship TA (GF) 0.066 0.0390 0.092 -0.0107 0.1421
Non-Flagship (GF) 0.103 0.0404 0.011 0.0241 0.1826
Flagship CoE Model 
Age
18-24 years – reference 
25-30 years 0.199 0.3735 0.594 -0.5328 0.9312
31 years and above 0.160 0.4597 0.728 -0.7411 1.0608
Education
Under 1 year – reference 
1-6 years 0.839 0.6837 0.220 -0.5012 2.1790
7 years and above 1.271 0.7366 0.085 -0.1730 2.7145
Income 
Under $150 – reference 
$150 - $250 -0.088 0.6299 0.889 -1.3227 1.1464
$250 - $300 -0.434 0.7031 0.537 -1.8120 0.9440
$300 - $500 -0.558 0.6134 0.363 -1.7599 0.6447
$500 and above -0.719 0.6575 0.274 -2.0082 0.5693
Marital status 
Single - reference
Having partner/boyfriend 0.495 0.5765 0.391 -0.6352 1.6248
Married (having husband) 1.189 0.6202 0.055 -0.0266 2.4046
Divorced/separated/widowed 0.960 0.5757 0.095 -0.1684 2.0885
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.604 0.4147 0.145 -1.4173 0.2083
Duration working in current workplace
Less than 12 months – reference
12 months and above 1.251 0.4912 0.011 0.2883 2.2138
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.880 0.4923 0.074 -1.8450 0.0848
Disclosed status as sex worker (0=no, 1=yes) -0.720 0.2978 0.016 -1.3040 -0.1366
Ever suspected having any STI (0=no, 1=yes) 0.647 0.3028 0.033 0.0530 1.2400
Risk index 
Low risk – reference 
Medium risk -0.029 0.3499 0.933 -0.7152 0.6563
High risk -0.074 0.3674 0.840 -0.7945 0.6457
Stigma and discrimination index 
Low – reference 
Medium -0.304 0.3748 0.417 -1.0385 0.4307
High -0.437 0.3633 0.229 -1.1489 0.2754
Constant -0.889 1.1920 0.456 -3.2256 1.4468
Table 4 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting STI screening and treatment in the past 6 
months (Continued) 
84
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Flagship TA (GF) Model 
Age
18-24 years – reference 
25-30 years 0.950 0.3593 0.008 0.2458 1.6542
31 years and above 1.327 0.4435 0.003 0.4573 2.1959
Education
Under 1 year – reference 
1-6 years -0.758 0.4462 0.090 -1.6321 0.1169
7 years and above -0.256 0.4692 0.586 -1.1752 0.6641
Income 
Under $150 – reference 
$150 - $250 0.330 0.4982 0.508 -0.6463 1.3065
$250 - $300 0.671 0.5487 0.222 -0.4048 1.7460
$300 - $500 0.146 0.4788 0.760 -0.7924 1.0846
$500 and above 0.015 0.5242 0.977 -1.0119 1.0427
Marital status 
Single - reference
Having partner/boyfriend 0.945 0.5735 0.099 -0.1793 2.0688
Married (having husband) -0.318 0.6720 0.636 -1.6346 0.9994
Divorced/separated/widowed -0.547 0.5754 0.342 -1.6749 0.5807
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.489 0.3392 0.149 -0.1757 1.1540
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.012 0.3952 0.975 -0.7869 0.7624
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.671 0.3748 0.073 -0.0632 1.4061
Disclosed status as sex worker (0=no, 1=yes) 0.496 0.2799 0.076 -0.0526 1.0447
Ever suspected having any STI (0=no, 1=yes) 0.875 0.2768 0.002 0.3319 1.4172
Risk index
Low risk – reference 
Medium risk 0.193 0.3010 0.522 -0.3973 0.7825
High risk -0.286 0.3997 0.474 -1.0694 0.4974
Stigma and discrimination index 
Low – reference 
Medium 0.172 0.3145 0.584 -0.4441 0.7887
High -0.148 0.3392 0.662 -0.8131 0.5166
Constant -1.941 0.8778 0.027 -3.6611 -0.2200
Non-Flagship (GF) – Model 
Age
18-24 years – reference 
25-30 years -0.245 0.3316 0.460 -0.8952 0.4048
31 years and above -0.360 0.3898 0.356 -1.1241 0.4038
Education
Under 1 year – reference 
1-6 years -0.007 0.3857 0.986 -0.7626 0.7494
7 years and above 0.041 0.4352 0.924 -0.8117 0.8943
Income 
Under $150 – reference 
$150 - $250 0.494 0.4428 0.265 -0.3739 1.3617
$250 - $300 0.673 0.4987 0.177 -0.3048 1.6503
$300 - $500 0.679 0.4591 0.139 -0.2204 1.5791
$500 and above 0.766 0.5801 0.186 -0.3705 1.9035
Table 4 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting STI screening and treatment in the past 6 
months (Continued) 
85
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Marital status 
Single - reference
Having partner/boyfriend 0.371 0.5680 0.514 -0.7423 1.4841
Married (having husband) 0.150 0.6600 0.820 -1.1432 1.4440
Divorced/separated/widowed 0.557 0.5581 0.319 -0.5373 1.6506
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.253 0.3865 0.513 -0.5047 1.0103
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.196 0.4537 0.665 -0.6929 1.0856
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.212 0.3972 0.593 -0.5662 0.9908
Disclosed status as sex worker (0=no, 1=yes) -0.803 0.2673 0.003 -1.3265 -0.2788
Ever suspected having any STI (0=no, 1=yes) 0.751 0.2798 0.007 0.2029 1.2997
Risk index
Low risk – reference 
Medium risk 1.141 0.3140 0.000 0.5259 1.7566
High risk 0.373 0.3176 0.240 -0.2491 0.9959
Stigma and discrimination index 
Low – reference 
Medium -0.229 0.3171 0.469 -0.8510 0.3921
High -0.182 0.3340 0.585 -0.8369 0.4725
Constant -1.506 0.7803 0.054 -3.0350 0.0238
No Program Model 
Age
18-24 years – reference 
25-30 years -0.282 0.2990 0.346 -0.8679 0.3041
31 years and above -0.092 0.3869 0.811 -0.8508 0.6660
Education
Under 1 year – reference 
1-6 years -0.075 0.4107 0.856 -0.8796 0.7305
7 years and above 0.004 0.4580 0.993 -0.8939 0.9014
Income 
Under $150 – reference 
$150 - $250 -0.044 0.3921 0.910 -0.8127 0.7243
$250 - $300 0.638 0.4625 0.167 -0.2680 1.5450
$300 - $500 -0.066 0.4278 0.877 -0.9045 0.7724
$500 and above 0.159 0.4946 0.747 -0.8099 1.1289
Marital status 
Single - reference
Having partner/boyfriend -0.970 0.5279 0.066 -2.0043 0.0649
Married (having husband) 0.522 0.6808 0.443 -0.8124 1.8562
Divorced/separated/widowed -0.288 0.5192 0.579 -1.3060 0.7292
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.732 0.6486 0.259 -2.0038 0.5389
Duration working in current workplace
Less than 12 months – reference
12 months and above 1.348 0.6957 0.053 -0.0152 2.7120
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.195 0.3239 0.548 -0.8293 0.4402
Disclosed status as sex worker (0=no, 1=yes) 0.092 0.2528 0.717 -0.4038 0.5873
Ever suspected having any STI (0=no, 1=yes) 0.194 0.2811 0.491 -0.3574 0.7447
Table 4 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting STI screening and treatment in the past 6 
months (Continued) 
86
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Risk index
Low risk – reference 
Medium risk 0.284 0.2958 0.338 -0.2962 0.8634
