Evaluation of the Impact of Malaria Control Interventions on All￾Cause Mortality in Children Under Five Years of Age in Kenya 2003- 2015 Kenya Malaria Impact Evaluation Group March 2017 Contents CONTENTS ...................................................................................................................................................... II LIST OF FIGURES.............................................................................................................................................. V LIST OF TABLES ............................................................................................................................................. VII ABBREVIATIONS .......................................................................................................................................... VIII ACKNOWLEDGMENTS.................................................................................................................................... IX EXECUTIVE SUMMARY.................................................................................................................................... X BACKGROUND ........................................................................................................................................................ X OBJECTIVES ........................................................................................................................................................... X EVALUATION DESIGN ............................................................................................................................................... X DATA SOURCES ...................................................................................................................................................... X EXPANSION OF INTERVENTIONS ................................................................................................................................ XI TRENDS IN COVERAGE OF INTERVENTIONS .................................................................................................................. XI Vector control............................................................................................................................................... xi Intermittent preventive treatment of malaria in pregnancy........................................................................ xi Case management of malaria..................................................................................................................... xii TRENDS IN MORBIDITY AND MORTALITY.....................................................................................................................XII Morbidity..................................................................................................................................................... xii Mortality ..................................................................................................................................................... xii CONTEXTUAL FACTORS............................................................................................................................................XII EVIDENCE THAT MALARIA INTERVENTIONS CONTRIBUTED TO A DECLINE IN ALL-CAUSE CHILD MORTALITY .............................XIII 1 INTRODUCTION ......................................................................................................................................1 1.1 CONTEXT .....................................................................................................................................................1 1.2 PURPOSE AND SCOPE .....................................................................................................................................1 1.3 EVALUATION DESIGN .....................................................................................................................................1 1.4 EVALUATION INDICATORS................................................................................................................................3 1.4.1 Insecticide-treated nets ..................................................................................................................4 1.4.2 Intermittent preventive treatment in pregnancy............................................................................4 1.4.3 Case management of malaria.........................................................................................................5 1.4.4 Malaria-related morbidity ..............................................................................................................5 1.4.5 Mortality .........................................................................................................................................6 1.5 DATA SOURCES .............................................................................................................................................6 1.6 LIMITATIONS OF EVALUATION DESIGN ...............................................................................................................6 2 COUNTRY BACKGROUND........................................................................................................................8 2.1 GEO-LOCATION AND WEATHER........................................................................................................................8 2.2 ADMINISTRATIVE DIVISIONS ............................................................................................................................9 2.3 DEMOGRAPHY AND ECONOMY.........................................................................................................................9 2.4 EDUCATION AND LITERACY ............................................................................................................................10 2.5 HEALTH SYSTEM IN KENYA ............................................................................................................................10 2.5.1 Health sector and devolution........................................................................................................10 2.5.2 Health service delivery structure...................................................................................................10 2.5.3 Human resources for health..........................................................................................................11 2.6 MALARIA CONTROL IN KENYA........................................................................................................................12 2.6.1 Epidemiology of malaria in Kenya ................................................................................................12 2.6.2 Malaria control strategy ...............................................................................................................18 2.6.3 Financing of malaria control interventions...................................................................................20 ii 3 MALARIA CONTROL INTERVENTIONS...............................................................................23 3.1 INSECTICIDE-TREATED NETS.........................................................................................................................................23 3.1.1 Background..............................................................................................................................................................................23 3.1.2 ITN policy in Kenya...............................................................................................................................................................23 3.1.3 ITN implementation .............................................................................................................................................................24 3.1.4 ITN coverage trends.............................................................................................................................................................26 3.1.5 ITN summary..........................................................................................................................................................................33 3.2 INDOOR RESIDUAL SPRAYING.......................................................................................................................................34 3.2.1 Background..............................................................................................................................................................................34 3.2.2 IRS policy..................................................................................................................................................................................34 3.2.3 IRS implementation...............................................................................................................................................................34 3.2.4 Trends in IRS implementation areas...............................................................................................................................35 3.2.5 IRS summary...........................................................................................................................................................................36 3.3 INTERMITTENT PREVENTIVE TREATMENT IN PREGNANCY ........................................................................................36 3.3.1 Background..............................................................................................................................................................................36 3.3.2 IPTp policy................................................................................................................................................................................36 3.3.3 IPTp implementation ............................................................................................................................................................37 3.3.4 SP resistance and implications for IPTp policy.............................................................................................................37 3.3.5 Trends in IPTp coverage ....................................................................................................................................................37 3.3.6 IPTp summary.........................................................................................................................................................................38 3.4 MALARIA CASE MANAGEMENT .....................................................................................................................................38 3.4.1 Background..............................................................................................................................................................................38 3.4.2 Case management policy....................................................................................................................................................38 3.4.3 Implementation of antimalarial and malaria diagnostic policy...............................................................................39 3.4.4 Trends in diagnostic capacity and AL availability........................................................................................................39 3.4.5 Trends in malaria case management.............................................................................................................................42 3.4.6 Equity in malaria case management...............................................................................................................................47 3.4.7 Malaria case management summary.............................................................................................................................48 4 TRENDS IN MALARIA MORBIDITY........................................................................................49 4.1 BACKGROUND ................................................................................................................................................................49 4.2 POPULATION-BASED TRENDS IN MALARIA MORBIDITY IN CHILDREN....................................................................49 4.2.1 Trends in prevalence of malaria parasites in children ..............................................................................................49 4.2.2 Trends in severe anemia prevalence (Hb<8g/dL) in children.................................................................................51 4.3 HEALTH FACILITY MALARIA MORBIDITY TRENDS.........................................................................................................52 4.3.1 Summary of malaria morbidity.........................................................................................................................................54 5 TRENDS IN ALL-CAUSE CHILD MORTALITY .....................................................................54 5.1 ALL-CAUSE CHILD MORTALITY....................................................................................................................................54 5.2 AGE-SPECIFIC MORTALITY.............................................................................................................................................56 5.3 EQUITY IN CHANGE IN MORTALITY.............................................................................................................................57 5.3.1 Summary of all-cause child mortality..............................................................................................................................58 6 TRENDS IN CONTEXTUAL FACTORS .................................................................................59 6.1 BACKGROUND ................................................................................................................................................................59 6.2 FUNDAMENTAL DETERMINANTS...................................................................................................................................60 6.2.1 Socioeconomic factors..........................................................................................................................................................60 6.2.2 Climatic variability..................................................................................................................................................................62 6.3 PROXIMATE DETERMINANTS .........................................................................................................................................64 6.3.1 Maternal and reproductive health indicators...............................................................................................................64 6.3.2 Child health and Immunization ........................................................................................................................................66 6.4 SUMMARY OF CONTEXTUAL FACTORS........................................................................................................................67 7 ADDITIONAL ANALYSES TO BOLSTER THE EVALUATION DESIGN...........................69 7.1 CASE STUDIES..................................................................................................................................................................69 7.1.1 Kilifi Health and Demographic Surveillance System data........................................................................................69 7.1.2 Siaya HDSS data...................................................................................................................................................................70 7.2 MULTIVARIABLE ANALYSIS .............................................................................................................................................73 7.2.1 Kaplan-Meier survival analysis..........................................................................................................................................73 iii 7.2.2 Regression analysis: Cox model.........................................................................................................................................74 8 PLAUSIBILITY ANALYSIS AND CONCLUSIONS................................................................79 8.1 INCREASED ACCESS TO AND USE OF MALARIA PREVENTION AND CONTROL INTERVENTIONS.........................79 8.2 IMPROVED MALARIA CASE MANAGEMENT ..................................................................................................................80 8.3 DECLINE IN MALARIA-RELATED MORBIDITY...............................................................................................................80 8.4 DECLINING ALL-CAUSE CHILD MORTALITY .................................................................................................................81 8.5 CONTEXTUAL FACTORS AND THE PLAUSIBILITY ARGUMENT ..................................................................................81 8.6 CONCLUSION..................................................................................................................................................................82 9 REFERENCES .............................................................................................................................83 iv List of Figures Figure 1.1: Framework for evaluating the impact of malaria control interventions on all-cause child mortality, Kenya 2003–2015 ................................................................................................................................................................2 Figure 1.2: Data sources used in the impact evaluation in Kenya, 1998–2015 ...............................................................6 Figure 2.1: Kenya maps of rainfall and temperature, 2015...................................................................................................8 Figure 2.2: Administrative map of the 47 counties in Kenya, 2015....................................................................................9 Figure 2.3: Malaria epidemiological zones in Kenya, 2015..................................................................................................13 Figure 2.4: Distribution of the dominant malaria vector species based on entomological monitoring sites in Kenya, 1900–2014 .............................................................................................................................................................15 Figure 2.5: Population-adjusted, modelled estimates of Plasmodium falciparum parasite prevalence in children ages 2–10 years (PfPR2-10) in Kenya from 2000–2015...........................................................................................16 Figure 2.6: Sources of malaria financing in Kenya, 2003–2014..........................................................................................21 Figure 2.7: Malaria expenditure by intervention in 2012 and 2014 in Kenya ................................................................22 Figure 3.1: Number of nets distributed by channel in Kenya, 2004–2015 .....................................................................26 Figure 3.2: Overall household ownership of at least one ITN in Kenya, 2003–2015 .................................................27 Figure 3.3: Household ownership of at least one insecticide-treated net by endemicity zone in Kenya, 2003– 2015.......................................................................................................................................................................................27 Figure 3.4: Household ownership of at least one insecticide-treated net by residence in Kenya, 2003–2015 ....28 Figure 3.5: Household ownership of at least one insecticide-treated net by wealth quintile in Kenya, 2003–2015 ................................................................................................................................................................................................28 Figure 3.6: Percentage of households with one insecticide-treated net per two persons in Kenya, 2003–2015 29 Figure 3.7: Insecticide-treated net use among general population, children under 5 years of age, and pregnant women among all households in Kenya, 2003–2015................................................................................................29 Figure 3.8: Insecticide-treated net use among general population, children under 5 years of age, and pregnant women in households with at least one ITN, 2003–2015.......................................................................................30 Figure 3.9: Insecticide-treated net use by household members in households with at least one ITN by malaria endemicity, 2003–2015.....................................................................................................................................................30 Figure 3.10: Malaria epidemiological zones covered by IRS implementation in Kenya from 2005–2012...............35 Figure 3.11: Pregnant women who received at least two doses of intermittent preventive treatment during pregnancy nationwide in Kenya, 2003–2015...............................................................................................................37 Figure 3.12: Pregnant women who received at least two doses of intermittent preventive treatment during their last pregnancy by endemicity in Kenya, 2003–2015 .......................................................................................37 Figure 3.13: National trends in malaria diagnostic capacity in public and non-profit health facilities in Kenya, 2010–2015 ...........................................................................................................................................................................39 Figure 3.14: National trends in the availability of artemether-lumefantrine at public-sector and non-profit health facilities on the day of the survey in Kenya, 2010–2015..........................................................................................41 Figure 3.15: National trends in artemether-lumefantrine stock-out measured in the 3 months prior to the survey, 2010–2015.............................................................................................................................................................41 Figure 3.16: Percentage of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing in Kenya, 2010-2015 .....................................................................42 Figure 3.17: Distribution of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing by endemicity zone in Kenya, 2010-2015................................42 Figure 3.18: Treatment seeking for children under 5 years of age with fever in the 2 weeks prior to the survey in Kenya, 2003–2015.........................................................................................................................................................43 Figure 3.19: Children under 5 years of age with fever in the 2 weeks prior to the survey who sought treatment and were treated with any antimalarial by endemicity zone in Kenya, 2003–2015......................44 Figure 3.20: Children under 5 years of age treated with an antimalarial who received the recommended first￾line treatment in Kenya, 2003–2015.............................................................................................................................44 Figure 3.21: Children under 5 years of age treated with an antimalarial who received the recommended first￾line treatment by endemicity zone in Kenya, 2003–2015 .......................................................................................46 Figure 3.22: Trends in the performance of malaria diagnosis and treatment in accordance with national case￾management guidelines among patients with fever seen at public and non-profit health facilities in Kenya, 2010–2015 ...........................................................................................................................................................................47 Figure 3.23: Children under 5 years of age treated with an antimalarial who received the recommended first￾line treatment by wealth quintile in Kenya, 2003–2015...........................................................................................48 Figure 4.1: Malaria parasite prevalence by microscopy in children under 5 years of age by malaria endemicity zone in Kenya, 2007–2015 ..............................................................................................................................................50 Figure 4.2: Severe anemia prevalence in children ages 6 to 59 months overall and by malaria endemicity zone in Kenya, 2007–2015 .............................................................................................................................................................51 v Figure 4.3: Monthly percentage of health facilities reporting data to the routine health information system by reporting platform in Kenya, 2012–2016.....................................................................................................................53 Figure 4.4: Malaria cases reported via the routine health information system in Kenya, 2011–2015 ....................54 Figure 5.1: Trends in all-cause child mortality in Kenya, 2003–2014 ..............................................................................55 Figure 5.2: Trends in all-cause child mortality by malaria-endemicity zone in Kenya, 2003–2014..........................55 Figure 5.3: Trends in age-specific all-cause mortality in Kenya, 2003–2014..................................................................56 Figure 5.4: Relative percentage change in age-specific all-cause mortality rates in Kenya, 2003–2014 ................56 Figure 6.1: Conceptual framework for the impact evaluation of the Kenya national malaria control program, 2003–2014 ...........................................................................................................................................................................59 Figure 6.2: Trends in gross domestic product per capita purchasing power parity and all-cause child mortality in Kenya, 2000–2014.........................................................................................................................................................60 Figure 6.3: Long-term anomalies and actual yearly rainfall in Kenya, 2000–2014 ........................................................62 Figure 6.4: Annual national mean, minimum, and maximum temperatures in Kenya, 2000–2014...........................64 Figure 7.1: Trends in inpatient malaria cases among children under 15 years of age at Kilifi County Hospital, in the coastal-endemic zone of Kenya, 2000–2015 .......................................................................................................69 Figure 7.2: Trends in all-cause and malaria-specific mortality among children under 5 years of age in Kilifi County Health and Demographic Surveillance System, 1990–2014.....................................................................70 Figure 7.3: Trends in inpatient malaria cases among children <5 years of age in two hospitals in Siaya County Health and Demographic Surveillance System in Kenya, 2003–2012...................................................................71 Figure 7.4: Trends in all-cause and malaria-specific mortality in children under 5 years of age in Siaya County Health and Demographic Surveillance System, 2003–2012....................................................................................73 Figure 7.5: Kaplan-Meier survival curves for children under 5 years of age by malaria intervention expansion period in Kenya, 2000–2014 ...........................................................................................................................................74 Figure 8.1: Summary of trends in coverage of malaria control interventions, malaria-specific morbidity, and all￾cause child mortality in Kenya, 2003–2015.................................................................................................................79 vi List of Tables Table 1.1: Population-based indicators......................................................................................................................................3 Table 2.1: Structure of the Kenya health system, 2015 ......................................................................................................10 Table 2.2: Health facilities by type and ownership in Kenya, 2015 ..................................................................................11 Table 2.3: Health worker to population ratios in Kenya, 2003–2014.............................................................................12 Table 2.4: Milestones in malaria policies and intervention implementation in Kenya, 1981–2015..........................19 Table 3.1: Nets distributed in Kenya, 2005–2015 ................................................................................................................24 Table 3.2: Use of Insecticide-treated nets among children under 5 years by background characteristic in Kenya, 2003–2015 ...........................................................................................................................................................................31 Table 3.3: Use of Insecticide-treated nets among pregnant women by background characteristic in Kenya, 2003–2015 ...........................................................................................................................................................................33 Table 3.4: Percent of targeted households protected by IRS in the highland epidemic-prone counties, 2005– 2012.......................................................................................................................................................................................35 Table 3.5: Distribution of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing, by demographic characteristic, in Kenya, 2010–2015.........43 Table 3.6: Use of recommended first-line treatment among children under 5 years of age who took an antimalarial, by demographic characteristic, in Kenya, 2003–2015 ....................................................................46 Table 4.1: Malaria parasite prevalence in children under 5 years of age by demographic characteristic in Kenya, 2007–2015 ...........................................................................................................................................................................50 Table 4.2: Severe anemia prevalence in children ages 6 to 59 months by demographic characteristic in Kenya, 2007–2015 ...........................................................................................................................................................................52 Table 5.1: All-cause child mortality, by demographic characteristic, in Kenya, 2003–2014....................................57 Table 6.1: Education and marital status of women ages 15-49 years in Kenya, 2003–2014......................................61 Table 6.2: Change in household attributes and asset ownership in Kenya, 2003–2014.............................................61 Table 6.3: Trends in maternal and reproductive health indicators in Kenya, 2003–2014..........................................66 Table 6.4: Summary of child health indicators in Kenya, 2003–2014 ..............................................................................67 Table 6.4: Summary of evidence associated with all-cause child mortality in Kenya, 2003–2015 ...........................68 vii Abbreviations ACCM all-cause child mortality ACT artemisinin-based combination therapy An Anopheles ANC antenatal care BSc bachelor of science CQ Chloroquine CHEW community health extension worker CHV community health volunteer DFID United Kingdom Department for International Development DHIS2 District Health Information Software 2 DPT diphtheria, pertussis, tetanus DVBD Division of Vector Borne and Neglected Diseases FBO faith-based organization GDP gross domestic product GOK Government of Kenya HDSS health and demographic surveillance system Hib Haemophilus influenzae type B IDSR Integrated Disease Surveillance and Response IGME United Nations Inter-agency Group for Child Mortality Estimation IPTp intermittent preventive treatment in pregnancy IRS indoor residual spraying ITN insecticide-treated net KDHS Kenya Demographic and Health Survey KEMRI Kenya Medical Research Institute KMIS Kenya Malaria Indicator Survey KNBS Kenya National Bureau of Statistics LiST Lives Saved Tool LLIN long-lasting insecticidal net MCH maternal and child health MCU Malaria Control Unit MERG Monitoring & Evaluation Reference Group MOH Ministry of Health MOPHS Ministry of Public Health and Sanitation NGO nongovernmental organization NMCP National Malaria Control Programme NMS National Malaria Strategy PfPR Plasmodium falciparum prevalence PMI United States President’s Malaria Initiative PPP purchasing power parity RBM Roll Back Malaria Partnership RDT rapid diagnostic test RHIS routine health information system SP sulfadoxine-pyrimethamine USAID United States Agency for International Development WHO World Health Organization viii Acknowledgments This evaluation was funded by the United States President’s Malaria Initiative (PMI) through the United States Agency for International Development (USAID). The report was produced through collaborative efforts of numerous partners. The National Malaria Control Program expresses appreciation to Abdisalan Noor, Robert W. Snow, Polycarp Mogeni, Philip Bejon, Tom Williams, Peter Macharia, Paul Ouma, Joseph Maina, Stephen Oloo, Ezekiel Gogo, and Lukio Olweny from the Kenya Medical Research Institute (KEMRI) Wellcome Trust Research Programme for their contributions to the intervention coverage and morbidity data analysis, including providing demographic surveillance data from Kilifi County; to Simon Kariuki and Vincent Were and the KEMRI-Centers for Disease Control and Prevention (CDC) program for mortality analysis and providing the demographic surveillance data from Siaya County. Thank you to the technical committee, including Ann M. Buff (PMI￾CDC), Christine Hershey (PMI-USAID), Patricia Njiri (Clinton Health Access Initiative), Robert Kimbui and Josea Rono (Management Sciences for Health), Tereza Kinyari (University of Nairobi), Steven Aloo (Population Services Kenya), David Gikungu (Kenya Meteorological Department), Evan Mathenge and Nathan Bakyaita (World Health Organization), and Elizabeth Juma (KEMRI). Hellen Gatakaa, Yazoume Ye, Agneta Mbithi, Samuel Oti (consultant), and Donatien Beguy (consultant) from MEASURE Evaluation and ICF, contributed to the design, mortality analysis and provided technical guidance during the report writing process. Thank you to the malaria surveillance, monitoring and evaluation technical working group for the overall oversight. Special thanks to the National Malaria Control Program team, led by Ejersa Waqo, Rebecca Kiptui, and Andrew Wamari, for their leadership and coordination of the evaluation. ix Executive Summary Background Malaria remains a significant cause of morbidity and mortality in Kenya, particularly among young children. Approximately 70 percent of the population is at risk of the disease, but the malaria burden in Kenya is not homogenous. There are four main malaria epidemiological zones with risk diversity determined largely by altitude, rainfall patterns, and temperature. The four transmission zones are malaria-endemic (i.e., Lake Victoria and Indian Ocean coast), epidemic-prone (i.e., western highlands), seasonal-risk (i.e., primarily arid and semi-arid areas of the north and northeast), and low-risk including Nairobi. Malaria accounts for 16 percent of outpatient visits to health facilities nationally. From 2003 to 2015, the Government of Kenya and partners increased funding for malaria control significantly, which resulted in expansion of key interventions, including insecticide-treated nets (ITNs), indoor residual spraying (IRS), intermittent preventive treatment during pregnancy (IPTp), and prompt and effective malaria case management through prompt parasitological diagnosis and treatment. Measuring the effects of these investments is necessary to inform the National Malaria Control Programme (NMCP), partners and donors and to provide evidence for future policies, strategies, and activities to further reduce the burden of malaria. Kenya’s Ministry of Health and the United States President’s Malaria Initiative (PMI), on behalf of the Monitoring and Evaluation Reference Group (MERG) of the Roll Back Malaria (RBM) Partnership, commissioned this evaluation to report the impact of malaria prevention and control investments during the period of 2003–2015 in Kenya. Objectives • Measure the degree to which malaria control interventions were implemented in Kenya. • Assess malaria-related morbidity, mortality, and contextual factors before, during, and after the implementation of malaria control interventions in Kenya. • Assess the plausible attribution of the expansion of malaria control interventions to changes in malaria-related morbidity, all-cause mortality, and malaria-related mortality among children under 5 years of age in Kenya. Evaluation Design The evaluation was based on pre- and post-assessments, which used a plausibility evaluation design, which measured changes in coverage