High risk 0.049 0.3254 0.880 -0.5888 0.6867
Stigma and discrimination index 
Low – reference 
Medium -0.476 0.3384 0.159 -1.1394 0.1870
High -0.213 0.2994 0.476 -0.8001 0.3736
Constant -0.238 0.7654 0.756 -1.7378 1.2624
87
Table 5 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting STI screening and treatment in the past 6 months 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Average probability of outcome by program exposure
No exposure 0.332 0.0201 0.000 0.2929 0.3719
Some exposure 0.265 0.0370 0.000 0.1930 0.3378
High exposure 0.615 0.0245 0.000 0.5671 0.6630
Comparison of probability of outcome between the 
levels of program exposure
No exposure – reference
Some exposure -0.067 0.0422 0.112 -0.1496 0.0157
High exposure 0.283 0.0316 0.000 0.2207 0.3446
No Exposure Model 
Age
18-24 years – reference 
25-30 years 0.192 0.2409 0.426 -0.2804 0.6637
31 years and above 0.427 0.2738 0.119 -0.1101 0.9633
Education
Under 1 year – reference 
1-6 years 0.106 0.3302 0.747 -0.5409 0.7536
7 years and above 0.305 0.3547 0.390 -0.3902 1.0003
Income 
Under $150 – reference 
$150 - $250 0.319 0.3126 0.308 -0.2941 0.9314
$250 - $300 0.758 0.3564 0.033 0.0597 1.4567
$300 - $500 0.014 0.3490 0.969 -0.6704 0.6977
$500 and above 0.543 0.3711 0.143 -0.1840 1.2706
Marital status 
Single - reference
Having partner/boyfriend 0.175 0.3845 0.649 -0.5787 0.9287
Married (having husband) -0.285 0.5247 0.587 -1.3136 0.7431
Divorced/separated/widowed -0.093 0.3933 0.813 -0.8637 0.6778
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.180 0.2696 0.503 -0.3479 0.7089
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.378 0.3265 0.247 -0.2618 1.0182
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.329 0.2412 0.172 -0.8019 0.1437
Disclosed status as sex worker (0=no, 1=yes) -0.407 0.1960 0.038 -0.7908 -0.0223
Ever suspected having any STI (0=no, 1=yes) 0.388 0.2022 0.055 -0.0085 0.7842
Risk index
Low risk – reference 
Medium risk 0.314 0.2160 0.146 -0.1094 0.7375
High risk 0.112 0.2416 0.642 -0.3614 0.5857
Stigma and discrimination index 
Low – reference 
Medium -0.454 0.2447 0.064 -0.9332 0.0262
High -0.139 0.2286 0.544 -0.5868 0.3094
Constant -1.304 0.6121 0.033 -2.5039 -0.1046
Some Exposure Model 
Age
18-24 years – reference 
25-30 years 0.212 0.4888 0.665 -0.7465 1.1696
31 years and above 0.246 0.5660 0.664 -0.8639 1.3549
88
Table 5 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting STI screening and treatment in the past 6 months 
(Continued) 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Education
Under 1 year – reference 
1-6 years -0.654 0.6989 0.349 -2.0239 0.7157
7 years and above -0.671 0.7827 0.391 -2.2051 0.8631
Income 
Under $150 – reference 
$150 - $250 0.265 0.6975 0.703 -1.1016 1.6325
$250 - $300 -0.638 0.8042 0.428 -2.2139 0.9387
$300 - $500 0.741 0.6846 0.279 -0.6011 2.0824
$500 and above 0.269 0.8456 0.750 -1.3884 1.9265
Marital status 
Single - reference
Having partner/boyfriend 0.083 0.8263 0.920 -1.5369 1.7023
Married(having husband) 1.380 0.9996 0.167 -0.5788 3.3394
Divorced/separated/widowed -0.048 0.8350 0.954 -1.6849 1.5883
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.937 0.7926 0.237 -0.6165 2.4902
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.905 0.8518 0.288 -2.5742 0.7649
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.265 0.5903 0.653 -0.8917 1.4222
Disclosed status as sex worker (0=no, 1=yes) 0.247 0.3943 0.531 -0.5259 1.0196
Ever suspected having any STI (0=no, 1=yes) -0.125 0.4545 0.784 -1.0156 0.7661
Risk index
Low risk – reference 
Medium risk -0.418 0.4752 0.379 -1.3492 0.5134
High risk -0.475 0.5491 0.387 -1.5516 0.6007
Stigma and discrimination index 
Low – reference 
Medium -0.465 0.5512 0.399 -1.5458 0.6150
High -0.021 0.5078 0.967 -1.0163 0.9743
Constant -1.015 1.3010 0.435 -3.5644 1.5353
High Exposure – Model 
Age
18-24 years – reference 
25-30 years 0.166 0.2754 0.546 -0.3733 0.7063
31 years and above 0.370 0.3054 0.226 -0.2286 0.9686
Education
Under 1 year – reference 
1-6 years 0.352 0.3659 0.336 -0.3648 1.0694
7 years and above 0.426 0.3793 0.262 -0.3176 1.1691
Income 
Under $150 – reference 
$150 - $250 -0.420 0.4990 0.399 -1.3985 0.5576
$250 - $300 -0.505 0.5533 0.362 -1.5892 0.5797
$300 - $500 -0.535 0.4920 0.276 -1.4997 0.4289
$500 and above -0.627 0.5098 0.219 -1.6264 0.3721
Marital status 
Single - reference
Having partner/boyfriend 0.132 0.4241 0.757 -0.6998 0.9628
Married(having husband) 0.282 0.4619 0.541 -0.6231 1.1876
Divorced/separated/widowed 0.079 0.3961 0.843 -0.6979 0.8549
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.159 0.3365 0.636 -0.8185 0.5004
89
Table 5 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting STI screening and treatment in the past 6 months 
(Continued) 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.344 0.3403 0.312 -0.3233 1.0108
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.105 0.2997 0.726 -0.6921 0.4825
Disclosed status as sex worker (0=no, 1=yes) -0.219 0.2164 0.310 -0.6435 0.2046
Ever suspected having any STI (0=no, 1=yes) 0.552 0.2332 0.018 0.0951 1.0092
Risk index
Low risk – reference 
Medium risk 0.315 0.2627 0.231 -0.2000 0.8298
High risk 0.016 0.2648 0.951 -0.5029 0.5351
Stigma and discrimination index 
Low – reference 
Medium 0.148 0.2294 0.519 -0.3017 0.5975
High 0.017 0.2663 0.949 -0.5050 0.5391
Constant 0.084 0.7167 0.907 -1.3211 1.4884
90
HIV Testing (Any Type)
Table 6 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV testing (any type) in the past 6 months 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Average probability of outcome by geographic area
Flagship CoE 0.585 0.0369 0.000 0.5130 0.6575
Flagship TA (GF) 0.610 0.0279 0.000 0.5550 0.6645
Non-Flagship (GF) 0.672 0.0264 0.000 0.6199 0.7233
No Program 0.481 0.0306 0.000 0.4211 0.5411
Comparison of probability of outcome between the 
geographic areas
No program – reference
Flagship CoE 0.104 0.0479 0.030 0.0103 0.1979
Flagship TA (GF) 0.129 0.0413 0.002 0.0476 0.2096
Non-Flagship (GF) 0.190 0.0403 0.000 0.1115 0.2694
Flagship CoE Model 
Age
18-24 years – reference 
25-30 years 0.092 0.3743 0.805 -0.6412 0.8260
31 years and above 0.436 0.4462 0.328 -0.4381 1.3108
Education
Under 1 year – reference 