of malaria interventions, malaria­related morbidity, and all-cause child mortality (ACCM) in children under age 5 years, while accounting for other contextual determinants of child survival during the evaluation period. The primary measure of impact was ACCM. Multivariable analysis was included to investigate associations between household ITN ownership and ACCM. Due to the variability in malaria transmission in Kenya, the analyses were disaggregated by malaria-epidemiological zones. Data Sources Data on intervention coverage, morbidity, and mortality were primarily from nationally representative, population-based household surveys. These surveys were the, 2003, 2008/9 and 2014 Kenya Demographic and Health Survey (KDHS) and 2007, x 2010 and 2015 Kenya Malaria Indicator Surveys. Data from other sources, including the routine health information system (RHIS) and health and demographic surveillance system (HDSS) (via case studies), were used to complement and triangulate findings and to provide examples of subnational changes. Expansion of Interventions Between 2003 and 2015, the country substantially increased coverage of malaria prevention interventions, particularly ITNs and IRS (in more limited targeted areas from 2005 to 2012). Mass distribution of free long-lasting insecticidal nets (LLINs) were conducted to expand coverage in children under age 5 years and pregnant women in 2006 and since 2011, to ensure universal coverage (i.e., at least one net per two persons per household). The country adopted IPTp in 1998 and incorporated the treatment as part of the routine antenatal care (ANC) package of services. Subsequently, efforts to improve coverage targeting both community and facility￾based health workers were implemented at various points of the evaluation period. Artemisinin-based combination therapy (ACT) was adopted as first-line treatment for uncomplicated malaria in 2004 and provided at no cost to patients in the public health sector in 2006. Access to accurate malaria diagnostic testing and effective case management according to treatment guidelines also significantly improved from 2011 to 2015. Trends in Coverage of Interventions Vector control National household ownership of at least one ITN increased significantly, from 8 (95 confidence interval [CI]: 7–9) percent in 2003 to 63 (95% CI: 59–66) percent in 2015; in areas specifically targeted for distribution, such as the lake-endemic region, ownership increased from 12 (95% CI: 9–16) percent in 2003 to 87 (95% CI: 82–90) percent in 2015. In households with at least one ITN, use among children under age 5 years increased from 66 (95% CI 66–77) percent in 2007 to 79 (95% CI 69–82) percent in 2015 and use among pregnant women increased from 70 (95% CI 63–75) to 82 (95% CI 76–87) percent during the same period. IRS implementation for epidemic response started in 1999 in parts of the western highland epidemic-prone and seasonal-risk zones. Due to poor surveillance and weak prediction systems, particularly in the highland epidemic-prone zone, a government￾led, coordinated IRS plan for epidemic prevention was introduced in 2005. IRS was conducted only in parts of counties within the highland epidemic-prone zone from 2005 to 2009. From 2010 to 2012, IRS was also conducted in parts of malaria-endemic counties bordering the highland epidemic-prone zone for vector control and malaria burden reduction. Intermittent preventive treatment of malaria in pregnancy Coverage with at least two doses of IPTp at the national level increased from 5 (95% CI: 4–6) percent to 15 (95% CI 13–17) percent from 2003 to 2008. Since 2009, IPTp has been targeted to the malaria-endemic zone only, in line with World Health Organization (WHO) recommendations. Uptake of IPTp2 increased in the coastal and lake malaria-endemic zones from 22 (95% CI: 12–35) percent and 29 (95% CI: 23– 35) percent in 2010 to 60 (95% CI: 50–68.3) percent and 53 (95% CI: 43–62) percent xi in 2015, respectively. Kenya adopted IPTp3 as policy in late 2015, putting it beyond the evaluation period. Case management of malaria Case management of children with fever also improved over this period. In 2010, 55 (95% CI: 48–63) percent of public-sector health facilities had malaria diagnostic capacity (via microscopy, rapid diagnostic tests [RDT] or both), and by 2015, 97 (95% CI: 93–99) percent had diagnostic capacity. The percentage of children under age 5 years with fever who sought treatment from appropriate providers was 60 (95% CI: 57–63) percent in 2003 and increased to 70 (95% CI: 66–74) percent in 2015. The proportion of children tested was 13 (95% CI: 10–17) percent in 2010 and increased to 39 (95% CI: 35–44) percent in 2015. The proportion of children who received the recommended first-line antimalarial treatment decreased from 42 (95% CI: 36–47) percent in 2003, when sulfadoxine-pyrimethamine (SP) was the first-line antimalarial, to a low of 11 (95% CI: 8–16) percent in 2007 one year after ACTs were introduced, and thereafter increased to 92 (95% CI: 88–95) percent in 2015. Overall, there was a reduction in the use of antimalarials to treat fever in children under age 5 years in non-endemic zones, perhaps indicating both a decline in malaria prevalence and an increase in the use of parasitological diagnosis before treatment. Trends in Morbidity and Mortality Morbidity In Kenya, malaria parasitemia prevalence among children ages 6–59 months declined over the evaluation period nationally from 9 (95% CI: 7–12) percent in 2010 to 5 (95% CI: 4–7) percent in 2015. The decrease observed nationally was mainly due to decreases in the malaria-endemic zone, where prevalence declined from 33 (95% CI: 26–41) percent in 2010 to 17 (95% CI: 12–21) percent in 2015. Similarly, at the national level, the prevalence of severe anemia (i.e., hemoglobin less than 8 g/dL) among children ages 6–59 months declined from 4 (95% CI: 3–5) percent in 2010 to 2 (95% CI: 2–3) percent in 2015. Reductions in severe anemia prevalence were observed in all malaria risk zones between 2010 and 2015, with a greater decline in the malaria-endemic zone. Mortality In Kenya, at the national level, ACCM declined from 115 deaths per 1,000 live births in 1999–2003 to 52 deaths per 1,000 live births in 2010–2014, a 54 percent reduction. There were similar reductions in mortality rates among neonatal, post-neonatal, infant, and child populations. The reduction in ACCM was greatest among children living in the highest-risk zone compared to those in the lower-risk zones. Data from two HDSS sites confirm the substantial reduction in ACCM in malaria-endemic zones. Contextual Factors During the evaluation period, there were substantial changes in fundamental and proximal determinants of child survival in Kenya. The gross domestic product (GDP) per capita purchasing power parity (PPP) increased from US$2,146 in 2003 to US$2,818 in 2014. Most standard household attributes and assets ownership improved; the largest increases were in ownership of a telephone and access to improved sources of water and electricity. At the individual level, more women had xii completed at least a primary education and women’s literacy improved from 2003 to 2014. Maternal health indicators, including health facility-based births, improved during the evaluation period. Improvements in child health included substantial increases in coverage of immunizations and vitamin A supplementation, and care￾seeking for acute respiratory infections and diarrhea. The prevalence of stunted growth in children under 5 years declined by 4 percent from 2003 to 2014. HIV prevalence among women ages 15 to 49 years declined from 9 percent in 2007 to 7 percent in 2012, and coverage of antiretroviral drugs increased from 39 percent in 2007 to 61 percent in 2012. Evidence That Malaria Interventions Contributed to a Decline in All￾cause Child Mortality The findings indicate that from 2003 to 2015, Kenya made remarkable progress towards increasing coverage of malaria prevention and control interventions for populations at risk of malaria. Household ownership and use of ITNs among pregnant women, young children, and general household members increased as did uptake of IPTp for prevention of malaria in pregnancy. Effective case management also improved nationally and particularly in the malaria-endemic zone. The expansion of malaria intervention coverage resulted in national reductions in malaria cases reported and prevalence of malaria parasitemia and severe anemia, with the greatest declines observed in the malaria-endemic zone. During the same period, Kenya experienced substantial socioeconomic progress and improvements in non-malaria health intervention coverage. However, these improvements alone are unlikely to account fully for the 54 percent reduction in under 5-year mortality between 2003 and 2014. Based on malaria prevention and control intervention coverage patterns, ACCM patterns in the malaria epidemiological zones, and timing, it is plausible that malaria prevention and control interventions contributed to the decline in under 5-year mortality in Kenya. The period of rapid reduction in under-5 mortality from 2003 to 2014 coincided with the rapid expansion of ITN distribution and with increased use of ITNs and ACTs. The Kaplan-Meir estimates clearly demonstrate improved survival during the 2010– 2014 period, which coincides with increased malaria intervention coverage, compared to the earlier periods prior to and during expansion of malaria interventions. The Cox proportional hazards regression analysis also indicates a stronger protective effect of ITN ownership on child survival in the lake-endemic area, where the potential to benefit from malaria interventions was higher compared to other epidemiological zones. Therefore, the declining trends in under 5-year mortality in Kenya are consistent with the expected impact of the expansion of malaria control interventions. xiii 1 Introduction 1.1 Context Malaria is one of the leading causes of morbidity and mortality in Kenya. It accounts for 16 percent of all outpatient attendance and 15 percent of all admissions to health facilities nationwide (Ministry of Health, 2013). Malaria has been estimated to cause 20 percent of all deaths in children under age 5 in Kenya (Ministry of Health, 2014a). Increasing evidence shows that the epidemiology and risk of malaria in Kenya have changed over the past 15 years. However, the country is still in the control phase of the malaria elimination continuum with malaria prevalence in some endemic counties as high as 38 percent (NMCP, KNBS and ICF-International, 2016). Since 2000, there has been a rapid increase in the expansion of malaria control interventions due to increased investments by partners and donors. Policy changes and new interventions have been informed by global and regional recommendations and changes in the malaria situation in Kenya. 1.2 Purpose and Scope Kenya’s Ministry of Health and the United States President’s Malaria Initiative (PMI), on behalf of the Roll Back Malaria (RBM) Partnership, commissioned an evaluation to measure the impact of expansion of key malaria interventions on all-cause mortality in children under 5 years of age during the evaluation period of 2003–2015. This report describes the trends in increased coverage of interventions as well as variations in coverage by malaria epidemiological zones. The evaluation also considers other contextual factors that may have contributed to the mortality decline over the period. The purpose of this report is to measure the impact of the massive investments and substantial expansion of malaria prevention and control interventions from 2003 to 2015. It also seeks to provide information to the Ministry of Health, National Malaria Control Programme (NMCP), and partners for evidence￾based decision making for future malaria control policies, strategies, and activities. 1.3 Evaluation Design The evaluation used a non-experimental design, a pre- and post-intervention expansion, to measure changes in malaria intervention coverage, malaria-related morbidity, and all-cause child mortality (ACCM) while documenting other contextual determinants of child survival and assembling a plausibility argument that malaria interventions contributed to the observed outcomes during the evaluation period (Rowe et al., 2009, Victora et al., 2011). Plausibility inferences assume that mortality reductions can be attributed to program efforts if improvements are found in population-level measurements of steps in the causal pathway of the impact model (Habicht et al., 1999). Specifically, for this impact evaluation, the underlying argument was the biological plausibility of causal association between malaria interventions, malaria-related morbidity, malaria-related mortality, and all-cause mortality in children under age 5 years. 1 If there is reduction in ACCM and anemia as malaria intervention coverage increases, the conclusion that malaria control activities reduced malaria-associated mortality becomes more plausible, assuming no major confounding factors are present (Carneiro et al., 2010). Indicators at the end of the evaluation period were compared with the counterfactual, which is the assumption that pre-intervention trends would have continued. Using plausibility inferences based on ecologic associations, these trends were then used to demonstrate whether mortality reductions were attributable to programmatic efforts to expand malaria control interventions. For this reason, the evaluation examined the levels and trends in ACCM, malaria parasitemia, severe anemia, coverage of malaria control interventions (i.e., long-lasting insecticidal nets [LLINs], indoor residual spraying [IRS], intermittent preventive treatment in pregnancy [IPTp], and prompt and effective case management), and other contextual determinants of child survival. The contextual factors were classified into fundamental and proximate determinants. Fundamental determinants included climatic factors, socioeconomic factors such as gross domestic product (GDP), education, and access to improved water and sanitation. Proximate determinants included access to health services, fertility-related risks, immunization coverage, HIV prevalence and interventions, and other predictors of maternal and child health such as nutrition and comorbidities, particularly diarrhea and pneumonia. The evaluation used all-cause mortality in children under 5 years as the measure of impact because there was no reliable measure of malaria-specific mortality at the national scale. The Mortality Task Force of Roll Back Malaria’s Monitoring & Evaluation Reference Group (2014), suggests that evaluation of full-coverage malaria control programs relies on an ecological study design, often referred to as a plausibility study design (Figure 1.1), because of challenges with measuring malaria￾specific mortality and defining a control or comparison group required for a traditional impact evaluation. Figure 1.1: Framework for evaluating the impact of malaria control interventions on all￾cause child mortality, Kenya 2003–2015 2 Table 1.1: Population-based indicators Intervention area Indicator Prevention Vector control 1. Proportion of households with at least one ITN . 2. Proportion of households with at least one ITN for every two people. 3. Proportion of population who slept under an ITN the previous night. 4. Proportion of children under 5 years who slept under an ITN the previous night. Note: LLIN – long-lasting insecticidal bed net; IRS – indoor residual spraying; ANC – antenatal care; GDP – gross domestic product; IPTp – intermittent preventive treatment of malaria in pregnancy; Vit. A – vitamin A Source: Ye et al 2017 In addition to trend analyses, survival analyses were performed to compare the survival probability of children ages 6 to 59 months before (2000–2004), during (2005–2009), and after (2010–2014) the expansion of malaria control interventions in Kenya. The evaluation assessed routine health information systems (HIS) data and health and demographic surveillance system (HDSS) data from two sites to futher support the plausibility argument. Furthermore, the Lives Saved Tool (LiST) (JHSPH, 2014), based on a mathematical model, was used to estimate the number of under-5 deaths potentially averted due to the expansion of LLINs between 2000 and 2014. The LiST model is used in this context as an advocacy tool and not necessarily as evidence of the impact of malaria interventions. Further details on the LiST model are in Annex 1. 1.4 Evaluation Indicators The evaluation used the indicators and definitions as recommended by Roll Back Malaria Monitoring & Evaluation Reference Group (RBM-MERG) and used by NMCP. The indicators are shown in Table 1.1. 3 5. Proportion of pregnant women who slept under an ITN the previous night. Prevention of malaria in pregnancy 6. Proportion of women who received at least two doses of sulfadoxine-pyrimethamine for IPTp for malaria during antenatal visits during last pregnancy. Case management Diagnosis 7. Proportion of children under 5 years with fever in the previous two weeks who had blood taken from a finger or heel for testing (which is a proxy for malaria testing). Treatment 8. Proportion of children under 5 years with fever in last two weeks for whom advice or treatment was sought from an appropriate provider. 9. Proportion of children under 5 years with fever who received any antimalarial treatment.* 10. Proportion of children under 5 years with fever who received the recommended antimalarial treatment within 24 hours.* 11. Proportion of children under 5 years with fever who received recommended first-line treatment (artemisinin combination treatment) among those who received any antimalarial treatment. Impact Measurement Morbidity 12. Proportion of children ages 6 to 59 months with malaria infection measured by microscopy. 13. Proportion of children ages 6 to 59 months with a hemoglobin of <8g/dl. Mortality 14. All-cause mortality in children under 5 years. Note: ITN – insecticide-treated bed net; IPTp – intermittent preventive treatment in pregnancy *These indicators are no longer recommended by Roll Back Malaria-Monitoring and Evaluation Reference Group but are included here because they are still used to track national malaria control program targets. Source: Roll Back Malaria-Monitoring and Evaluation Reference Group. Household Survey Indicators for Malaria Control, 2013. 1.4.1 Insecticide-treated nets RBM-MERG net indicators report on both ownership and use. ITN ownership is a household-level indicator, whereas ITN use is an individual-level indicator. Use at the population level was measured historically in populations with the greatest risk of malaria morbidity and mortality, such as children under 5 years and pregnant women. 