1-6 years 0.582 0.7098 0.412 -0.8091 1.9733
7 years and above 0.866 0.7302 0.235 -0.5649 2.2976
Income 
Under $150 – reference 
$150 - $250 0.350 0.6331 0.580 -0.8909 1.5908
$250 - $300 0.484 0.6996 0.489 -0.8874 1.8550
$300 - $500 0.386 0.6377 0.545 -0.8643 1.6355
$500 and above -0.036 0.6492 0.955 -1.3087 1.2362
Marital status 
Single - reference
Having partner/boyfriend 0.848 0.5316 0.111 -0.1936 1.8901
Married (having husband) 1.429 0.5860 0.015 0.2805 2.5776
Divorced/separated/widowed 0.676 0.5104 0.185 -0.3241 1.6769
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.485 0.3780 0.200 -1.2256 0.2562
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.837 0.4163 0.044 0.0211 1.6531
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.669 0.4363 0.125 -1.5241 0.1862
Disclosed status as sex worker (0=no, 1=yes) -0.405 0.3115 0.194 -1.0154 0.2056
Ever suspected contracting HIV (0=no, 1=yes) 0.211 0.3468 0.544 -0.4690 0.8903
Risk index
Low risk – reference 
Medium risk 0.231 0.3454 0.503 -0.4457 0.9082
High risk 0.217 0.3837 0.571 -0.5348 0.9691
Stigma and discrimination index 
Low – reference 
Medium -0.216 0.3413 0.527 -0.8845 0.4532
High -0.652 0.3925 0.097 -1.4210 0.1177
Constant -0.844 1.1794 0.474 -3.1554 1.4679
91
Table 6 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV testing (any type) in the past 6 months
(Continued) 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Flagship TA (GF) Model 
Age
18-24 years – reference 
25-30 years 1.028 0.3473 0.003 0.3477 1.7091
31 years and above 1.116 0.4237 0.008 0.2853 1.9460
Education
Under 1 year – reference 
1-6 years -0.375 0.4441 0.399 -1.2453 0.4954
7 years and above -0.126 0.4676 0.788 -1.0423 0.7906
Income 
Under $150 – reference 
$150 - $250 1.370 0.4728 0.004 0.4435 2.2971
$250 - $300 0.710 0.5287 0.179 -0.3258 1.7467
$300 - $500 0.497 0.4414 0.260 -0.3683 1.3621
$500 and above 0.689 0.5133 0.179 -0.3168 1.6954
Marital status 
Single - reference
Having partner/boyfriend -0.031 0.5736 0.957 -1.1554 1.0931
Married (having husband) -0.974 0.6713 0.147 -2.2897 0.3418
Divorced/separated/widowed -1.123 0.5445 0.039 -2.1902 -0.0559
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.488 0.3515 0.165 -0.2007 1.1772
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.288 0.4172 0.489 -0.5293 1.1062
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.341 0.3423 0.319 -0.3300 1.0118
Disclosed status as sex worker (0=no, 1=yes) 0.067 0.2843 0.814 -0.4904 0.6241
Ever suspected contracting HIV (0=no, 1=yes) 0.011 0.2765 0.970 -0.5313 0.5524
Risk index
Low risk – reference 
Medium risk 0.281 0.3102 0.365 -0.3267 0.8891
High risk -0.379 0.3797 0.319 -1.1227 0.3656
Stigma and discrimination index 
Low – reference 
Medium -0.011 0.3282 0.974 -0.6539 0.6326
High -0.104 0.3422 0.760 -0.7751 0.5663
Constant -0.595 0.9037 0.510 -2.3661 1.1762
Non-Flagship (GF) – Model 
Age
18-24 years – reference 
25-30 years 0.080 0.3541 0.822 -0.6142 0.7737
31 years and above -1.098 0.4139 0.008 -1.9089 -0.2865
Education
Under 1 year – reference 
1-6 years -1.417 0.4976 0.004 -2.3921 -0.4415
7 years and above -0.960 0.5605 0.087 -2.0585 0.1388
Income 
Under $150 – reference 
$150 - $250 0.456 0.4436 0.304 -0.4134 1.3255
$250 - $300 0.944 0.5575 0.090 -0.1484 2.0372
$300 - $500 0.894 0.4861 0.066 -0.0586 1.8468
$500 and above 1.363 0.6435 0.034 0.1016 2.6242
92
Table 6 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV testing (any type) in the past 6 months
(Continued) 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Marital status 
Single - reference
Having partner/boyfriend -0.487 0.5319 0.360 -1.5294 0.5555
Married (having husband) -0.520 0.6489 0.423 -1.7917 0.7518
Divorced/separated/widowed 0.098 0.5372 0.855 -0.9545 1.1514
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.091 0.4530 0.840 -0.7966 0.9792
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.758 0.5032 0.132 -0.2285 1.7441
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.219 0.4157 0.598 -1.0338 0.5959
Disclosed status as sex worker (0=no, 1=yes) -0.119 0.2856 0.678 -0.6783 0.4412
Ever suspected contracting HIV (0=no, 1=yes) -0.328 0.2905 0.259 -0.8974 0.2414
Risk index
Low risk – reference 
Medium risk 1.016 0.3284 0.002 0.3719 1.6594
High risk 0.460 0.3414 0.178 -0.2092 1.1289
Stigma and discrimination index 
Low – reference 
Medium -0.216 0.3435 0.529 -0.8894 0.4571
High 0.031 0.3362 0.928 -0.6283 0.6894
Constant 1.108 0.8582 0.197 -0.5745 2.7897
No Program Model 
Age
18-24 years – reference 
25-30 years -0.295 0.2925 0.314 -0.8679 0.2787
31 years and above -0.030 0.3488 0.932 -0.7134 0.6538
Education
Under 1 year – reference 
1-6 years 0.325 0.3862 0.400 -0.4321 1.0817
7 years and above 0.522 0.4273 0.222 -0.3153 1.3598
Income 
Under $150 – reference 
$150 - $250 0.050 0.3510 0.886 -0.6375 0.7383
$250 - $300 -0.100 0.4476 0.823 -0.9772 0.7774
$300 - $500 0.141 0.3998 0.724 -0.6425 0.9247
$500 and above -0.044 0.4901 0.928 -1.0047 0.9165
Marital status 
Single - reference
Having partner/boyfriend -0.702 0.5758 0.223 -1.8302 0.4269
Married (having husband) -0.156 0.7031 0.825 -1.5337 1.2224
Divorced/separated/widowed -0.846 0.5507 0.124 -1.9256 0.2331
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.236 0.6243 0.705 -0.9872 1.4600
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.071 0.6868 0.918 -1.4167 1.2754
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.054 0.2956 0.854 -0.6338 0.5251
Disclosed status as sex worker (0=no, 1=yes) -0.128 0.2512 0.610 -0.6203 0.3642
Ever suspected contracting HIV (0=no, 1=yes) -0.521 0.2801 0.063 -1.0704 0.0277
93
Table 6 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV testing (any type) in the past 6 months
(Continued) 
Variable Probability Robust Std. 
Err. Sig. Level 95% Conf. Interval
Risk index
Low risk – reference 
Medium risk 0.476 0.2796 0.089 -0.0719 1.0242
High risk 0.147 0.3347 0.659 -0.5085 0.8035
Stigma and discrimination index 
Low – reference 
Medium -0.137 0.3503 0.697 -0.8232 0.5501
High -0.391 0.2916 0.181 -0.9621 0.1811
Constant 0.470 0.7644 0.538 -1.0277 1.9687
94
Table 7 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV testing (any type) in the past 6 months 
Variable Probability Robust Std. 
Err.
Sig. 
Level 95% Conf. Interval
Average probability of outcome by program exposure
No exposure 0.440 0.0209 0.000 0.3994 0.4812
Some exposure 0.377 0.0390 0.000 0.3011 0.4539
High exposure 0.794 0.0198 0.000 0.7551 0.8328
Comparison of probability of outcome between the levels 
of program exposure
No exposure – reference
Some exposure -0.063 0.0445 0.158 -0.1500 0.0244
High exposure 0.354 0.0288 0.000 0.2973 0.4100
No Exposure Model 
Age
18-24 years – reference 
25-30 years -0.033 0.2179 0.879 -0.4604 0.3938
31 years and above 0.143 0.2558 0.575 -0.3579 0.6447
Education
Under 1 year – reference 
1-6 years -0.025 0.3151 0.938 -0.6422 0.5930
7 years and above 0.135 0.3368 0.688 -0.5250 0.7952
Income 
Under $150 – reference 
$150 - $250 0.351 0.2773 0.206 -0.1926 0.8944
$250 - $300 0.234 0.3392 0.490 -0.4307 0.8990
$300 - $500 0.212 0.3043 0.486 -0.3846 0.8082
$500 and above 0.737 0.3474 0.034 0.0563 1.4180
Marital status 
Single - reference
Having partner/boyfriend -0.105 0.3594 0.771 -0.8091 0.5995
Married(having husband) -0.570 0.4753 0.231 -1.5011 0.3621
Divorced/separated/widowed -0.546 0.3543 0.124 -1.2401 0.1488
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.088 0.2590 0.734 -0.4196 0.5956
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.336 0.3103 0.278 -0.2716 0.9445
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.387 0.2248 0.085 -0.8278 0.0533
Disclosed status as sex worker (0=no, 1=yes) -0.349 0.1773 0.049 -0.6960 -0.0010
Ever suspected contracting HIV (0=no, 1=yes) -0.372 0.1891 0.049 -0.7425 -0.0011
Risk index
Low risk – reference 
Medium risk 0.549 0.2018 0.007 0.1530 0.9442
High risk -0.044 0.2316 0.851 -0.4976 0.4103
Stigma and discrimination index 
Low – reference 
Medium 0.166 0.2315 0.474 -0.2880 0.6197
High 0.146 0.2152 0.496 -0.2754 0.5682
Constant -0.183 0.5458 0.738 -1.2525 0.8869
Some Exposure Model 
Age
18-24 years – reference 
25-30 years 0.484 0.5551 0.383 -0.6038 1.5721
31 years and above -0.085 0.6453 0.895 -1.3497 1.1797
95
Table 7 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV testing (any type) in the past 6 months 
(Continued) 
Variable Probability Robust Std. 
Err.
Sig. 
Level 95% Conf. Interval
Education
Under 1 year – reference 
1-6 years 0.195 0.7035 0.781 -1.1834 1.5744
7 years and above -0.096 0.7225 0.894 -1.5120 1.3202
Income 
Under $150 – reference 
$150 - $250 0.574 0.6761 0.396 -0.7507 1.8995
$250 - $300 -0.411 0.7816 0.599 -1.9426 1.1212
$300 - $500 1.573 0.7163 0.028 0.1687 2.9767
$500 and above 1.063 0.7378 0.150 -0.3835 2.5087
Marital status 
Single - reference
Having partner/boyfriend 0.952 0.9909 0.337 -0.9904 2.8939
Married(having husband) 2.472 1.1485 0.031 0.2213 4.7232
Divorced/separated/widowed 1.598 1.0067 0.112 -0.3754 3.5708
Duration living in current location 
Less than 12 months – reference 
12 months and above 2.142 0.8040 0.008 0.5660 3.7178
Duration working in current workplace
Less than 12 months – reference
12 months and above -2.061 0.9020 0.022 -3.8290 -0.2932
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.099 0.5819 0.866 -1.0421 1.2391
Disclosed status as sex worker (0=no, 1=yes) 0.540 0.3875 0.163 -0.2191 1.3001
Ever suspected contracting HIV (0=no, 1=yes) 0.311 0.4304 0.470 -0.5323 1.1548
Risk index
Low risk – reference 
Medium risk -1.337 0.5818 0.022 -2.4779 -0.1971
High risk -0.413 0.5887 0.483 -1.5669 0.7408
Stigma and discrimination index 
Low – reference 
Medium -0.077 0.5326 0.885 -1.1210 0.9669
High 1.138 0.5255 0.030 0.1084 2.1684
Constant -3.529 1.4627 0.016 -6.3959 -0.6620
High Exposure – Model 
Age
18-24 years – reference 
25-30 years 0.286 0.3365 0.394 -0.3729 0.9459
31 years and above 0.214 0.3399 0.530 -0.4526 0.8800
Education
Under 1 year – reference 
1-6 years -0.513 0.4723 0.277 -1.4388 0.4124
7 years and above -0.391 0.4940 0.428 -1.3598 0.5768
Income 
Under $150 – reference 
$150 - $250 -0.102 0.6052 0.867 -1.2879 1.0845
$250 - $300 0.009 0.6772 0.990 -1.3185 1.3360
$300 - $500 -0.450 0.5930 0.448 -1.6120 0.7125
$500 and above -0.685 0.6074 0.259 -1.8756 0.5052
Marital status 
Single - reference
Having partner/boyfriend 0.421 0.5539 0.447 -0.6648 1.5063
Married(having husband) 0.343 0.5975 0.566 -0.8284 1.5137
Divorced/separated/widowed -0.103 0.4986 0.837 -1.0800 0.8746
96
Table 7 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV testing (any type) in the past 6 months 
(Continued) 
Variable Probability Robust Std. 
Err.
Sig. 