1.4.2 Intermittent preventive treatment in pregnancy 4 IPTp coverage is measured by a RBM-MERG population-based indicator. Until October 2012, IPTp was defined as at least two doses of sulfadoxine-pyrimethamine (SP) after quickening and at least one month apart; these recommendations have since changed to give IPTp at each antenatal care (ANC) visit after the first trimester (WHO, 2012). This evaluation reported on two or more doses, which is in line with the Kenya policy that was in place during the evaluation period. IPTp3 was included in the revised strategy in October 2015 and a circular on IPTp3 implementation was sent to counties in October 2016, falling outside the period under evaluation. 1.4.3 Case management of malaria RBM-MERG population-based indicators measure some elements of diagnosis and treatment of malaria. Facility-based data are often better suited to monitoring trends in malaria case management (such as proportion of suspected malaria cases tested, test positivity rate, and proportion of confirmed malaria cases receiving artemisinin-based combination therapy [ACT]) and are included in this report where relevant. Population￾based surveys measure the proportion of children with fever receiving diagnostic tests for malaria, which is measured by a proxy indicator. Having blood taken from a finger or heel stick is a proxy indicator for having had a diagnostic test. Questions on care-seeking behavior for fever in children under 5 years and the type and timing of treatment with antimalarial drugs are also included. 1.4.4 Malaria-related morbidity The prevalence of severe anemia (i.e., hemoglobin levels less than 8 g/dL) and prevalence of parasitemia in children ages 6 to 59 months were two outcomes examined. Severe anemia is a potential impact measure for total malaria-related disease burden, measurable at the population level with less seasonal variations than parasite prevalence (McElroy et al., 2000, Menendez et al., 1997, Snow et al., 1997). The prevalence of parasites is the most direct measure of malaria burden. Cross￾sectional household surveys provide national and epidemiological-zone estimates of parasite prevalence to measure the impact of malaria control interventions. In addition, morbidity analyses were supplemented by routine surveillance data on malaria cases (i.e., malaria test positivity rates). HIV Bias in Mortality Measurement Mortality estimates in this report were not adjusted for HIV. In high-prevalence countries, maternal AIDS-related deaths result in missing birth histories of children with elevated mortality risk. The United Nations Inter-agency Group for Child Mortality Estimation has developed methods of HIV adjustment for child mortality estimates and recommends using this adjustment in countries where more than 5 percent of adult women are infected with HIV. However, other analyses of potential HIV bias by Rajaratnam et al. (2010) using Demographic and Health Surveys data from 21 countries show substantial variation in effect on child mortality estimates in both directions, even in countries where HIV prevalence exceeds 20 percent. Data from the Kenya AIDS Indicator Survey show that HIV prevalence among women ages 15 to 49 declined from 9.0 percent in 2007 to 6.9 percent in 2012 but remained far below 20 percent, indicating that any bias introduced in child mortality estimates by HIV has likely not changed over the study period. Finally, improvement in coverage of antiretroviral drugs (from 39.2 percent in 2007 to 60.5 percent in 2012, among all infected) and prevention of mother-to-child transmission over the evaluation period should reduce any favorable bias (exaggeration of mortality decline) because birth histories of HIV￾positive mothers are progressively more likely to be included over time due to improved survival of HIV-infected women. Therefore, in this evaluation report, the mortality estimation methods do not take into account potential selection bias arising from high HIV prevalence, hi h b li it ti f th t 5 1.4.5 Mortality In line with RBM-MERG guidance, the principal measure of impact in this evaluation is ACCM (Rowe et al., 2007, RBM, 2013, Rajaratnam et al., 2010). ACCM is preferable because population-level national data on malaria-specific mortality are not available. In addition, there are concerns about the sensitivity and specificity of the verbal autopsy method for detecting malaria deaths (Snow et al., 1997). Furthermore, malaria is thought to make an indirect contribution to ACCM that is equivalent to 50 percent to 100 percent of the mortality that can be directly attributed to malaria (Murphy and Breman, 2001). Mortality estimates used in this evaluation are derived from multiple Demographic and Health Surveys data sets rather than mortality estimates available from the United Nations Inter-agency Group for Child Mortality Estimation (IGME) (UN IGME, 2014). The level and detail of stratification needed to inform the plausibility design of this evaluation was not possible using IGME estimates. Furthermore, it is more appropriate to assess impact based on actual measured morbidity and mortality, rather than modeled estimates. 1.5 Data Sources Intervention coverage, morbidity, and mortality data were obtained primarily from large population-based household surveys, including the Kenya Demographic and Health Survey (KDHS) in 1998, 2003, 2008/2009, and 2014 and the Kenya Malaria Indicator Survey (KMIS) in 2007, 2010, and 2015. Other sources of data that included HDSS data complemented findings and served as examples of subnational changes. Routine health data available via the HIS and integrated disease surveillance and response (IDSR) data were reviewed to complement survey findings as was data from quality of care (QOC) surveys for malaria case management. The annual economic survey reports were used to describe health system factors. Historical meteorological data were used to describe rainfall and temperature variation which occurred during the evaluation period. The evaluation identified other potential data sources, and selected the most relevant (Figure 1.2) after assessing the validity, strengths, and limitations of each source. A more detailed description of the data sets, survey methods, sample sizes, and other statistical parameters is in Annex 2. Figure 1.2: Data sources used in the impact evaluation in Kenya, 1998–2015 Note: KDHS – Kenya Demographic and Health Survey; HDSS – health and demographic surveillance system; HMIS –health management information system; IDSR – Integrated Disease Surveillance and Response; KMIS – Kenya Malaria Indicator Survey; QOC – quality of care for malaria case management 1.6 Limitations of Evaluation Design The impact evaluation was limited to the scope of the secondary data sets that were included in the analysis. Each of these sources has relative strengths and limitations, 6 particularly with respect to the level of data aggregation. By applying the defined inclusion criteria across the data sets, the evaluation attempted to minimize the effect of inherent limitations. The main shortcoming of plausibility inferences, as compared to probability inferences (derived typically from randomized controlled trial study designs), is that the observed difference and magnitude of changes in ACCM might not be entirely attributable to the expansion of malaria control interventions, and the role of other contextual factors cannot be adequately measured and controlled. 7 2 Country Background 2.1 Geo-location and Weather Kenya is in East Africa and lies between 5 degrees north and 5 degrees south latitude and between 24 and 31 degrees east longitude. The country is bordered by Ethiopia to the north, South Sudan to the northwest, Uganda to the west, Tanzania to the south, and Somalia and the Indian Ocean to the east. The country covers a total area of 582,646 square kilometers, 2 percent of which comprises water bodies. It also has diverse physical features: Mount Kenya, the second highest peak in Africa, Lake Victoria, and the Great Rift Valley, which runs from Lake Turkana in the north through the Great Lakes region to the south. The country has a number of large rivers, including the Tana, Galana, Turkwel, and Nzoia. Eighty percent of the land area is arid or semi-arid. The country has a hot and humid tropical climate along the 400 km Indian Ocean coastline and Lake Victoria, is temperate in the savanna grasslands around the capital, Nairobi, and is increasingly cooler toward Mount Kenya. The arid and semi-arid areas, the savanna plateau and the coastal hinterland, have considerably lower seasonal rainfall (Figure 2.1a) with an annual average of less than 250 to 500 mm. The Lake Victoria region and the western and central highlands receive the highest rainfall in the country and exhibit less seasonality. The “long rains” occur from March to June. The “short rains” occur from October to December. The hottest period is from February to March, and the coldest period is from July to mid-August. The varied topography and altitude contribute to large variations in ambient temperature (Figure 2.1b). Figure 2.1: Kenya maps of rainfall and temperature, 2015 a) Mean monthly rainfall (mm) b) Mean temperature ( o C) Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 8 2.2 Administrative Divisions Administratively, the country was divided into 47 counties in 2013, which replaced the previous administrative divisions of eight provinces and nearly 300 districts. The counties are shown in Figure 2.2. Figure 2.2: Administrative map of the 47 counties in Kenya, 2015 Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 2.3 Demography and Economy Kenya’s population was 38.6 million in 2009 and estimated at 43.0 million in 2014, with an inter-censual growth rate of 2.9 percent per annum (KNBS, 2015). The total fertility rate in 2014 was 4.8 (KNBS and ICF-Macro, 2015), while overall life expectancy at birth was 61.5 years (UNDP, 2015). Kenya is home to people of diverse cultures with more than 42 ethnic groups and as many languages. The Kenyan economy is predominantly agricultural, with a strong industrial base; agriculture is the livelihood for 80 percent of the population. The performance of the Kenyan economy during the period under review was varied, with an average growth rate of 5.1 percent (World Bank, 2016). The highest growth rate recorded was 8.4 percent in 2010 and the lowest rate was 0.2 percent in 2008, which coincided with the global recession. The GDP in 2014 was 60.9 billion current international dollars (Intl$) compared with 12.7 billion in 2003. The gross national income per capita increased from Intl$1,690 in 2003 to Intl$2,940 in 2014 (World 9 Bank, 2016). At the start of the millennium, approximately 50 percent of the population lived below the poverty line, dropping to 43 percent in 2012. Kenya’s human development index score increased by 22.6 percent from 0.447 in 2003 to 0.548 in 2014, positioning the country at 145 among 188 countries (World Bank, 2016). 2.4 Education and Literacy Kenya had a national adult literacy rate of 72 percent, with higher literacy among males (78 percent) compared to females (67 percent), from 2008 to 2012 (World Bank, 2014). Youth ages 15 to 24 years had a literacy rate of 83 percent for males and 82 percent for females (World Bank, 2014). 2.5 Health System in Kenya 2.5.1 Health sector and devolution In March 2013, Kenya transitioned to a devolved system of government that moved fiscal, planning, and oversight functions from central authorities to 47 new counties. One of the devolved functions was health, with specific functions assigned to national and county governments to facilitate progressive realization of the right to health by all. Policies and guidelines, norms and standards, capacity building, national referral hospitals, and technical assistance were assigned to the national government; county governments became responsible for health service delivery, including routine malaria prevention and control interventions and services. 2.5.2 Health service delivery structure Kenya’s health delivery system is pyramidal, consisting of community, primary (i..e., dispensaries and health centers), secondary (i.e., county referral hospitals), and tertiary (i.e., national referral hospitals) levels of care (Table 2.1). Table 2.1: Structure of the Kenya health system, 2015 Current structure Previous structure Tier 1 Community health services Level 1 Community health services Tier 2 Primary care facilities Level 2 Dispensaries and clinics Level 3 Health centers Tier 3 County referral facilities Level 4 Primary care hospitals Level 5 Secondary care hospitals Tier 4 National referral facilities Level 6 National referral hospitals The six-level health service delivery system was in existence prior to devolution and was transitioned to a four-tier system in 2013 (Table 2.1), which consists of: • Community health services: Includes all community-based health activities such as malaria community case management services delivered by community health volunteers (CHV). • Primary care facilities: Includes all dispensaries, health centers, and maternity homes where the majority of malaria prevention and control interventions such as LLINs, IPTp and case management are delivered. • County referral facilities: Includes hospitals that are operated and managed by county governments; most hospitals have large-volume outpatient 10 departments that provide malaria prevention and control interventions such as LLINs, IPTp and case management. • National referral facilities: Includes specialized referral hospitals that are operated and managed by the national government. At the community level, CHVs provide health promotion and limited preventive and curative services. They are supervised by community health extension workers (CHEW); all CHVs and CHEWs are linked to a primary-care facility. Primary care is delivered through dispensaries, health centers and outpatient department clinics at hospitals where curative, reproductive, maternal and child health, and preventive services are offered. County hospitals provide both primary and secondary care while national referral hospitals provide primarily tertiary care services. The main provider of health services is the Ministry of Health, which manages 4,201 (44 percent) of 9,506 health facilities (Table 2.2). The private non-profit sector also plays a significant role in service provision, with faith-based organizations (FBO) operating 11 percent and nongovernmental organizations (NGO) operating slightly more than 3 percent of all health facilities. The private for-profit sector operates 38 percent of health facilities, the majority of which are outpatient clinics. Overall, 95 percent of health facilities are primary-care facilities, and 5 percent are hospitals. Table 2.2 shows health facilities in Kenya in 2015 by type and ownership. Table 2.2: Health facilities by type and ownership in Kenya, 2015 Public Private non-profit Private Tier MOH Total Local Other Gov’t FBO NGO Private Authorities Level 2 3,195 83 250 768 293 3,155 7,744 Level 3 739 26 19 188 35 292 1,299 Level 4 252 0 2 67 7 115 443 Level 5 11 1 0 1 0 3 16 Level 6 4 0 0 0 0 0 4 Total 4,201 110 271 1,024 335 3,565 9,506 Note: FBO – faith-based organization; MOH – Ministry of Health; NGO – nongovernmental organization Source: Kenya Master Facility List, May 2015 2.5.3 Human resources for health The health worker to population ratio has improved over the period under review, with the number of medical personnel increasing from 192 per 100,000 population in 2003 to 282 per 100,000 in 2014 (KNBS, 2004, KNBS, 2015). Table 2.3 shows the cadres of health worker to population ratios. 11 Table 2.3: Health worker to population ratios in Kenya, 2003–2014 Type of personnel 2003 2014 Number Number per 100,000 population Number Number per 100,000 population Doctors 4,813 15 9,149 21 Dentists 772 3 1,90 3 Pharmacists 1,881 6 2,355 5 Pharmaceutical Technologists 1,405 4 7,041 16 BSc. Nurses - - 2,418 6 Registered Nurses 9,869 33 41,371 96 Enrolled Nurses 30,212 100 27,186 63 Clinical Officers 4,804 16 15,960 37 Public Health Officers 1,216 3.6 9,039 21 Bsc – Bachelor of Science degree Source: Kenya National Bureau of Statistics, 2004 and 2015. 2.6 Malaria Control in Kenya 2.6.1 Epidemiology of malaria in Kenya There are four main malaria epidemiological zones, with diversity in risk determined largely by altitude, rainfall patterns, and temperature. These four transmission zones are malaria endemic, epidemic-prone, seasonal transmission, and low transmission (Figure 2.3). Based on population-adjusted estimates of Plasmodium falciparum prevalence (PfPR) among children ages 2 to 10 years (PfPR2-10), 29 percent of Kenya’s 2015 population live in low- risk and malaria-free areas, 22 percent live in areas of seasonal malaria risk, 20 percent live in the highland epidemic-prone areas, and 29 percent live in malaria-endemic areas. 12 Figure 2.3: Malaria epidemiological zones in Kenya, 2015 Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 1. Endemic zone: The zone includes areas of stable malaria that have altitudes ranging from 0 to 1,300 meters around Lake Victoria in western Kenya and in the coast regions. The map above (Figure 2.3) shows the lake and coast regions as two different shades of dark brown; these two regions together comprise the malaria-endemic zone. Rainfall, temperature, and humidity are the determinants of the perennial transmission of malaria. Transmission is intense throughout the year, with annual entomological inoculation rates ranging between 30 and 100 (NMCP et al., 2016). In 2015, malaria parasite prevalence in children ages 6 to 59 months was 5.3 percent in the coast region and 16.6 percent in the lake Victoria region (NMCP et al., 2016). 2. Highland epidemic-prone zone: These areas border the endemic zone in the western part of the country. Malaria transmission in the western highlands is seasonal with considerable year-to-year variation. The entire population is vulnerable and case-fatality rates during an epidemic can be greater than in endemic regions. Malaria parasite prevalence in children ages 6 to 59 months was 2.6 percent in the epidemic-prone zone (NMCP et al., 2016). 3. Seasonal-risk zone: These are arid and semi-arid areas of northern and south￾eastern parts of Kenya that may experience short periods of intense malaria transmission during the rainfall season. Temperatures are usually high, and water pools created during the rainy season provide the malaria vector breeding sites. Malaria transmission is seasonal with considerable year-to-year variation. The whole population is vulnerable, and case-fatality rates during an epidemic can be up to 10 times greater than in the endemic region. Malaria parasite prevalence in children ages 6 to 59 months was 0.5 percent in the seasonal malaria risk zone (NMCP et al., 2016). 