Level 95% Conf. Interval
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.541 0.4089 0.186 -0.2608 1.3420
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.165 0.4255 0.699 -0.9984 0.6693
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.394 0.3987 0.323 -1.1758 0.3873
Disclosed status as sex worker (0=no, 1=yes) 0.202 0.2505 0.419 -0.2886 0.6934
Ever suspected contracting HIV (0=no, 1=yes) -0.413 0.2475 0.095 -0.8977 0.0725
Risk index
Low risk – reference 
Medium risk 0.255 0.2853 0.372 -0.3044 0.8138
High risk 0.050 0.3176 0.875 -0.5726 0.6726
Stigma and discrimination index 
Low – reference 
Medium -0.230 0.2649 0.385 -0.7491 0.2891
High 0.012 0.3338 0.972 -0.6427 0.6659
Constant 1.875 0.8574 0.029 0.1945 3.5555
97
HIV Finger Prick Testing 
Table 8 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV finger prick testing in the past 6 months 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Average probability of outcome by geographic area
Flagship CoE 0.487 0.0357 0.000 0.4170 0.5571
Flagship TA (GF) 0.523 0.0280 0.000 0.4675 0.5775
Non-Flagship (GF) 0.566 0.0283 0.000 0.5109 0.6220
No Program 0.368 0.0276 0.000 0.3143 0.4226
Comparison of probability of outcome between the 
geographic areas
No program – reference
Flagship CoE 0.119 0.0454 0.009 0.0297 0.2075
Flagship TA (GF) 0.154 0.0392 0.000 0.0772 0.2309
Non-Flagship (GF) 0.198 0.0394 0.000 0.1207 0.2753
Flagship CoE Model 
Age
18-24 years – reference 
25-30 years 0.142 0.3560 0.691 -0.5562 0.8394
31 years and above 0.321 0.4217 0.447 -0.5060 1.1473
Education
Under 1 year – reference 
1-6 years 0.176 0.6984 0.801 -1.1932 1.5445
7 years and above 0.641 0.7185 0.372 -0.7675 2.0489
Income 
Under $150 – reference 
$150 - $250 -0.297 0.6126 0.627 -1.4982 0.9032
$250 - $300 0.262 0.6944 0.706 -1.0990 1.6232
$300 - $500 -0.141 0.6059 0.817 -1.3281 1.0469
$500 and above -0.590 0.6393 0.356 -1.8430 0.6631
Marital status 
Single - reference
Having partner/boyfriend 0.839 0.5089 0.099 -0.1582 1.8366
Married (having husband) 1.406 0.5704 0.014 0.2877 2.5235
Divorced/separated/widowed 0.911 0.4944 0.065 -0.0579 1.8799
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.633 0.3986 0.113 -1.4138 0.1487
Duration working in current workplace
Less than 12 months – reference
12 months and above 1.222 0.4492 0.007 0.3412 2.1021
Duration engaged in sex work
Less than 12 months – reference
12 months and above -1.050 0.4401 0.017 -1.9130 -0.1877
Disclosed status as sex worker (0=no, 1=yes) -0.229 0.3127 0.463 -0.8423 0.3835
Ever suspected contracting HIV (0=no, 1=yes) 0.374 0.3338 0.263 -0.2805 1.0282
Risk index
Low risk – reference 
Medium risk -0.188 0.3424 0.583 -0.8589 0.4832
High risk 0.464 0.3796 0.222 -0.2800 1.2081
Stigma and discrimination index 
Low – reference 
Medium -0.281 0.3311 0.396 -0.9297 0.3680
High -0.772 0.4006 0.054 -1.5574 0.0128
Constant -0.491 1.1178 0.661 -2.6815 1.7002
98
Table 8 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV finger prick testing in the past 6 months 
(Continued) 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Flagship TA (GF) Model 
Age
18-24 years – reference 
25-30 years 1.112 0.3601 0.002 0.4064 1.8179
31 years and above 1.116 0.4298 0.009 0.2733 1.9581
Education
Under 1 year – reference 
1-6 years -0.148 0.4348 0.733 -1.0005 0.7040
7 years and above 0.135 0.4546 0.767 -0.7561 1.0259
Income 
Under $150 – reference 
$150 - $250 1.512 0.4881 0.002 0.5553 2.4688
$250 - $300 0.898 0.5508 0.103 -0.1819 1.9771
$300 - $500 0.007 0.4603 0.988 -0.8955 0.9088
$500 and above 0.381 0.5295 0.472 -0.6569 1.4185
Marital status 
Single - reference
Having partner/boyfriend -0.131 0.5460 0.811 -1.2008 0.9393
Married (having husband) -1.401 0.6841 0.041 -2.7418 -0.0603
Divorced/separated/widowed -1.392 0.5376 0.010 -2.4460 -0.3387
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.295 0.3509 0.400 -0.3926 0.9829
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.833 0.4056 0.040 0.0385 1.6284
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.052 0.3574 0.885 -0.7523 0.6487
Disclosed status as sex worker (0=no, 1=yes) 0.037 0.2858 0.898 -0.5234 0.5968
Ever suspected contracting HIV (0=no, 1=yes) -0.215 0.2774 0.439 -0.7584 0.3289
Risk index
Low risk – reference 
Medium risk 0.226 0.2969 0.446 -0.3556 0.8084
High risk -0.196 0.3918 0.617 -0.9637 0.5722
Stigma and discrimination index 
Low – reference 
Medium 0.000 0.3295 0.999 -0.6455 0.6461
High -0.396 0.3407 0.245 -1.0634 0.2721
Constant -0.669 0.8926 0.454 -2.4181 1.0807
Non-Flagship (GF) – Model 
Age
18-24 years – reference 
25-30 years 0.145 0.3305 0.661 -0.5027 0.7928
31 years and above -0.837 0.3887 0.031 -1.5991 -0.0755
Education
Under 1 year – reference 
1-6 years -0.780 0.4182 0.062 -1.5996 0.0399
7 years and above -0.359 0.4659 0.441 -1.2724 0.5539
Income 
Under $150 – reference 
$150 - $250 0.602 0.4273 0.159 -0.2356 1.4394
$250 - $300 1.063 0.5139 0.039 0.0556 2.0701
$300 - $500 0.971 0.4662 0.037 0.0569 1.8843
$500 and above 0.653 0.5869 0.266 -0.4969 1.8037
99
Table 8 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV finger prick testing in the past 6 months 
(Continued) 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Marital status 
Single - reference
Having partner/boyfriend -0.667 0.5246 0.203 -1.6953 0.3610
Married (having husband) -0.412 0.6134 0.502 -1.6144 0.7901
Divorced/separated/widowed -0.276 0.5081 0.586 -1.2723 0.7194
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.522 0.4450 0.241 -0.3506 1.3938
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.070 0.4852 0.885 -0.8807 1.0212
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.017 0.4023 0.966 -0.8053 0.7715
Disclosed status as sex worker (0=no, 1=yes) 0.129 0.2669 0.630 -0.3946 0.6518
Ever suspected contracting HIV (0=no, 1=yes) -0.224 0.2736 0.413 -0.7601 0.3123
Risk index
Low risk – reference 
Medium risk 1.269 0.3202 0.000 0.6411 1.8962
High risk 0.451 0.3227 0.162 -0.1818 1.0834
Stigma and discrimination index 
Low – reference 
Medium -0.605 0.3187 0.058 -1.2293 0.0200
High -0.001 0.3149 0.998 -0.6180 0.6162
Constant 0.033 0.7581 0.965 -1.4530 1.5187
No Program Model 
Age
18-24 years – reference 
25-30 years -0.948 0.3051 0.002 -1.5463 -0.3503
31 years and above -0.513 0.3812 0.178 -1.2605 0.2339
Education
Under 1 year – reference 
1-6 years 0.106 0.4141 0.798 -0.7057 0.9175
7 years and above 0.307 0.4603 0.504 -0.5949 1.2095
Income 
Under $150 – reference 
$150 - $250 0.151 0.3621 0.676 -0.5586 0.8610
$250 - $300 0.069 0.4717 0.883 -0.8553 0.9938
$300 - $500 -0.119 0.4047 0.768 -0.9125 0.6737
$500 and above -0.057 0.5213 0.913 -1.0789 0.9646
Marital status 
Single - reference
Having partner/boyfriend 0.579 0.5075 0.254 -0.4161 1.5734
Married (having husband) 1.046 0.6377 0.101 -0.2035 2.2962
Divorced/separated/widowed 0.453 0.4970 0.362 -0.5211 1.4271
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.533 0.4599 0.247 -0.3685 1.4344
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.480 0.5372 0.372 -1.5324 0.5732
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.772 0.3015 0.010 0.1810 1.3630
Disclosed status as sex worker (0=no, 1=yes) -0.251 0.2548 0.325 -0.7503 0.2485
Ever suspected contracting HIV (0=no, 1=yes) -0.739 0.3163 0.020 -1.3586 -0.1186
100
Table 8 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of getting HIV finger prick testing in the past 6 months 
(Continued) 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Risk index
Low risk – reference 
Medium risk 0.689 0.2966 0.020 0.1077 1.2702
High risk 0.105 0.3182 0.742 -0.5191 0.7284
Stigma and discrimination index 
Low – reference 
Medium 0.394 0.3520 0.263 -0.2957 1.0842
High -0.002 0.3184 0.995 -0.6258 0.6221
Constant -1.376 0.7433 0.064 -2.8330 0.0808
101
Table 9 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV finger prick testing in the past 6 months 
Variable Probability Robust 
Std. Err.
Sig. 
Level 95% Conf. Interval
Average probability of outcome by program exposure
No exposure 0.325 0.0200 0.000 0.2862 0.3646
Some exposure 0.205 0.0378 0.000 0.1305 0.2787
High exposure 0.723 0.0216 0.000 0.6811 0.7656
Comparison of probability of outcome between the levels of 
program exposure 
No exposure – reference
Some exposure -0.121 0.0430 0.005 -0.2050 -0.0366
High exposure 0.398 0.0294 0.000 0.3403 0.4556
No Exposure Model 
Age
18-24 years – reference 
25-30 years 0.070 0.2318 0.761 -0.3840 0.5248
31 years and above -0.006 0.2772 0.982 -0.5497 0.5369
Education
Under 1 year – reference 
1-6 years -0.058 0.3280 0.860 -0.7008 0.5851
7 years and above 0.032 0.3591 0.930 -0.6721 0.7356
Income 
Under $150 – reference 
$150 - $250 0.175 0.2919 0.549 -0.3972 0.7470
$250 - $300 0.114 0.3562 0.749 -0.5840 0.8123
$300 - $500 0.076 0.3221 0.813 -0.5549 0.7076
$500 and above 0.575 0.3597 0.110 -0.1296 1.2803
Marital status 
Single - reference
Having partner/boyfriend 0.047 0.3780 0.900 -0.6933 0.7883
Married(having husband) -0.183 0.5030 0.716 -1.1688 0.8029
Divorced/separated/widowed -0.382 0.3855 0.322 -1.1372 0.3739
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.157 0.2696 0.560 -0.3713 0.6854
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.235 0.3280 0.473 -0.4074 0.8782
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.055 0.2344 0.815 -0.4046 0.5143
Disclosed status as sex worker (0=no, 1=yes) -0.411 0.1890 0.029 -0.7817 -0.0410
Ever suspected contracting HIV (0=no, 1=yes) -0.377 0.2040 0.064 -0.7773 0.0225
Risk index
Low risk – reference 
Medium risk 0.523 0.2137 0.014 0.1043 0.9421
High risk -0.065 0.2545 0.797 -0.5644 0.4334
Stigma and discrimination index 
Low – reference 
Medium 0.150 0.2518 0.552 -0.3436 0.6433
High 0.265 0.2322 0.253 -0.1899 0.7203
Constant -0.943 0.5877 0.109 -2.0950 0.2087
Some Exposure Model 
Age
18-24 years – reference 
25-30 years -1.111 0.8367 0.184 -2.7504 0.5293
31 years and above 0.062 0.8089 0.939 -1.5237 1.6472
102
Table 9 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV finger prick testing in the past 6 months 
(Continued) 
Variable Probability Robust 
Std. Err.
Sig. 
Level 95% Conf. Interval
Education
Under 1 year – reference 
1-6 years -0.102 0.9041 0.910 -1.8737 1.6704
7 years and above 0.108 1.0316 0.916 -1.9137 2.1302
Income 
Under $150 – reference 
$150 - $250 0.426 0.8402 0.612 -1.2211 2.0724
$250 - $300 0.093 0.9354 0.920 -1.7399 1.9267
$300 - $500 1.123 0.8078 0.164 -0.4603 2.7063
$500 and above -0.763 1.3347 0.568 -3.3786 1.8533
Marital status 
Single - reference
Having partner/boyfriend 0.403 1.0597 0.704 -1.6741 2.4799
Married(having husband) 1.783 1.1465 0.120 -0.4636 4.0305
Divorced/separated/widowed 0.635 0.9445 0.501 -1.2160 2.4863
Duration living in current location 
Less than 12 months – reference 
12 months and above 1.744 1.0147 0.086 -0.2443 3.7332
Duration working in current workplace
Less than 12 months – reference
12 months and above -2.972 1.1763 0.012 -5.2772 -0.6660
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.126 0.7466 0.866 -1.3378 1.5890
Disclosed status as sex worker (0=no, 1=yes) 0.459 0.5399 0.395 -0.5990 1.5174
Ever suspected contracting HIV (0=no, 1=yes) -0.322 0.5638 0.568 -1.4270 0.7829
Risk index
Low risk – reference 
Medium risk -0.869 0.7418 0.242 -2.3226 0.5853
High risk -0.073 0.7331 0.921 -1.5097 1.3642
Stigma and discrimination index 
Low – reference 
Medium 0.053 0.5897 0.929 -1.1029 1.2087
High 0.065 0.6048 0.914 -1.1201 1.2507
Constant -1.888 1.5159 0.213 -4.8589 1.0835
High Exposure – Model 
Age
18-24 years – reference 
25-30 years 0.153 0.2950 0.604 -0.4250 0.7312
31 years and above -0.132 0.3180 0.679 -0.7550 0.4917
Education
Under 1 year – reference 
1-6 years -0.253 0.3895 0.515 -1.0168 0.5101
7 years and above -0.074 0.4046 0.854 -0.8675 0.7187
Income 
Under $150 – reference 
$150 - $250 0.351 0.4871 0.471 -0.6036 1.3058
$250 - $300 0.342 0.5495 0.534 -0.7352 1.4188
$300 - $500 -0.358 0.4712 0.448 -1.2811 0.5660
$500 and above -0.723 0.4948 0.144 -1.6926 0.2471
Marital status 
Single - reference
Having partner/boyfriend 0.214 0.5270 0.684 -0.8188 1.2472
Married(having husband) 0.240 0.5561 0.666 -0.8503 1.3297
Divorced/separated/widowed -0.140 0.4821 0.771 -1.0851 0.8049
103
Table 9 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of getting HIV finger prick testing in the past 6 months 
(Continued) 
Variable Probability Robust 
Std. Err.
Sig. 