13 4. Low-risk zone: These areas include the central highlands including Nairobi. Mean daily temperatures are usually too low to allow completion of the sporogonic cycle of the malaria parasite in the vector. However, increasing temperatures and changes in the hydrological cycle associated with climate change are likely to increase the areas suitable for malaria vector breeding, with the possible introduction of malaria transmission in areas where it has never existed. Malaria parasite prevalence in children ages 6 to 59 months was 0.4 percent (NMCP et al., 2016). 2.6.1.1 Plasmodium species Plasmodium falciparum is the main cause of malaria in Kenya, accounting for 96 percent of infections, 83 percent are pure infections and 13 percent are mixed infections with P. malariae or P. ovale. Four percent are due to P. malariae. Parasitemia surveys have not detected P. vivax since 2007. 2.6.1.2 Vector species The major malaria vectors in Kenya are Anopheles (An.) gambiae s.s., An. arabiensis, An. merus, and An. funestus. An. merus has a distribution largely within a 25-kilometer inland extent from the Kenya coast and is an important secondary vector within its range. The vectors are abundant during and immediately after the rainy season when larval habitats are abundant. An. arabiensis and An. gambiae s.s. appear to be sympatric in their distribution; however, there is evidence that An. arabiensis has begun to displace An. gambiae s.s. as the more dominant vector where both coincide (Bayoh et al., 2010). Molecular characterisation of the members of the An. funestus complex in Kenya has only been possible in the last decade (Kamau et al., 2002); where An. funestus has been previously reported might have been predominantly An. funestus s.s (Kamau et al., 2003, Kawada et al., 2012). The distribution of the predominant vectors, based on where entomologic monitoring occurred, is shown in Figures 2.4a– 2.4d. Although some malaria vectors have recently been found to bite outdoors, over 90 percent of bites still occur indoors thus LLINs and IRS are still appropriate primary tools for malaria vector control (Bayoh et al., 2014). 14 Figure 2.4: Distribution of the dominant malaria vector species based on entomological monitoring sites in Kenya, 1900–2014 Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 2.6.1.3 Malaria disease burden In 2014, malaria accounted for approximately 16 percent of outpatient attendance nationally (Ministry of Health, 2014a), and 15 to 20 percent of pediatric hospital admissions in malaria-endemic areas. Prior to 2007, malaria was the leading cause of morbidity and mortality. Malaria accounted for 30 to 50 percent of outpatient admissions, 20 percent of all hospital admissions, 3 to 5 percent of all deaths, and 20 percent of all childhood deaths (WHO, 2007b). Data from the Kilifi HDSS site, in the coastal- endemic zone, showed that malaria slide-positive acute-illness admissions accounted for 56 percent of all acute-illness admissions in children <14 years of age in 1998, declining to 7 percent in 2008, but rising to 24 percent in 2014 (Mogeni et al., 2016). Data from the Kilifi HDSS (PfPR2-10 4–8 percent) and Siaya HDSS (PfPR2-10 27–38 percent) both show a malaria-specific mortality in children under age 5 years ranging from 2.5 to 7 per 1,000 person years (Mogeni et al., 2016). The national malaria prevalence among children 6 months to 14 years of age was 8 percent in 2015, which was a decrease from 11 percent in 2010 (NMCP et al., 2016). Prevalence was highest among children ages 10 to 14 years at 11 percent compared with 15 5 percent in children ages 6 to 59 months (NMCP et al., 2016). The lake-endemic zone had the highest prevalence nationwide at 27 percent, which was a decrease from 38 percent in 2010 (NMCP et al., 2016). The malaria maps in Figure 2.5 show the change in malaria parasite prevalence in children ages 2 to 10 years across Kenya from 2000 to 2015 based on Baysian geostatistical modelling estimates that predicted PfPR2-10 at a 1x1 km resolution (Ministry of Health, 2016). The data that underlie the maps were assembled from approximately 2,700 cross-sectional community surveys conducted from 1975 to 2015. The surveys were from over 2,000 geographically unique sites and included approximately 400,000 individuals (Ministry of Health, 2016). Eighty percent of the surveys were from rural areas and about half were from school health surveillance (Ministry of Health, 2016). Figure 2.5: Population-adjusted, modelled estimates of Plasmodium falciparum parasite prevalence in children ages 2–10 years (PfPR2-10) in Kenya from 2000–2015 a) 2000 b) 2005 c) 2010 d) 2015 Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence 16 to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 17 2.6.2 Malaria control strategy Kenya has a long and storied history of malaria prevention and control. Malaria control strategies were implemented from the British colonial era at the beginning of the twentieth century through the WHO Global Malaria Eradication Program era in the 1960s through the 1980s that coincided with independence and into the present era, which started in the early twenty-first century. The strategies have included prophylaxis with quinine, pyrimethamine, and chloroquine, treatment with chloroquine or quinine and vector control with mosquito nets, IRS, and larviciding in selected areas. In 1992, the Ministry of Health developed the first comprehensive National Malaria Plan of Action for 1992–1997 (Ministry of Health, 1992). The 1997–1998 El Niño Southern Oscillation Cycle was followed by malaria epidemics in the western epidemic-prone highlands that resulted in high mortality rates. In 1998, following the epidemics and recognition of the poor outcomes partly due to chloroquine resistance, SP replaced chloroquine as first-line treatment for malaria and IRS was adopted as a strategy for malaria epidemic prevention and response (Ministry of Health, 1998). In addition, IPTp with SP was adopted as a strategy replacing chloroquine chemoprophylaxis in pregnancy (Ministry of Health, 1998). The first National Malaria Strategy (NMS) 2001–2010 was developed in 2001 (Ministry of Health, 2001b). The control strategies were case management with SP, IPTp, implementation of the first national ITN program covering pregnant women and children under 5 years of age, and IRS for epidemic prevention and control in the highland epidemic-prone areas. The plan also provided a series of indicator targets congruent with the RBM targets established in Abuja, Nigeria in April 2000. These targets were as follows: 60 percent of vulnerable children and pregnant mothers sleep under an ITN; 60 percent of children with a fever in the last two weeks access effective treatment within 24–48 hours of symptom onset; and 60 percent of pregnant women access IPTp with SP during the second and third trimesters by 2006. The second NMS 2009–2017 was launched in 2009 and revised in 2014, extending its implementation period and targets to include 2018 (MOPHS, 2009a; Ministry of Health, 2014a). The focus of the second strategy was scaling up to universal coverage with vector control and other preventive methods for all persons at risk and universal access to malaria diagnosis and treatment. Additional implementation milestones, including the launch of malaria control policies and guidelines, are shown in Table 2.4. 18 Table 2.4: Milestones in malaria policies and intervention implementation in Kenya, 1981– 2015 Year Milestones 1981 Kenya Anti-Malaria Strategy launched by the Division of Vector Borne Diseases 1980s Increasing chloroquine resistance and no effective sustained vector control method result in the rebound of malaria. 1992 World Bank re-emphasizes significance of malaria control for economic and social development. The Global Malaria Control Strategy adopted at the WHO Ministerial Conference. Ministry of Health develops the National Malaria Plan of Action 1992–1997. 1994 The Malaria Control Unit (MCU) created in the DVBD as the operational National Malaria Control Programme. 1998 Malaria epidemics occur in the western highlands after the 1997–1998 El Niño Southern Oscillation Cycle. First National Guidelines for Diagnosis, Treatment and Prevention of Malaria launched. Sulfadoxine-pyrimethamine (SP) replaces chloroquine as first-line treatment for malaria. Indoor residual spraying (IRS) adopted as a strategy for malaria epidemic prevention and response. Intermittent preventive treatment in pregnancy (IPTp) with SP adopted. Demographic and Health Survey conducted. 1999 First Health Sector Strategic Plan (HSSP) 1999–2004, which recognizes malaria as the highest priority public health issue. Guidelines for Malaria Epidemic Preparedness and Control in Kenya launched. 2000 The MCU in DVBD is elevated to a division and renamed the Division of Malaria Control (DOMC) within the Ministry of Health. 2001 The first Kenya National Malaria Strategy (NMS) 2001–2010 launched. Insecticide-Treated Nets (ITN) Strategy: 2001–2006 launched. Subsidized ITNs distributed to pregnant women and children under 5 at maternal and child health clinics (MCH) under a cost-sharing mechanism. 2003 Demographic and Health Survey conducted. 2004 Artemisinin-based combination therapy (ACT) adopted as first-line treatment for malaria. Amodiaquine adopted as interim first-line treatment before ACTs become available in the public sector. ITNs distributed free to pregnant women and children under 5 years at MCH clinics. Distribution of socially-marketed, subsidized nets to community members launched. 2005 Conventional ITNs, which need retreatment with insecticide at intervals, replaced by long-lasting insecticidal bed nets (LLINs). 2006 National Guidelines for Diagnosis, Treatment and Prevention of Malaria–2nd Edition launched. ACTs become available in the public sector at no cost to patients. Malaria rapid diagnostic test (RDTs) become available in the public sector in low￾transmission zones. First free mass net distribution campaign targeting children under 5 years and pregnant women conducted. Integrated Disease Surveillance and Response launched. 2007 First Malaria Indicator Survey conducted. 19 Year Milestones 2008 Post-election violence results in population displacements with movement between malaria risk zones and disruption of health services. Pharmacy and Poisons Board bans importation and sale of non-recommended antimalarials, including oral artemisinin-based monotherapies. 2009 Demographic and Health Survey conducted . Malaria program performance review conducted. Kenya National Malaria Strategy 2009–2017 launched. Kenya Malaria Monitoring and Evaluation Plan 2009–2017 launched. IRS adopted as a strategy for malaria control in endemic areas. IPTp with SP implementation restricted to malaria-endemic areas. 2010 Malaria Indicator Survey conducted. Integrated Vector Management Policy Guidelines for Kenya 2010 launched. National Malaria Policy Document 2010 launched. Malaria Communication Strategy 2010–2013 launched. National Guidelines for Diagnosis, Treatment and Prevention of Malaria –3rd Edition, which places emphasis on parasitological diagnosis of malaria before treatment. Subsidized ACTs introduced in the private sector through the Affordable Medicine Facility– malaria. 2011 Demographic and Health Survey conducted . Guidelines for Malaria Epidemic Preparedness and Response revised. Free universal mass net distribution campaign conducted in Nyanza, Western, and Rift Valley provinces. District Health Information Software 2 (DHIS2) launched nationally. 2012 Malaria RDTs become available in all public health facilities nationally. Free universal mass net distribution campaign conducted in Coast and Rift Valley provinces. 2013 Devolution of health services to county level and creation of County Health Management Teams to manage health service delivery. Mid-term review of the Kenya NMS 2009–2017 conducted. National Guidelines for Diagnosis, Treatment and Prevention of Malaria –4th Edition; artesunate replaced quinine as first-line treatment for severe malaria. 2014 Demographic and Health Survey conducted . Free universal mass net distribution campaign conducted in six counties in western Kenya. Revised Kenya Malaria Strategy 2009–2018 and Kenya Malaria Monitoring and Evaluation Plan 2009–2018 launched. 2015 Malaria Indicator Survey conducted. Free universal mass net distribution campaign conducted in 17 counties in coastal and western Kenya. Table 2.4: Milestones in malaria policies and intervention implementation in Kenya, 1981– 2015 2.6.3 Financing of malaria control interventions Total health expenditures for malaria control came mainly from three sources including the Government of Kenya (GOK), external donors and partners, and households. GOK financing for malaria activities includes procurement of medicines for severe malaria, diagnostics, and insecticides as well as funding for program management activities at the national level. The country’s health budget as a 20 proportion of GDP remained fairly constant at about 5 percent between 2003 and 2014 (WHO, 2014). However, the per capita expenditure on health, including external donor resources, began a steady increase in 2004 and together with per capita government expenditure, increased significantly from 2010 (WHO, 2014). The increase in health expenditures coincided with increased funding for infrastructure and human resources in the health budget and external funding for universal coverage with LLINs. Kenya’s major sources of external funding for malaria include the Global Fund to fight AIDS, Tuberculosis and Malaria (Global Fund), United States President’s Malaria Initiative (PMI), United Kingdom Department for International Development (DFID), World Bank, and other multilateral agencies and NGOs. These external resources have been the backbone of the fight against malaria in Kenya. While each source has had different financial planning and disbursement timelines, disbursements and procurements for commodities have been accounted for in the calendar year in which they were received. As shown in Figure 2.6, donor financing accounts for over 90 percent of malaria commodities and interventions. However, government expenditures on health service delivery are not easily quantified and thus not included in the figure. About half of malaria expenditures in 2012 and in 2014 was spent on vector control interventions (i.e., LLINs and IRS), about 15 percent was spent on antimalarial treatments, and 10 percent was spent on diagnostics (Figure 2.7). The data in Figure 2.7 is presented as an example; funding could be reliably allocated to specific intervention strategies during 2012 (i.e., a year without funding for a universal coverage mass LLIN distribution) and 2014 (i.e., a year with a universal coverage mass LLIN distribution). Figure 2.6: Sources of malaria financing in Kenya, 2003–2014 Note: GOK – Government of Kenya; PMI – U.S. President’s Malaria Initiative; DFID – Department for International Development; UNICEF – United Nations Children’s Fund; MACEPA – Malaria Control and Elimination Partnership in Africa Source: Adapted from World Health Organization. World Malaria Report 2015. Geneva: World Health Organization, 2016. 21 Figure 2.7: Malaria expenditure by intervention in 2012 and 2014 in Kenya Note: LLIN – long-lasting insecticidal net Source: 2012 chart adapted from Malaria Control Unit, Annual Malaria Report 2012. Nairobi, Kenya: Ministry of Public Health and Sanitation, 2013; 2014 chart adapted from World Health Organization. World Malaria Report 2015. Geneva: World Health Organization, 2016. 22 3 Malaria Control Interventions 3.1 Insecticide-Treated Nets 3.1.1 Background Evidence from clinical trials, including those undertaken in Kenya (Nevill et al., 1996, Phillips-Howard et al., 2003), has shown that ITNs provided significant protective efficacy against malaria infection, over 50 percent reduced risk of clinical disease, and a nearly 20 percent reduction in all-cause childhood mortality (Lengeler, 2004). To achieve the RBM milestones, protection of the population in malaria-endemic countries with ITNs was therefore considered the main preventive tool to reduce the burden of the disease (RBM and WHO, 2000, Roll Back Malaria, 2005). The major ITN issues debated at the time, however, were the appropriate target population, sources of funding, and the most effective and efficient way to scale up (Webster et al., 2005). Initial consensus in the malaria research and control community was around prioritizing children under age 5 years and pregnant women (Roll Back Malaria, 2005). The decision was based on the assumption that the majority of the sub-Saharan African population was exposed to stable malaria transmission, and these population sub￾groups remained vulnerable because of lack of immunity due to age-limited exposure or compromised immunity due to pregnancy. The target set by the RBM partnership for 60 percent coverage of essential prevention and disease management interventions by 2006 was readily adopted in national malaria control strategies in almost all sub￾Saharan Africa countries, including Kenya. 3.1.2 ITN policy in Kenya The Kenya National Malaria Strategy (NMS) 2001–2010, launched in April 2001, aimed to ensure that 60 percent of at-risk children and pregnant women slept under an ITN by 2006 (Ministry of Health, 2001b). However, insufficient funding and the need to re-treat conventional nets every 6 months with insecticide affected achievement of targets. The NMS 2001–2010 did not outline any specific delivery approach but envisaged the targeting of children and pregnant women with subsidized ITNs. In 2000, the Ministry of Health and partners developed an ITN strategy paper (Ministry of Health, 2001a) in which various approaches to expand ITN coverage were outlined to reach a target of 60 percent of populations at risk by 2005. At this time, the main strategies for scaling up ITNs in Africa were through social marketing via the retail sector combined with limited distribution of nets at a subsidized cost through MCH clinics (Noor et al., 2007). By 2007, however, evidence began to emerge from Kenya that free mass distribution was the most effective and equitable mechanism to achieve the scale up of ITNs (Noor et al., 2007) and had a significant impact on reducing mortality among children under age 5 years (Fegan et al., 2007). This evidence was followed shortly by the WHO recommendation for universal coverage of ITNs by free mass distribution as the main method for scale-up supplemented with routine distribution and other mechanisms (WHO, 2007a). The Kenya NMS 2009–2017, which replaced the NMS 2001–2010, reflected the WHO recommendation for universal coverage and targeted 80 percent population coverage 23 of LLINs among those at risk (MOPHS, 2009). Free mass distribution was the main mechanism for scale-up combined with free routine distribution through MCH clinics and limited social marketing in targeted areas (MOPHS, 2009). Following an empirical mapping of malaria transmission in the country (Noor et al., 2009), 23 counties were targeted for subsequent free mass distribution campaigns in the lake- and coastal￾endemic and highland epidemic-prone zones as well as specific sub-counties with focal malaria transmission (e.g., irrigation areas). 