Level 95% Conf. Interval
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.213 0.3612 0.555 -0.4945 0.9214
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.343 0.3715 0.356 -0.3855 1.0707
Duration engaged in sex work
Less than 12 months – reference
12 months and above -0.716 0.3491 0.040 -1.3999 -0.0316
Disclosed status as sex worker (0=no, 1=yes) 0.166 0.2257 0.461 -0.2761 0.6087
Ever suspected contracting HIV (0=no, 1=yes) -0.340 0.2266 0.134 -0.7840 0.1043
Risk index
Low risk – reference 
Medium risk 0.140 0.2577 0.586 -0.3648 0.6454
High risk 0.227 0.2902 0.433 -0.3413 0.7963
Stigma and discrimination index 
Low – reference 
Medium -0.303 0.2526 0.231 -0.7978 0.1922
High -0.178 0.2921 0.542 -0.7509 0.3942
Constant 1.469 0.7408 0.047 0.0173 2.9214
104
Stigma and Discrimination 
Table 10 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of being in high stigma and discrimination 
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Average probability of outcome by geographic area
Flagship CoE 0.419 0.0334 0.000 0.353 0.484
Flagship TA (GF) 0.508 0.0288 0.000 0.451 0.564
Non-Flagship (GF) 0.425 0.0287 0.000 0.369 0.481
No Program 0.617 0.0284 0.000 0.561 0.673
Comparison of probability of outcome between the 
geographic areas 
No program – reference
Flagship CoE -0.198 0.0436 0.000 -0.284 -0.113
Flagship TA (GF) -0.109 0.0402 0.007 -0.188 -0.031
Non-Flagship (GF) -0.192 0.0401 0.000 -0.271 -0.113
Flagship CoE Model 
Age
18-24 years – reference 
25-30 years -0.951 0.3840 0.013 -1.704 -0.198
31 years and above -0.422 0.4555 0.354 -1.315 0.471
Education
Under 1 year – reference 
1-6 years -1.052 0.7060 0.136 -2.436 0.331
7 years and above -2.078 0.7360 0.005 -3.521 -0.635
Income 
Under $150 – reference 
$150 - $250 -0.670 0.5916 0.257 -1.830 0.489
$250 - $300 -1.026 0.6576 0.119 -2.315 0.263
$300 - $500 -0.954 0.5565 0.086 -2.045 0.136
$500 and above -0.826 0.6307 0.190 -2.062 0.410
Marital status 
Single - reference
Having partner/boyfriend 0.158 0.5382 0.769 -0.897 1.213
Married (having husband) -0.024 0.6101 0.969 -1.219 1.172
Divorced/separated/widowed 0.033 0.5430 0.951 -1.031 1.098
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.544 0.4078 0.183 -1.343 0.256
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.220 0.4854 0.651 -0.732 1.171
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.120 0.4797 0.802 -0.820 1.060
Disclosed status as sex worker (0=no, 1=yes) 0.051 0.3117 0.871 -0.560 0.662
Ever suspected contracting HIV (0=no, 1=yes) 0.502 0.3321 0.130 -0.149 1.153
Risk index
Low risk – reference 
Medium risk -0.219 0.3569 0.539 -0.919 0.480
High risk 0.311 0.3602 0.388 -0.395 1.017
Constant 2.014 1.0091 0.046 0.036 3.992
Flagship TA (GF) Model 
Age
18-24 years – reference 
25-30 years -0.873 0.3420 0.011 -1.543 -0.202
31 years and above -0.764 0.4081 0.061 -1.564 0.036
105
Table 10 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of being in high stigma and discrimination (Continued)
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval
Education
Under 1 year – reference 
1-6 years -0.212 0.4208 0.614 -1.037 0.613
7 years and above -1.008 0.4566 0.027 -1.903 -0.113
Income 
Under $150 – reference 
$150 - $250 0.696 0.4706 0.139 -0.226 1.619
$250 - $300 -0.567 0.5491 0.302 -1.643 0.509
$300 - $500 0.202 0.4702 0.667 -0.719 1.124
$500 and above -0.015 0.4933 0.976 -0.981 0.952
Marital status 
Single - reference
Having partner/boyfriend -0.456 0.5672 0.421 -1.568 0.655
Married (having husband) -0.979 0.6558 0.136 -2.264 0.307
Divorced/separated/widowed -0.567 0.5585 0.310 -1.662 0.528
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.602 0.3521 0.087 -1.292 0.088
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.179 0.3835 0.641 -0.931 0.573
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.170 0.3460 0.623 -0.508 0.848
Disclosed status as sex worker (0=no, 1=yes) -0.319 0.2653 0.229 -0.839 0.201
Ever suspected contracting HIV (0=no, 1=yes) 0.008 0.2677 0.977 -0.517 0.532
Risk index
Low risk – reference 
Medium risk 0.109 0.2889 0.705 -0.457 0.676
High risk 0.039 0.3584 0.914 -0.664 0.741
Constant 1.805 0.8395 0.032 0.160 3.450
Non-Flagship (GF) – Model 
Age
18-24 years – reference 
25-30 years -0.817 0.3299 0.013 -1.464 -0.171
31 years and above -0.326 0.3710 0.379 -1.053 0.401
Education
Under 1 year – reference 
1-6 years 0.605 0.4193 0.149 -0.217 1.427
7 years and above -0.194 0.4652 0.676 -1.106 0.717
Income 
Under $150 – reference 
$150 - $250 -0.761 0.4218 0.071 -1.588 0.065
$250 - $300 -1.011 0.5359 0.059 -2.061 0.039
$300 - $500 -1.267 0.4454 0.004 -2.140 -0.394
$500 and above -0.844 0.5064 0.095 -1.837 0.148
Marital status 
Single - reference
Having partner/boyfriend 0.507 0.5710 0.374 -0.612 1.626
Married (having husband) -0.105 0.6704 0.876 -1.419 1.209
Divorced/separated/widowed 0.222 0.5604 0.692 -0.876 1.320
Duration living in current location 
Less than 12 months – reference 
12 months and above 0.076 0.4100 0.853 -0.728 0.880
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.547 0.4564 0.230 -1.442 0.347
106
Table 10 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of 
geographic area on the probability of being in high stigma and discrimination (Continued)
Variable Probability Robust 
Std. Err. Sig. Level 95% Conf. Interval 
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.265 0.4122 0.520 -0.543 1.073
Disclosed status as sex worker (0=no, 1=yes) 0.372 0.2618 0.155 -0.141 0.885
Ever suspected contracting HIV (0=no, 1=yes) 0.502 0.2696 0.062 -0.026 1.031
Risk index
Low risk – reference 
Medium risk -0.374 0.3108 0.228 -0.984 0.235
High risk -0.640 0.3127 0.041 -1.253 -0.027
Constant 0.391 0.7674 0.610 -1.113 1.895
No Program Model 
Age
18-24 years – reference 
25-30 years 0.235 0.3059 0.442 -0.364 0.835
31 years and above 0.637 0.3700 0.085 -0.088 1.363
Education
Under 1 year – reference 
1-6 years 0.818 0.3464 0.018 0.139 1.497
7 years and above -0.178 0.3825 0.642 -0.928 0.572
Income 
Under $150 – reference 
$150 - $250 -0.253 0.3885 0.515 -1.014 0.508
$250 - $300 -0.404 0.4910 0.411 -1.366 0.558
$300 - $500 -0.950 0.4220 0.024 -1.777 -0.122
$500 and above -0.337 0.5420 0.534 -1.399 0.725
Marital status 
Single - reference
Having partner/boyfriend 0.478 0.5976 0.424 -0.694 1.649
Married (having husband) -0.024 0.6967 0.973 -1.389 1.342
Divorced/separated/widowed -0.136 0.5688 0.811 -1.251 0.979
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.717 0.5919 0.226 -1.877 0.443
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.201 0.6632 0.762 -1.099 1.501
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.390 0.3208 0.224 -0.238 1.019
Disclosed status as sex worker (0=no, 1=yes) -0.260 0.2641 0.325 -0.778 0.258
Ever suspected contracting HIV (0=no, 1=yes) -0.270 0.2868 0.346 -0.832 0.292
Risk index
Low risk – reference 
Medium risk -0.002 0.3051 0.994 -0.600 0.596
High risk 0.341 0.3468 0.326 -0.339 1.020
Constant 0.500 0.7077 0.479 -0.887 1.888
107
Table 11 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of being in high stigma and discrimination 
Variable Probability Robust 
Std. Err.
Sig. 
Level 95% Conf. Interval
Average probability of outcome by program exposure
No exposure 0.596 0.0203 0.000 0.556 0.636
Some exposure 0.522 0.0420 0.000 0.440 0.604
High exposure 0.378 0.0230 0.000 0.333 0.423
Comparison of probability of outcome between the levels of 
program exposure 
No exposure – reference
Some exposure -0.074 0.0466 0.113 -0.165 0.017
High exposure -0.218 0.0304 0.000 -0.277 -0.158
No Exposure Model 
Age
18-24 years – reference 
25-30 years -0.347 0.2176 0.111 -0.773 0.080
31 years and above -0.068 0.2523 0.788 -0.562 0.427
Education
Under 1 year – reference 
1-6 years -0.021 0.2946 0.944 -0.598 0.557
7 years and above -0.433 0.3159 0.170 -1.053 0.186
Income 
Under $150 – reference 
$150 - $250 -0.084 0.2849 0.768 -0.642 0.474
$250 - $300 -0.278 0.3429 0.418 -0.950 0.394
$300 - $500 -0.483 0.3023 0.110 -1.075 0.110
$500 and above -0.462 0.3412 0.176 -1.130 0.207
Marital status 
Single - reference
Having partner/boyfriend -0.259 0.3898 0.507 -1.022 0.505
Married(having husband) -0.372 0.4762 0.435 -1.305 0.561
Divorced/separated/widowed -0.569 0.3879 0.142 -1.329 0.191
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.418 0.2573 0.104 -0.923 0.086
Duration working in current workplace
Less than 12 months – reference
12 months and above 0.259 0.3061 0.398 -0.341 0.859
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.118 0.2245 0.598 -0.322 0.558
Disclosed status as sex worker (0=no, 1=yes) 0.128 0.1801 0.478 -0.225 0.481
Ever suspected contracting HIV (0=no, 1=yes) 0.387 0.1952 0.047 0.004 0.770
Risk index
Low risk – reference 
Medium risk 0.076 0.2058 0.714 -0.328 0.479
High risk 0.045 0.2191 0.836 -0.384 0.475
Constant 1.217 0.5307 0.022 0.176 2.257
Some Exposure Model 
Age
18-24 years – reference 
25-30 years -0.444 0.4978 0.373 -1.419 0.532
31 years and above -0.117 0.5957 0.844 -1.285 1.050
Education
Under 1 year – reference 
1-6 years -0.454 0.5374 0.398 -1.507 0.599
7 years and above -0.409 0.6041 0.498 -1.593 0.775
108
Table 11 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of being in high stigma and discrimination (Continued) 
Variable Probability Robust 
Std. Err.
Sig. 
Level 95% Conf. Interval
Income 
Under $150 – reference 
$150 - $250 -0.855 0.6293 0.174 -2.088 0.379
$250 - $300 -1.205 0.7325 0.100 -2.640 0.231
$300 - $500 -0.630 0.6347 0.321 -1.874 0.614
$500 and above -1.502 0.7893 0.057 -3.049 0.045
Marital status 
Single - reference
Having partner/boyfriend 0.714 0.7304 0.328 -0.718 2.145
Married(having husband) 0.389 0.8971 0.665 -1.369 2.147
Divorced/separated/widowed 0.312 0.7637 0.683 -1.185 1.809
Duration living in current location 
Less than 12 months – reference 
12 months and above -1.131 0.6473 0.081 -2.399 0.138
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.337 0.7745 0.663 -1.855 1.181
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.910 0.6969 0.192 -0.456 2.276
Disclosed status as sex worker (0=no, 1=yes) -0.373 0.3915 0.341 -1.140 0.394
Ever suspected contracting HIV (0=no, 1=yes) -1.017 0.4218 0.016 -1.844 -0.191
Risk index
Low risk – reference 
Medium risk -0.093 0.4955 0.851 -1.064 0.878
High risk 0.297 0.4811 0.536 -0.645 1.240
Constant 1.863 1.0182 0.067 -0.132 3.859
High Exposure – Model 
Age
18-24 years – reference 
25-30 years -0.565 0.2613 0.031 -1.077 -0.053
31 years and above -0.496 0.3256 0.128 -1.134 0.142
Education
Under 1 year – reference 
1-6 years 0.193 0.3565 0.589 -0.506 0.891
7 years and above -0.817 0.3852 0.034 -1.572 -0.062
Income 
Under $150 – reference 
$150 - $250 0.010 0.3972 0.979 -0.768 0.789
$250 - $300 -0.521 0.4578 0.255 -1.419 0.376
$300 - $500 -0.819 0.4033 0.042 -1.610 -0.029
$500 and above -0.346 0.4277 0.418 -1.185 0.492
Marital status 
Single - reference
Having partner/boyfriend 0.347 0.4367 0.427 -0.509 1.203
Married(having husband) -0.339 0.5238 0.517 -1.366 0.687
Divorced/separated/widowed 0.206 0.4153 0.620 -0.608 1.020
Duration living in current location 
Less than 12 months – reference 
12 months and above -0.532 0.3336 0.111 -1.186 0.122
Duration working in current workplace
Less than 12 months – reference
12 months and above -0.091 0.3406 0.790 -0.758 0.577
Duration engaged in sex work
Less than 12 months – reference
12 months and above 0.167 0.3133 0.593 -0.447 0.782
Disclosed status as sex worker (0=no, 1=yes) 0.011 0.2153 0.959 -0.411 0.433
Ever suspected contracting HIV (0=no, 1=yes) 0.425 0.2190 0.052 -0.004 0.854
109
Table 11 Inverse-probability-weighed regression adjustment (IPWRA) model: The effect of program 
exposure on the probability of being in high stigma and discrimination (Continued) 
Risk index
Low risk – reference 
Medium risk -0.002 0.2556 0.994 -0.503 0.499
High risk -0.034 0.2657 0.899 -0.554 0.487
Constant 0.236 0.6430 0.713 -1.024 1.497
Cost Allocation
Management Costs
The mean management cost was $83,650.66 ($36,020.00-$165,280.00; STD $71,020.90). The costs of 
insurance, benefits, and salary for program and non-program staff , the executive director, HDM/program 
manager, M&E officer, Training & Development, FP/HIV Counselor, UIC contract staff, contact tracing 
officer etc. were included under the Management costs at the central and field offices. The total cost of 
management/support for Flagship CoE was $165,280, $49,652 for Flagship TA and $36,020 for Non-Flagship 
sites. 
IP Field office – Outreach staff Costs
The mean IP field office cost was $24,618.33 ($36,020.00-$165,280.00; STD $13,914.28). The IP Field office 
– outreach staff costs refer to the costs associated with incentives and phone cards for outreach workers 
and salary for the EW manager. The total outreach cost at the IP field offices was $38,306 for Flagship CoE, 
$25,061 for Flagship –TA program and $10,488 for Non-Flagship sites. This element accounted for 18% of 
the total Flagship CoE costs.
Flagship TA Costs 
Flagship Technical Assistance was the core component of the SMARTgirl program and used as a key 
differentiating factor between Flagship-TA and Non-Flagship sites where Flagship-TA sites received TA from 
the Flagship CoE program. The cost of the flagship TA incudes salary, benefit, and insurance for program 
and non-program staff, transportation, communication, Training/workshop for implementing partners, 
Monitoring & evaluation (including field visits) and, other (overhead and consultant) costs. The cost 
allocated to flagship TA was $16,361 and, this constituted about 12 percent of the Flagship-TA program 
costs. This allocation was only application to Flagship-TA sites. 
Administration Costs
The mean management cost was $44,884.66 ($11,381-$106,422.00; STD $53,363.02). The administration 
costs refer to the costs associated with non-program staff, (the finance manager), office expenses (office 
rental, electricity, water, office supplies, copier, printer, and ink) and capital costs (assets lasting more than 
one year, e.g. furniture and equipment). Capital costs were straight-line amortized over a useful life of 5 
years to annualize the costs. The total administration costs both at central and filed offices was $106,422 
for Flagship-CoE program, $11,381 for Flagship-TA and $16,851 for the Non-Flagship program and it 
accounted for about 49 percent of the total Flagship CoE costs.