3.1.3 ITN implementation 3.1.3.1 Distribution strategy Since 2001, several mechanisms for ITN distribution to populations at risk were implemented in Kenya (Noor et al., 2007, Snow et al., 2010, Noor et al., 2010). Initially, ITNs were accessed mainly from the private-for-profit retail sector and via limited focal distributions by research projects or NGOs (Snow et al., 2010, Shretta, 1999). Distribution of highly subsidized ITNs through MCH clinics and by social marketing via the retail sector and community-based organizations started in 2004. From May 2005 forward, LLINs largely replaced conventional ITNs (Noor et al., 2007). Kenya implemented the first free mass LLIN distribution campaign in 2006, with over 3 million LLINs distributed to children under age 5 years and pregnant women (Noor et al., 2007). In 2008, a national campaign to re-treat conventional nets with KO-TAB 1-2-3 and replace torn or damaged nets was undertaken in 55 districts (Snow et al., 2009). A total of 1.93 million nets were re-treated, and 207,290 torn nets were replaced. The first universal coverage (i.e., one LLIN per two people per household) mass LLIN distribution occurred in three phases from 2011–2012 and the second in four phases from 2014–2015 in 23 counties and parts of three counties (i.e., areas with focal malaria transmission due to irrigation). Concurrently, free routine LLIN distribution via MCH clinics continued in 36 counties and social marketing of LLINs through community-based organizations continued in the 14 malaria-endemic counties. Table 3.1 summarizes ITN and LLIN distributions in Kenya since 2005 when major expansion efforts began and Figure 3.1 illustrates the mechanisms of net distribution since 2004. Table 3.1: Nets distributed in Kenya, 2005–2015 Year of Distribution Type of net distribution mechanism Number distributed Funders 2005–2015 Routine LLIN & ITNs 23,343,797 DFID, PMI, Global Fund 2006 2011 2012 2013 Free mass campaign Free mass campaign Free mass campaign Free mass campaign LLIN LLIN LLIN LLIN 3,099,473 8,321,603 2,790,749 17,325 Global Fund Global Fund, PMI, DFID, World Vision Global Fund, DFID, World Bank World Bank 24 2014 Free mass campaign LLIN 3,286,767 Global Fund, PMI 2015 Free mass campaign LLIN 9,367,569 Global Fund, PMI Total 50,227,283 Note: LLIN – long-lasting insecticidal net; ITN – insecticide-treated net; DFID – Department for International Development; PMI– U.S. President’s Malaria Initiative Source: National Malaria Control Program data 25 Figure 3.1: Number of nets distributed by channel in Kenya, 2004–2015 Note: ITN – insecticide-treated net Source: National Malaria Control Program data 3.1.3.2 Insecticide resistance Insecticide resistance related to ITNs was first reported in Western Kenya and has continued to increase as pyrethroid-based LLINs and indoor residual spraying (IRS) using pyrethroids became the mainstays of vector control (Ochomo et al., 2014). High levels of pyrethroid resistance, as well as widespread resistance, have been observed in An. arabiensis and An. gambiae s.s. Holes from use have been shown to permit mosquito entry and feeding, thus providing little protection against these vectors. All An. gambiae s.s samples collected resting in LLINs with holes have been found to be homozygous for the knock down resistance kdr genotype L1014S (Ochomo et al., 2013). An. funestus populations in Western Kenya and Uganda have also been found to have extensive resistance to both pyrethroids and DDT, thus presenting challenges for the future control of these vectors (Mulamba et al., 2014). In contrast, An. arabiensis populations from an area of low malaria transmission around a rice irrigation scheme in Central Kenya showed complete susceptibility to all insecticide groups (Kamau and Vulule, 2006); while at the coast, An. gambiae s.l. collected from in Kilifi, Malindi, and Taveta showed different levels of resistance to deltamethrin, lambdacyhalothrin, and bendiocarb(NMCP, 2016). Despite increasing resistance to pyrethroids by the main malaria vectors, studies conducted in western Kenya from 2011 to 2014 did not find any evidence of association between insecticide resistance and malaria infection in children under 5 years of age (NMCP, 2016). 3.1.4 ITN coverage trends 3.1.4.1 Household ITN ownership In 2003, just before the start of major ITN programmatic expansion efforts in Kenya, 8 percent of households in the country owned at least one ITN (Figure 3.2). In 2007, following the first free mass campaign in 2006, ownership of at least one ITN had increased to nearly half of all households. By 2015, household ITN ownership 26 100 Percentage of households with at least one ITN 90 80 70 60 50 40 30 20 10 0 5 38 42 33 42 49 6 43 57 46 52 52 8 46 58 50 75 73 13 55 66 58 70 73 12 57 73 59 81 87 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic nationally was 63 percent, although not all counties or epidemiologic zones are targeted for net distribution. Figure 3.2: Overall household ownership of at least one ITN in Kenya, 2003–2015 100 80 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Percentage of households with at least one ITN 8 48 56 48 59 63 0 20 40 60 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) In the lake- and coastal-endemic counties, household ITN ownership increased from 12 percent and 13 percent in 2003 to 87 percent and 73 percent in 2015, respectively (Figure 3.3). The malaria-endemic counties benefit from three concurrent net distribution channels: universal coverage mass distribution every 3 years, routine distribution to pregnant women and infants via MCH clinics, and social-marketing to rural households via NGOs and community-based organizations. Figure 3.3: Household ownership of at least one insecticide-treated net by endemicity zone in Kenya, 2003–2015 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) The gap between urban and rural ownership of at least one ITN also decreased, and by 2015, both residences had similar rates, with ownership slightly above 60 percent 27 100 Percentage of households with at least one ITN 80 60 40 20 0 3 35 49 49 51 49 4 50 58 43 61 63 6 44 60 43 64 68 8 55 55 44 57 62 15 51 56 50 61 68 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Most poor Second Middle Fourth Least poor 100 Percentage of households with at least one ITN 80 63 62 58 61 55 50 56 60 48 KMIS 2010 KDHS 2014 KMIS 2015 46 48 40 20 14 6 0 KDHS 2003 KMIS 2007 KDHS 2008 Rural Urban (Figure 3.4). In contrast, by 2015, the households in the poorest quintile owned nearly 20 percent fewer ITNs than those in the highest wealth quintiles (Figure 3.5). Figure 3.4: Household ownership of at least one insecticide-treated net by residence in Kenya, 2003–2015 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.5: Household ownership of at least one insecticide-treated net by wealth quintile in Kenya, 2003–2015 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.6 shows the percentage of households that have attained universal coverage of ITNs from 2003 to 2015. As with ITN ownership, universal coverage of ITNs at the household level increased significantly from 4 percent in 2003 to 40 percent in 2015. In 28 5 28 35 31 43 48 6 39 47 40 54 56 3 22 28 26 37 44 0 10 20 30 40 50 60 70 80 90 100 Percentage who slept under ITN the night before survey KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Total population Under 5 Pregnant women 100 Percentage of households with one ITN for every 2 people 4 18 27 20 35 0 20 40 80 60 40 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 the lake- and coastal-endemic zones, universal coverage increased from 6 to 54 percent and 7 to 46 percent, respectively, from 2003 to 2015. Figure 3.6: Percentage of households with one insecticide-treated net per two persons in Kenya, 2003–2015 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 3.1.4.2 ITN use by population, children under age 5, and pregnant women Among all households Use of ITNs, measured as the percentage of all household members sleeping under an ITN the night before the survey, was 5 percent among all ages in 2003, rising to 48 percent in 2015. Use of ITNs among children under 5 years increased from 6 percent in 2003 to 40 percent in 2010 and to 56 percent in 2015. Similarly, use among pregnant women increased from 3 percent in 2003 to 44 percent in 2015 (Figure 3.7). Figure 3.7: Insecticide-treated net use among general population, children under 5 years of age, and pregnant women among all households in Kenya, 2003–2015 Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 29 100 82 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Total Population Under 5 Pregnant Women Percentage who slept under an ITN the night before survey in households with at least one ITN 59 54 58 60 67 71 72 66 69 69 77 79 70 76 71 77 0 20 40 60 80 Among households with at least one ITN In households with at least one ITN, ITN use by all household members was much higher, increasing from 59 percent in 2003 to 71 percent in 2015. Among children under 5 years in households with at least one ITN, use of ITNs increased from 72 percent in 2003 to 79 percent in 2015. Due to overall low ITN coverage, there were insufficient numbers of women using ITNs recorded in the KDHS of 2003 to allow analysis. However, in the KMIS 2007, approximately 70 percent of pregnant women in households with at least one ITN slept under a net the night before the survey, which increased to 82 percent in 2015 (Figure 3.8). Use of ITNs in the lake-endemic zone increased from 53 percent to 76 percent in households with at least one ITN (Figure 3.9). Figure 3.8: Insecticide-treated net use among general population, children under 5 years of age, and pregnant women in households with at least one ITN, 2003–2015 Note: ITN – insecticide-treated net; KDHS 2003 had only 48 pregnant women in households with ITNs, which was too few to analyse. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.9: Insecticide-treated net use by household members in households with at least one ITN by malaria endemicity, 2003–2015 30 70 54 64 59 69 68 66 59 55 59 58 64 48 46 51 60 64 70 72 62 64 67 75 74 53 54 59 57 70 76 0 10 20 30 40 50 60 70 80 90 100 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Percentage who slept under an ITN the night before survey in households with at least one ITN Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic Note: ITN – insecticide-treated net Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 3.1.4.3 Equity in ITN use Use of ITNs among children under 5 years of age in all households The gaps in ITN use by location of residence and wealth quintile were reduced between 2003 and 2015. There were no disparities in ITN use between children under 5 years in urban and rural areas by 2015. In 2003, 10 times more children under 5 years of age used ITNs in the highest compared to the lowest wealth quintile; by 2015, the ITN-use gap was reduced to 1.7 times between the highest and lowest wealth quintiles (Table 3.2). This suggests that the current disparity in ITN use by wealth quintile is largely due to disparities in access to ITNs (Table 3.2). Table 3.2: Use of Insecticide-treated nets among children under 5 years by background characteristic in Kenya, 2003–2015 KDHS 2003 KMIS 2010 KMIS 2015 Percentage point change 2003–2015 Characteristic % (95%CI) N % (95%CI) N % (95%CI) N (95% CI) p-value* Sex Male 6.5 (5.5-7.8) 2,873 40.8 (38.3-43.3) 2,370 55.4 (50.3-60.4) 2,038 48.9 (46.5-51.2) <0.0001 Female 5.5 (4.3-6.8) 2,844 39.8 (37.4-42.3) 2,364 56.8 (52.1-61.3) 1,998 51.3 (49.0-53.6) <0.0001 Residence Urban 12.1 (9.5-15.3) 1,430 45.4 (39.1-51.9) 553 59.8 (51.4-67.7) 1,562 47.7 (44.7-50.7) <0.0001 Rural 4.7 (3.8-5.9) 4,287 39.4 (37.7-41.1) 4,181 54.4 (48.9-59.8) 2,474 49.7 (47.7-51.8) <0.0001 Wealth Quintiles Lowest 1.5 (0.8-2.9) 1,426 39.3 (36.9-41.8) 2,040 40.0 (32.0-48.4) 1,290 38.5 (35.8-41.3) <0.0001 Second 3.0 (1.8-4.9) 1,089 39.4 (32.4-46.8) 232 56.6 (50.5-62.5) 867 53.6 (50.1-57.1) <0.0001 Middle 6.2 (4.4-8.5) 1,062 37.8 (34.9-40.8) 1,348 60.6 (54.4-66.5) 682 54.3 (50.4-58.3) <0.0001 Fourth 7.4 (5.4-10.1) 946 43.5 (38.3-48.9) 539 63.2 (55.0-70.7) 606 55.8 (51.6-60.0) <0.0001 Highest 14.6 (11.6-18.1) 1,194 46.1 (40.2-52.1) 565 66.6 (56.0-75.8) 591 52.1 (47.8-56.4) <0.0001 Total 6.0 (5.1-7.1) 5,711 40.3 (38.6-42.0) 4,734 56.1 (51.6-60.5) 4,036 50.1 (48.4-51.7) <0.0001 Note: N=weighted number of children (denominator); An insecticide-treated net (ITN) is (1) a factory-treated net that does not require any further treatment (i.e., long-lasting insecticidal net or LLIN), or (2) a pre-treated net obtained within 31 the past 12 months, or (3) a net that has been soaked with insecticide within the past 12 months. *Baseline for the chi-square test is 2003 compared with results of KMIS 2015; data presented for baseline 2003, midline 2010 and endline 2015. CI – confidence interval Source: Kenya Demographic and Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 32 Table 3.3: Use of Insecticide-treated nets among pregnant women by background characteristic in Kenya, 2003–2015 Characteristic KDHS 2003 KMIS 2010 KMIS 2015 Percentage point change 2003–2015 (95% CI) p-value* % (95%CI) N % (95%CI) N % (95%CI) N Residence Urban Rural 6.6 (3.7-11.3) 5.1 (3.4-7.6) 174 463 38.1 (14.6-68.9) 39.3 (33.0-46.0) 47 335 60.0 (49.1-70.0) 56.3 (47.4-64.9) 164 204 53.4 (45.1-61.8) 51.2 (44.1-58.3) <0.0001 <0.0001 Wealth Quintiles Lowest Second Middle Fourth Highest 1.7 (0.4-6.9) 4.0 (1.6-9.5) 6.6 (2.8-14.6) 7.4 (3.7-14.1) 7.6 (4.4-12.8) 143 121 107 109 157 43.3 (32.5-54.8) 31.9 (15.2-55) 38.2 (28.5-49.0) 46.8 (32.3-61.8) 29.3 (13.6-52.2) 140 22 113 50 57 35.0(23.5-48.6) 63.7(50.8-74.8) 71.5 (56.0-83.2) 52.4 (37.7-66.8) 67.0 (51.9-79.3) 111 71 56 52 78 33.7 (24.6-42.8) 59.2 (47.5-71.0) 64.9 (52.2-77.6) 44.6 (30.1-59.0) 59.0 (47.8-70.3) <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 Total 5.4 (3.9-7.5) 637 39.1 (32.1-46.7) 382 57.8 (51.0-64.2) 368 52.5 (47.2-57.9) <0.0001 Note: N=weighted number of children (denominator); An insecticide -treated net (ITN) is (1) a factory -treated net that does not require any further treatment (i.e., long -lasting insecticidal net or LLIN), or (2) a pre-treated net obtained within the past 12 months, or (3) a net that has been soaked with insecticide within the past 12 months. *Baseline for the chi-square test is 2003 compared with results of KMIS 2015; data presented for baseline 2003, midline 2010 and endline 2015. CI – confidence interval Source: Kenya Demographic and Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Use of ITNs among pregnant women in all households There were no differences in ITN use among pregnant women based on place of residence. In 2003, 4.5 times more pregnant women used ITNs in the highest compared to the lowest wealth quintile; by 2015, the ITN-use gap was reduced to 1.9 times between the highest and lowest wealth quintiles (Table 3.3). 3.1.5 ITN summary • Between 2003 and 2015, over 50 million free ITNs were distributed. • Both overall ITN ownership, from 8 to 63 percent, and use, from 5 to 48 percent, increased significantly from 2003 to 2015. • In the targeted lake- and coastal-endemic zones, with three concurrent ITN distribution channels, ownership increased from 12 to 87 percent and 13 to 73 percent, respectively, from 2003 to 2015. • In households with at least one ITN, use among children under 5 years and pregnant women increased from 66 to 79 percent and 70 to 82 percent, respectively, from 2007 to 2015. • Overall, universal coverage (i.e., at least one ITN per two person per household) or access increased from 4 to 40 percent from 2003 to 2015. • In the targeted lake- and coastal-endemic zones, with three concurrent ITN distribution channels, universal coverage or access increased from 6 to 54 percent and 7 to 46, respectively, from 2003 to 2015. • Although disparities by residence and socioeconomic status have been substantially reduced, the poorest quintile of the population continues to have lower ITN ownership and reported use compared to those in the wealthiest quintile of the population. • Widespread resistance to pyrethroids and DDT has been documented in the main malaria vectors in western Kenya; however, evidence shows that LLINs remain effective as a malaria prevention intervention. 33 3.2 Indoor Residual Spraying 3.2.1 Background In Kenya, IRS use started as early as 1944 in Kericho, located in the highland epidemic-prone zone, using DDT. Through the 1950s, IRS activities continued, focusing on farming and irrigation areas across the country. By the late 1970s, IRS activities had all but ceased, with malaria treated presumptively through the primary health care system as a febrile disease. The suspension of vector control activities and the gradual increase in chloroquine resistance led to a resurgence in malaria in the endemic areas in 1980s and 1990s. The 1997–1998 El Niño resulted in massive malaria epidemics across the country. In response to these epidemics, the government reintroduced IRS in 2000, focusing on highland epidemic-prone districts and using pyrethroids, and from 2005 to 2009, IRS was the main vector control strategy for epidemic prevention and response in the highland epidemic-prone zone. From 2010, the country adopted IRS as a vector control strategy for reduction of disease burden and implemented the strategy in three high-burden districts in the lake-endemic zone through 2012. 3.2.2 IRS policy In 2005, the GOK introduced a coordinated approach for implementing IRS to prevent epidemics in the highland epidemic-rone zone. In the 2009 Integrated Vector Management Policy Guidelines, Kenya set a series of regulatory and implementation guidelines for IRS, including a recommendation to use IRS as a complementary vector control intervention in endemic areas with high stable transmission of malaria (MOPHS 2009b). The Pest Control Products Board was identified as the lead organization responsible for the registration of insecticides for public health use, including IRS, using “sufficient evidence of efficacy, cost-effectiveness, safety and conformity to WHO specifications” (add reference here). Since the start of contemporary IRS efforts in Kenya in 2000, pyrethroids have been the main insecticide used for IRS. With the emergence of high levels of pyrethroid resistance demonstrated in western Kenya, the NMCP and partners developed an IRS business plan to guide IRS implementation between 2015 and 2018 (Ministry of Health, 2015a) andan insecticide resistance management strategy in line with the WHO strategy for mitigation of insecticide resistance (Ministry of Health, 2015b). 3.2.3 IRS implementation Since 2002, IRS has been largely focused in highland epidemic-prone counties with a goal of 85 percent coverage of targeted households (Ministry of Health, 2015a). The IRS program was implemented with funding mainly from the GOK, Global Fund, DFID, PMI, and WHO. The GOK and WHO supported the intervention from 2002 to 2005; the GOK, Global Fund and DFID supported it from 2005 to 2008; and the GOK and PMI provided support from 2008 to 2012. Lambdacyhalothrin, deltamethrin, and alpha￾cypermethrin (i.e., all pyrethroid insecticides) have been used in a single yearly spray cycle implemented just before the malaria high-transmission season, which generally starts in June. In 2010, IRS began in districts in the malaria endemic counties of Migori and Homa Bay in an effort to rapidly reduce transmission. From 2005 to 2010, IRS coverage in target areas increased from less than 100,000house units covering about 300,000 people to 1.6 million 34 Year Target housing units Housing units sprayed % Housing unit Population protected coverage Year 2005 680,000 93,000 13.7 279,000 housing units covering 4.8 million people or almost 90 percent of the target population in the districts as shown in Table 3.4. Due to global concerns around increasing resistance to insecticides, the NMCP began monitoring insecticide resistance in Kenya in 2009. The Kenya Medical Research Institute (KEMRI) started insecticide resistance monitoring in lake-endemic and highland epidemic-prone counties in 2009 (Kiambo Njagi, personal communication). Additional studies in the coastal-endemic and lake-endemic counties investigating the combined effectiveness of ITNs and IRS in a series of controlled trials started in 2010 (Charles Mbogo, personal communication). Figure 3.10 presents the malaria epidemiological zones covered by IRS from 2005 to 2012. The IRS program was stopped between 2013 and 2015 as the GOK reviewed and adopted strategic and regulatory frameworks for the implementation of non-pyrethroid insecticides (Ministry of Health, 2015a). Figure 3.10: Malaria epidemiological zones covered by IRS implementation in Kenya from 2005–2012 Note: While data are provided by county, indoor residual spraying did not target entire counties for either epidemic prevention and response or burden reduction. Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 3.2.4 Trends in IRS implementation areas During the implementation period, IRS was conducted in parts of highland epidemic￾prone and lake-endemic counties. The coverage rose from less than 15 percent of all targeted structures in 2005 to over 90 percent in 2012. The total population protected varied by year based on the number of target house units sprayed as shown in Table 3.4. Table 3.4: Percent of targeted households protected by IRS in the highland epidemic￾prone counties, 2005–2012 35 Year 2006 680,000 110,000 16.2 330,000 Year 2007 1,554,431 1,171,073 75.3 3,459,207 Year 2008 1,063,043 921,621 86.7 3,061,966 Year 2009 1,246,369 612,822 49.2 1,717,470 Year 2010 1,756,548 1,562,974 89.0 4,770,392 Year 2011 547,966 485,043 89.0 1,832,090 Year 2012 693,060 643,292 93.0 2,435,836 IRS – indoor residual spraying Source: Ministry of Health. The Epidemiology and Control Profile of Malaria in Kenya: Reviewing the Evidence to Guide the Future of Vector Control. Nairobi, Kenya: Ministry of Health, 2016. 3.2.5 IRS summary • From 2000 to 2009, IRS using pyrethroids was implemented for malaria epidemic prevention and response in part of 12 highland epidemic-prone counties. • From 2010 to 2012, IRS using pyrethroids was implemented in parts of three malaria-endemic counties, which bordered highland epidemic-prone counties, for burden reduction and in highland epidemic-prone counties only for epidemic response. • From 2013 to 2015, no IRS was implemented as the GOK reviewed and adopted strategic and regulatory frameworks for use of non-pyrethroid insecticides to mitigate emerging pyrethroid resistance. 3.3 Intermittent Preventive Treatment in Pregnancy 3.3.1 Background Due to the substantial risks posed to the life and well-being of the pregnant mother, foetus and neonate, malaria in pregnancy is considered an important health problem in malaria endemic countries (WHO, 2015). In addition to protection with ITNs during pregnancy, IPTp with SP beginning early in the second trimester is the other main malaria prevention approach in pregnancy. Following the results of a meta￾analysis of efficacy studies (Garner and Gülmezoglu, 2006), WHO initially recommended at least 2 doses of IPTp during pregnancy (WHO, 2007c) for all pregnant women at risk of malaria. As SP resistance increased, studies showed that at least 3 doses of IPTp conferred greater protection, and consequently, WHO revised its recommendations to give IPTp at each ANC vist after the first trimester in areas of moderate-to-high malaria transmission (WHO, 2012). 3.3.2 IPTp policy Both in the first and second NMS, IPTp was a key approach to prevent malaria in pregnant women in Kenya (MOPHS 2009a). Initially, IPTp was recommended nationally starting in 1998, but as the understanding of the heterogeneity of malaria in Kenya improved (Noor et al., 2009), the policy was changed to implementation only in areas of moderate to high transmission (MOPHS, 2009a), which included the 14 counties of the lake- and coastal-endemic zones. In 2011, the NMCP issued simplified IPTp guidance to all health facilities in the lake- and coastal-endemic zones providing focused ANC services to encourage a doses of IPTp with SP at each scheduled ANC visit after quickening, with a minimum of two doses at least 4 weeks apart, in order help meet national targets. In late 2014, Kenya revised the IPTp strategy to explicitly state that all pregnant women should receive at least three doses of IPTp with SP during ANC visits in the 14 malaria-endemic counties (Ministry of Health, 2014a). 36 However, the revised IPTp strategy was not operationalized until 2015, and only IPTp doses 1 and 2 were reported via the routine HIS through the end of 2015. 3.3.3 IPTp implementation In 2002, the focused ANC and malaria in pregnancy program was rolled out nationwide to improve coverage of IPTp with SP. Starting in 2004, the Global Fund supported scaling up IPTp integrated with reproductive health services. Subsequently, funding from the Global Fund, PMI and other sources has been used to ensure the delivery of IPTp with SP routinely in targeted endemic areas. 3.3.4 SP resistance and implications for IPTp policy Several studies showed rapidly expanding resistance levels and by 2003, SP had been designated a failed drug for the treatment of uncomplicated malaria in the Kenya (Bousema et al., 2003). However, the evidence of SP resistance on intermittent preventive treatment of malaria in pregnant women remains inconclusive (WHO, 2007c) Subsequently, the WHO still recommends SP for IPTp (WHO, 2012). 3.3.5 Trends in IPTp coverage Data on IPTp coverage was obtained from national household survey data and defined using a denominator of women ages 15–49 with a live birth in the 2 years preceding the survey. In the lake and coastal malaria-endemic zones where the intervention is targeted, coverage of IPTp2 increased from 9 and 26 percent in 2007 to 53 and 60 percent in 2015, respectively (Figure 3.12). Figure 3.11: Pregnant women who received at least two doses of intermittent preventive treatment during pregnancy nationwide in Kenya, 2003–2015 100 80 60 35 40 KDHS2003 KMIS2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Proportion of pregnant women receiving two or more doses of 5 12 15 26 18 0 20 IPTp Note: IPTp – intermittent preventive treatment in pregnancy Source: Kenya Demographic and Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.12: Pregnant women who received at least two doses of intermittent preventive treatment during their last pregnancy by endemicity in Kenya, 2003–2015 37 100 80 60 53 55 60 40 33 30 31 26 29 22 29 23 23 23 17 14 14 18 14 14 20 11 9 12 7 7 5 7 5 6 3 0 KDHS2003 KMIS2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Low risk 2 Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic Proportion of pregnant women receiving two or more doses of IPTp Note: IPTp – intermittent preventive treatment in pregnancy Source: Kenya Demographic and Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 3.3.5.1 Equity in IPTp coverage Although coverage increased across all socioeconomic classes from 2003 to 2015, coverage was marginally higher among the poorest households. Between 2003 and 2010, mothers with a secondary school or higher education were likely to receive at least two doses of IPTp2 compared to those with no education. However, by 2015, this difference had disappeared 3.3.6 IPTp summary • IPTp2 increased from 9 and 26 percent in 2007 to 53 and 60 percent in 2015 in the lake and coastal malaria endemic zones, respectively. 3.4 Malaria Case Management 3.4.1 Background Prompt diagnosis and treatment with effective drugs is the cornerstone to reducing morbidity and mortality. The WHO has recommended artemisinin-based combination therapies (ACTs) for the treatment of uncomplicated malaria across malaria-endemic countries, especially those in sub-Saharan Africa where P. falciparum is the main parasite. Under the Test, Treat, and Track policy, the WHO also recommends the confirmation of parasitemia in people with malaria symptoms when they visit health facilities and treatment with ACTs for only those who are positive. In Kenya, ACTs are available, free in the public health sector and at subsidized prices in the private sector, following the introduction of the Affordable Medicines Facility￾malaria (AMFm) in 2010 3.4.2 Case management policy Prior to 1998, the first-line drug for the treatment of uncomplicated malaria was chloroquine (CQ). Although CQ was a very effective drug for many years, more than 40 percent of malaria infections in children treated with CQ failed to clear by day 7 on the Kenyan coast by 1987 (Brandling-Bennett et al., 1988, Watkins et al., 1984, Spencer et al., 38 1984), and by 1991, widespread CQ-treated clinical failures had been reported across the country. In 1998, CQ was replaced with SP as the recommended first-line treatment, and national treatment guidelines were revised (Ministry of Health, 1998). However, by 2003, SP had failed across the country for the treatment of clinical malaria (Bousema et al., 2003). In 2004, the national policy changed from SP to an ACT, artemether-lumefantrine (AL), as the recommended first-line treatment of uncomplicated malaria Ministry of Health, 2006). In 2006, the GOK made malaria RDTs available in public-sector facilities in low-transmission zones, and in 2009, the GOK extended the case management policy to include parasitological testing, via malaria RDTs or microscopy, for all patients with suspected malaria (MOPHS, 2009a). Prior to 2009, only patients over 5 years of age required parasitological testing before treatment; fevers in children under 5 years were presumptively treated as malaria in line with the integrated management of childhood illnesses (IMCI) guidelines (Ministry of Health, 2006). In late 2012, the GOK initiated a national rollout of malaria RDTs to increase parasitological testing prior to treatment, and by 2015, 97 percent of public-sector health facilities had malaria diagnostic capacity (Machini et al., 2016). Since 2010 in focused areas of the lake-endemic counties, community case management of malaria has been ongoing with Global Fund support, and integrated community case management (iCCM) has been ongoing with support from DFID, United National Children’s Fund (UNICEF) and WHO. The NMS 2009–2017 targets for malaria case management are that 100 percent of health facilities have AL and malaria diagnostics and 100 percent of patients with fever who present to health workers should receive parasitological diagnosis of malaria and recommended treatment (MOPHS, 2009a, Ministry of Health, 2014a). 3.4.3 Implementation of antimalarial and malaria diagnostic policy Between 2006 and 2015, the main malaria case-management activities included the following: procurement and distribution of SP, AL, malaria RDTs, microscopes and microscopy supplies; development, revision, and distribution of new case￾management guidelines (MOPHS, 2010, MOPHS, 2012, Ministry of Health, 2014b) and job aids for health workers; national in-service trainings for front-line health workers linked to guideline revisions for both case management and malaria diagnostics (Ministry of Health, 2014b); and strengthening of supervision and quality assurance. 3.4.4 Trends in diagnostic capacity and AL availability The evaluation used the biannual nationally representative, health-facility surveys on quality of care for malaria in public and non-profit health facilities to assess trends in case management and malaria commodity availability. Between January 2010 and December 2015, ten health facility surveys were conducted. There were no reliable routine data on commodity availability before 2010. Compared to baseline in 2010, there was near universal capacity of health facilities to provide parasitological malaria diagnosis by the end of 2015 (Figure 3.13), representing a 42 percentage point increase in health facilities providing parasitological diagnosis. The increase in diagnostic capacity was due entirely to the increased availability of malaria RDTs, as a result of policy changes at the end of 2012. Figure 3.13: National trends in malaria diagnostic capacity in public and non-profit health facilities in Kenya, 2010–2015 39 Proportion of Facilities 100 80 60 40 51 55 53 58 54 59 54 65 76 56 31 91 70 51 77 47 40 91 69 49 98 91 44 97 84 44 20 8 9 13 17 0 Baseline 2010 FU1 2010 FU2 2011 FU3 2012 FU4 2012 FU5 2013 FU6 2014 FU7 2014 FU8 2015 FU9 2015 Malaria microscopy Malaria RDT Any malaria diagnostics Note: FU=Follow up Source: Machini B, Memusi D, Njiru P, Kigen S, Kimbui R, Amboko B, Zurovac D, Kiptui R, Waqo E. Monitoring outpatient malaria case management under the 2010 diagnostic and treatment policy in Kenya: progress January 2010–December 2015. Nairobi, Kenya: Ministry of Health, February, 2016. The commodity assessments on survey days showed that the availability of at least one AL pack was very high from 2010 to 2015 with the exception of 2014 (Figure 3.14). The proportion of health facilities stocking all four AL weight-band packs was significantly lower and more erratic throughout the monitoring period. The lack of malaria commodities observed at the health-facility level was a function of primarily the transition to a ‘pull’ system of commodity distribution following devolution of health service delivery to counties. 40 100 94 97 93 92 97 95 92 89 82 Proportion of facilities with AL 80 60 40 20 65 72 45 61 72 72 44 76 23 43 46 0 Baseline 2010 FU1 2010 FU2 2011 FU3 2012 FU4 2012 FU5 2013 FU6 2014 FU7 2014 FU8 2015 FU9 2015 One or more AL packs All AL packs Figure 3.14: National trends in the availability of artemether-lumefantrine at public-sector and non-profit health facilities on the day of the survey in Kenya, 2010–2015 Note: FU –f ollow up; AL – artemether-lumefantrine Source: Machini B, Memusi D, Njiru P, Kigen S, Kimbui R, Amboko B, Zurovac D, Kiptui R, Waqo E. Monitoring outpatient malaria case management under the 2010 diagnostic and treatment policy in Kenya: progress January 2010–December 2015. Nairobi, Kenya: Ministry of Health, February, 2016. Stock-out of malaria medicine was defined as stock-out of at least 7 consecutive days over a 3-month period prior to the surveys. A substantial decline in AL stock-outs was observed with the proportion of health facilities reporting simultaneous stock-out of all four AL packs declining from 27 percent in 2010 to 12 percent during the last survey round (Figure 3.15). Figure 3.15: National trends in artemether-lumefantrine stock-out measured in the 3 months prior to the survey, 2010–2015 100 60 52 45 39 45 22 56 73 74 54 27 21 6 9 22 7 21 24 29 12 0 10 20 30 40 50 60 70 80 90 Proportion of facilities with AL stock-out Baseline FU1 FU2 FU3 FU4 FU5 FU6 FU7 FU8 FU9 2010 2010 2011 2012 2012 2013 2014 2014 2015 2015 One or more AL packs Note: FU=Follow up; AL – artemether-lumefantrine Source: Machini B, Memusi D, Njiru P, Kigen S, Kimbui R, Amboko B, Zurovac D, Kiptui R, Waqo E. Monitoring outpatient malaria case management under the 2010 diagnostic and treatment policy in Kenya: progress January 2010–December 2015. Nairobi, Kenya: Ministry of Health, February, 2016. 41 18 29 28 13 32 21 11 26 37 14 37 44 12 45 58 0 20 40 60 80 100 Proportion of children with fever in the last two weeks who received a finger or heel prick KMIS 2010 KDHS 2014 KMIS 2015 Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic 13 35 39 0 20 40 60 80 100 Proportion of children with fever in the last two weeks who received a finger or heel prick KMIS 2010 KDHS 2014 KMIS 2015 3.4.5 Trends in malaria case management 3.4.5.1 Population-based indicators Diagnostic testing in children Universal parasitological diagnosis to confirm malaria in febrile patients before treatment was introduced in 2009. However, malaria RDTs were not available widely until 2014. Prior to 2009, presumptive treatment for malaria was the norm and was policy for children under 5 years consistent with IMCI guidance (Ministry of Health, 2006). Diagnostic testing was measured at the population level using a proxy indicator defined as the proportion of children under 5 years with fever who had blood taken from a finger or heel stick. Nationally, the proportion of children tested was 13 percent in 2010 and increased to 39 percent in 2015 (Figure 3.16). Figure 3.16: Percentage of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing in Kenya, 2010-2015 Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) The lake- and coastal-endemic zones experienced the largest increases in diagnostic testing of children under 5 years. In the lake-endemic zone, diagnostic testing increased from 12 percent in 2010 to 58 percent in 2015, and in the coastal-endemic zone, testing increased from 14 percent in 2010 to 44 percent in 2015 (Figure 3.17). Figure 3.17: Distribution of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing by endemicity zone in Kenya, 2010-2015 Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 42 Background characteristic KMIS 2010 KDHS 2014 KMIS 2015 Percentage point change 2003-2005 (95% CI) p-value % (95% CI) n % (95% CI) n % (95% CI) n Sex Male Female Residence Urban Rural Wealth Quintiles Lowest Second Middle Fourth Highest Total 14.1 (9.8-19.8) 11.9 (8.4-16.6) 27.9 (15.7-44.5) 11.7(8.8-15.4) 9.1 (5.9-13.9) 7.7 (3.3-16.9) 12.7 (8.8-18.2) 12.4 (6.6-22.0) 37.3 (24.4-52.4) 13.0 (10.0-16.7) 445 429 85 789 396 48 236 109 85 874 35.2 (32.5-38.0) 35.0 (32.3-37.7) 38.8 (35.1-42.7) 33.3 (31.0-35.8) 30.6 (27.3-34.1) 37.0 (33.3-40.9) 31.9 (27.8-36.2) 34.1 (29.7-38.7) 44.5 (39.3-49.9) 35.1 (33.1-37.1) 2,402 2,340 1,484 3,258 1,524 1,138 836 706 538 4742 36.7 (31.9-41.8) 42.0 (36.3-47.9) 44.0 (36.7-51.5) 37.2 (31.8-42.9) 29.2 (23.1-36.3) 42.3 (33.9-51.3) 39.7 (31.3-48.7) 37.5 (28.6-47.3) 51.8 (41.7-61.8) 39.2 (34.9-43.7) 654 636 460 830 393 287 241 201 168 1290 22.5 (17.6-27.5) 30.1 (25.2-35.0) 15.7 (5.1-26.3) 25.6 (21.6-29.5) 20.2 (14.9-25.5) 33.8 (24.1-43.5) 27.1 (19.6-34.6) 24.5 (15.3-33.6) 14.1 (1.4-26.9) 26.2 (22.7-29.7) <0.0001 <0.0001 0.007 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.033 <0.0001 Note: n=weighted number of children (denominator); baseline for the chi-square test is 2010 compared with results from 2015; data presented for 2010, 2014 and 2015. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Table 3.5 shows the demographic characteristics of children under 5 years of age with fever in the 2 weeks preceding the interview who had blood taken from a finger or heel. Significant increases were observed in both urban and rural areas from 2010 to 2015, with coverage increasing from 28 percent to 44 percent in urban areas and from 12 percent to 37 percent in rural areas. Table 3.5: Distribution of children under 5 years of age with fever in the 2 weeks before the survey who had blood taken from a finger or heel for testing, by demographic characteristic, in Kenya, 2010– 2015 3.4.5.1.1 Treatment in children In 2003, the percentage of children under the age of 5 years who had fever in the last 2 weeks and sought treatment for fever at an appropriate source (i.e., public and private health facilities, pharmacies and drug stores) was 60 percent. The percentage declined to 51 percent in 2007 and increased to 70 percent by 2015 (Figure 3.18). The percentage of children with fever who sought treatment and were given any antimalarials was 26 percent in 2003, reached a peak of 37 percent in 2010 then declined to 27 percent in 2015. Figure 3.19 shows that the percentage of children under 5 years with fever who were treated with any antimalarial decreased in all transmission zones except the lake-endemic zone from 2003 to 2015. Figure 3.18: Treatment seeking for children under 5 years of age with fever in the 2 weeks prior to the survey in Kenya, 2003–2015 60 51 58 50 72 70 0 20 40 60 80 100 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Proportion of children under 5 years who sought treatment for fever 43 100 Proportion of children under 5 years who sought treatment for fever and were treated with antimalarials 80 60 39 41 51 40 39 31 34 37 35 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 12 15 9 9 9 19 25 16 12 27 24 21 26 21 21 22 12 18 27 30 40 20 0 Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic 55 92 Proportion of children under 5 years treated with the recommended first-line antimalarial 80 60 40 20 0 42 11 34 52 86 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Note: Treatment from an appropriate source includes public and private health facilities, pharmacies and drug stores. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.19: Children under 5 years of age with fever in the 2 weeks prior to the survey who sought treatment and were treated with any antimalarial by endemicity zone in Kenya, 2003– 2015 Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Among children under 5 years who received treatment for malaria, 42 percent of children received the recommended first-line treatment (i.e., SP) in 2003 compared to 92 percent of children who received the recommended first-line treatment (i.e., AL) in 2015 (Figure 3.20). A similar trend was observed by malaria-endemicity zones (Figure 3.21). Figure 3.20: Children under 5 years of age treated with an antimalarial who received the recommended first-line treatment in Kenya, 2003–2015 100 44 Note: From 2003–2005, the recommended first-line antimalarial was sulfadoxine-pyrimethamine; from 2006–2015, the recommended first-line antimalarial was artemether-lumefantrine. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 45 41 20 44 29 70 82 39 21 39 55 68 84 52 14 43 44 71 90 34 3 49 65 87 95 40 7 23 58 93 94 0 20 40 60 80 100 Proportion of children under 5 years treated with the recommended first-line antimalarial KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic Demographic characteristic Sex KDHS 2003 KMIS 2010 KMIS 2015 Percentage point change 2003–2015 % n % n % n (95% CI) p-value Male Female 40.3 (33.6-47.3) 42.7 (35.3-50.5) 278 296 53.5 (43.6-63.2) 50.4 (39.3-61.5) 167 136 91.0 (84.3-95.0) 92.3 (86.4-95.8) 169 167 50.8 (43.7-58.0) 49.6 (42.7-56.6) <0.0001 <0.0001 Residence Urban Rural 42.8 (36.6-49.3) 34.1 (24.5-45.3) 447 127 47.1 (24.6-70.8) 52.7 (44.5-60.8) 40 263 92.9 (82.6-97.3) 91.2 (85.7-94.7) 115 221 50.3 (43.8-56.8) 57.5 (48.5-66.6) <0.0001 <0.0001 Mother’s education None 44.3 (32.9-56.4) 97 56.6 (37.0-74.3) 41 76.8 (51.6-91.1) 39 32.6 (16.1-49.1) 0.001 Primary 39.3 (33.2-45.8) 354 54.4 (45.1-63.4) 191 90.0 (83.8-94.0) 201 50.8 (44.2-57.3) <0.0001 Secondary 46.2 (35.7-57.1) 123 45.3 (31.4-60.1) 71 97.4 (91.0-99.3) 96 51.6 (42.3-60.8) <0.0001 Total 41.5 (36.0-47.3) 574 52.2 (44.4-59.9) 303 91.6 (87.1-94.7) 336 50.2 (45.2-55.2) <0.0001 Note: n=weighted number of children (denominator); baseline for the chi-square test is 2003 compared with results from 2015; data presented for baseline 2003, midline 2010 and endline 2015. From 2003–2005, the recommended first-line antimalarial was sulfadoxine-pyrimethamine; from 2006–2015, the recommended first-line antimalarial was artemether￾lumefantrine. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Figure 3.21: Children under 5 years of age treated with an antimalarial who received the recommended first-line treatment by endemicity zone in Kenya, 2003–2015 Note: From 2003–2005, the recommended first-line antimalarial was sulfadoxine-pyrimethamine; from 2006–2015, the recommended first-line antimalarial was artemether-lumefantrine. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) Treatment with the recommended first-line antimalarial did not vary by sex or residence of the child. By 2015, there was a substantial disparity in use of the recommended treatment by mother’s education, with 97 percent of the children whose mothers had a secondary or higher education receiving the recommended treatment versus 77 percent of children whose mothers had no education (Table 3.6). Table 3.6: Use of recommended first-line treatment among children under 5 years of age who took an antimalarial, by demographic characteristic, in Kenya, 2003–2015 3.4.5.2 Health-facility-based indicators The evaluation used the biannual nationally-representative, health-facility surveys, conducted between 2010 and 2015, on quality of care in public and non-profit health 46 facilities to assess trends in malaria case management. In line with national case￾management guidelines (MOPHS, 2010), a composite malaria case-management performance indicator was developed. To meet the performance indicator, the following criteria had to be met: 1) the febrile patient was tested for malaria; 2) if a positive test result was reported, the patient was treated with AL, and 3) if a negative test result was reported, the patient was not treated for malaria. Overall, the composite performance indicator significantly improved from 28.1 percent in 2010 to 61.2 percent in 2015. The recommended first-line treatment, AL, was used for the majority of patients with positive test results between 2010 and 2015. Among patients with a negative test result, a significant decline in treatment with an antimalarial was observed in 2015 compared to the 2010 (52 percent versus 7 percent) (Figure 3.22). Figure 3.22: Trends in the performance of malaria diagnosis and treatment in accordance with national case-management guidelines among patients with fever seen at public and non-profit health facilities in Kenya, 2010–2015 99 ts 100 83 90 89 93 90 83 88 91 80 75 ien f Febrile Pat 60 53 43 50 44 57 40 58 44 58 48 63 55 57 47 67 60 69 62 66 61 ion o 40 28 36 30 31 21 Proport 20 17 17 10 9 8 0 Baseline FU1 FU2 FU3 FU4 FU5 FU6 FU7 FU8 FU9 2010 2010 2011 2012 2012 2013 2014 2014 2015 2015 Composite performance* Malaria test performed Test +ve Rx with AL Test -ve Rx with any AM Note: FU – follow up, AM – antimalarial; +ve – positive; -ve – negative *Composite performance indicator comprised three criteria: 1) febrile patient was tested for malaria; 2) if a positive test result was reported, the patient was treated with artemether-lumefantrine, and 3) if a negative test result was reported, the patient was not treated with an antimalarial. Source: Machini B, Memusi D, Njiru P, Kigen S, Kimbui R, Amboko B, Zurovac D, Kiptui R, Waqo E. Monitoring outpatient malaria case management under the 2010 diagnostic and treatment policy in Kenya: progress January 2010–December 2015. Nairobi, Kenya: Ministry of Health, February, 2016. 3.4.6 Equity in malaria case management Treatment with the recommended first-line antimalarial was variable with socioeconomic status across the national surveys with no discernible sustained trend (Figure 3.23). 47 Figure 3.23: Children under 5 years of age treated with an antimalarial who received the recommended first-line treatment by wealth quintile in Kenya, 2003–2015 36 7 28 55 83 84 48 21 29 50 88 94 38 9 37 59 89 92 53 8 26 41 86 97 40 9 50 45 72 88 0 10 20 30 40 50 60 70 80 90 100 KDHS 2003 KMIS 2007 KDHS 2008 KMIS 2010 KDHS 2014 KMIS 2015 Proportion of children under 5 years treated with the recommended first-line antimalarial Lowest Second Middle Fourth Highest Note: From 2003–2005, the recommended first-line antimalarial was sulfadoxine-pyrimethamine; from 2006–2015, the recommended first-line antimalarial was artemether-lumefantrine. Source: Kenya Demographic Health Survey (KDHS); Kenya Malaria Indicator Survey (KMIS) 3.4.7 Malaria case management summary • The availability of malaria diagnostics increased from 55 to 97 percent in public sector health facilities from 2010 to 2015, with all of the gains from increased availability of malaria RDTs. • The recommended first-line medication, AL, was available in 90 percent or more of public sector health facilities from 2010 to 2015, except for during 2014. • Healthcare worker adherence to national treatment guidelines more than doubled from 28 to 61 percent nationally from 2010 to 2015 in public sector health facilities. • Overall, children under 5 years of age treated with an antimalarial who received the recommended first-line medication increased from 42 to 92 percent from 2003 to 2015. • In the lake- and coastal-endemic zones, children under 5 years of age treated with an antimalarial who received the recommended first-line medication increased from 40 to 94 percent and from 34 to 95 percent, respectively, from 2003 to 2015. 48 4 Trends in Malaria Morbidity 4.1 Background The most commonly used measure of malaria transmission is the malaria parasite prevalence rate (Smith and Hay, 2009, Hay et al., 2008), measured during cross￾sectional population surveys. In stable malaria-transmission areas, such as the lake￾and coastal-endemic areas, severe anemia is a common consequence of malaria infection in children and contributes to a large proportion of deaths due to malaria (Korenromp et al., 2004, Odhiambo et al., 2008). Infection with the malaria P. falciparum parasite causes the destruction of erythrocytes, or red blood cells, and reduces erythrocyte production in the bone marrow leading to anemia (Haldar and Mohandas, 2009). Changes in the prevalence of malaria parasites and severe anemia in young children in high-transmission settings are, therefore, strong signals of the changes in the burden of malaria. For purposes of monitoring the impact of malaria control interventions, severe anemia is defined as blood hemoglobin levels <8 g/dL (Korenromp et al., 2011). Prior to the national expansion of malaria RDTs in 2012, most malaria cases seen at health facilities were diagnosed by clinical signs and symptoms, and fevers among children were empirically treated as malaria. Since 2010, national guidelines have recommended parasitological diagnosis for all persons suspected of having malaria (MOPHS, 2010)), but regular reporting of test results only started in earnest in 2015 (Githinji et al., 2017). Therefore, available routine health data from the health management information system, District Health Information Software 2 (DHIS2), do not provide sufficient data points for malaria indicators over time to inform the impact evaluation. Alternatively, malaria inpatient data from a set of hospitals (Okiro et al., 2009, 2010, 2011, 2013) have been used to describe the trends in confirmed malaria cases. 4.2 Population-Based Trends in Malaria Morbidity in Children There have been three national Kenya malaria indicator surveys that measured the prevalence of malaria parasitemia and anemia among children under 5 years of age at the community level in 2007, 2010 and 2015. Each of the three surveys measured malaria parasitemia and anemia in different age groups: 3 to 59 months (2007), 3 months to 14 years (2010), and 6 months to 14 years (2015). For comparison purposes, data for children ages 6 to 59 months were analysed across all surveys. 4.2.1 Trends in prevalence of malaria parasites in children Malaria parasite prevalence in children under 5 years of age, as measured using microscopy, was 3.3 percent nationally in 2007, rose to 9.2 percent in 2010 and declined to 5.0 percent in 2015. In each survey, parasite prevalence was highest in the lake- and coastal-endemic areas. The rise in 2010 was mainly attributable to an increase in the parasite prevalence rate in the lake-endemic zone. However, by 2015, the parasite prevalence (16.6 percent) had been reduced by almost half in the lake￾endemic zone compared to the 2010 prevalence (32.9 percent). In the coastal￾endemic zone, the parasite prevalence rose from 3.1 percent in 2010 to 5.3 percent in 2015 (Figure 4.1). 49 Figure 4.1: Malaria parasite prevalence by microscopy in children under 5 years of age by malaria endemicity zone in Kenya, 2007–2015 45 32.9 Total (National) Low risk Seasonal Highland Coastal Endemic Lake Endemic transmission epidemic KMIS 2007 KMIS 2010 KMIS 2015 Percentage of children ages 6 to 59 months infected with malaria 3.3 0.0 0.2 0.4 9.2 4.0 8.7 0.1 0.2 1.8 3.1 5.0 0.4 0.5 2.6 5.3 16.6 0 5 10 15 20 25 30 35 40 Note: Parasite prevalence data not available for Kenya demographic and health surveys due to lack of biomarker testing. Source: Kenya Malaria Indicator Survey (KMIS) Across the three surveys, malaria parasitemia prevalence was highest in the poorest quintiles and in rural areas (Table 4.1). Table 4.1: Malaria parasite prevalence in children under 5 years of age by demographic characteristic in Kenya, 2007–2015 Background KMIS 2007 KMIS 2010 KMIS 2015 Percentage characteristic point change (95% p-value* CI) %(95% CI) n %(95% CI) n %(95% CI) n Sex Male 3.4 (2.4-4.9) 2,360 9.7 (7.3-12.9) 2,058 4.9 (3.6-6.7) 1,733 1.5 (0.3-2.8) 0.015 Female 3.2 (2.1-4.9) 2,320 8.7 (6.1-12.2) 2,094 5.0 (3.6-6.8) 1,707 1.8 (0.5-3.0) 0.004 Residence Urban 1.4 (0.5-3.9) 645 1.0 (0.4-2.7) 479 1.9 (1.1-3.2) 1,301 0.5 (-0.6-1.7) 0.404 Rural 3.6 (2.4-5.2) 4,035 10.7 (8.0-14.2) 3,673 6.3 (4.6-8.5) 2,139 2.7 (1.5-3.9) <0.0001 Wealth Quintiles Lowest 3.9 (1.6-9.1) 1,060 13.8 (9.8-19.1) 1,807 5.7 (3.6-9.1) 1,104 1.8 (0.0-3.6) 0.0406 Second 4.0 (2.6-6.1) 1,163 7.6 (3.9-14.1) 214 8.9 (6.4-12.4) 745 4.8 (2.5-7.2) <0.0001 Middle 4.2 (2.4-7.2) 845 10.7 (7.4-15.2) 1,179 4.7 (3.1-7.2) 609 0.6 (-1.5-2.8) 0.570 Fourth 2.8 (1.7-4.5) 982 3.7 (2.1-6.6) 461 2.9 (1.7-5.1) 526 0.1 (-1.6-1.9) 0.908 Highest 0.7 (0.3-1.7) 630 0.6 (0.1-2.3) 491 0.9 (0.3-2.9) 456 0.2 (-0.8-1.3) 0.645 Note: n=weighted number of children (denominator); *baseline for the chi-square test is 2007 compared with results from 2015; data presented for baseline 2007, midline 2010 and endline 2015. Parasite prevalence data not available for Kenya demographic and health surveys due to lack of biomarker testing. Source: Kenya Malaria Indicator Survey (KMIS) 50 4.0 1.3 4.2 2.6 7.6 4.9 4.0 0.6 5.4 2.7 3.2 6.9 2.2 0.5 3.1 2.3 2.2 3.4 0.0 5.0 10.0 15.0 20.0 Total (National) Low risk Seasonal transmission Highland epidemic Coastal Endemic Lake Endemic Percentage of children 6-59 months with severe anaemia KMIS 2007 KMIS 2010 KMIS 2015 4.2.2 Trends in severe anemia prevalence (Hb<8g/dL) in children Nationally, severe anemia showed similar trends as malaria parasitemia, rising from 4.4 percent in 2007 to 5.1 percent in 2010 before declining to 2.6 percent in 2015. Although severe anemia was highest in the coastal-endemic zone in 2007, declines were observed in each subsequent survey. However, severe anemia increased from 2007 to 2010 in the lake-endemic, highland epidemic-prone and seasonal-risk zones before declining in 2015; however, the declines were not statistically significant. Similar to malaria parasitemia, by 2015, the severe anemia prevalence (3.4 percent) had been reduced by over half in the lake-endemic zone compared to the 2010 prevalence (6.9 percent) (Figure 4.2). Figure 4.2: Severe anemia prevalence in children ages 6 to 59 months overall and by malaria endemicity zone in Kenya, 2007–2015 Note: Parasite prevalence data not available for Kenya demographic and health surveys due to lack of biomarker testing. Source: Kenya Malaria Indicator Survey (KMIS) Severe anemia declined across all wealth quintiles except the second quintile between 2003 and 2015. Between 2003 and 2015, severe anemia in children living in both urban and rural areas also declined significantly (Table 4.2). 51 Table 4.2: Severe anemia prevalence in children ages 6 to 59 months by demographic characteristic in Kenya, 2007–2015 Characteristic KMIS 2007 %(95% CI) n KMIS 2010 %(95% CI) n KMIS 2015 %(95% CI) n Percentage point change (95% CI) * p-value Sex Male Female Residence Urban Rural Wealth Quintiles Lowest Second Middle Fourth Highest 4.4 (3.4-5.6) 3.6 (2.7-4.6) 6.6 (4.5-9.6) 3.6 (2.8-4.5) 6.2 (4.2-9.1) 3.2 (2.1-4.8) 3.5 (2.3-5.4) 3.3 (2.2-4.9) 4.3 (2.9-6.3) 2,353 2,305 644 4,014 1,056 1,158 836 978 630 4.8 (3.5-6.4) 3.2 (2.3-4.5) 1.6 (0.7-3.9) 4.4 (3.4-5.6) 5.2 (3.8-7.1) 3.4 (1.4-8.2) 3.7 (2.4-5.8) 4.0 (2.5-6.5) 1.7 (0.8-3.7) 2,058 2,094 479 3,673 1,807 214 1,179 461 491 2.5 (1.7-3.6) 1.9 (1.3 -2.9) 1.6 (0.9-2.6) 2.5 (1.7-3.6) 2.9 (1.8-4.7) 3.6 (2.2-5.9) 1.7 (0.9-3.0) 1.1 (0.4-2.7) 1.1 (0.5-2.4) 1,732 1,707 1,304 2,135 1,104 744 606 527 458 -1.9 (-3.0-0.8) -1.7 (-2.7-0.7) -5.1 (-7.1-3.0) -1.1 (-2.0-0.3) -3.3 (-5.0-1.5) 0.4 (-1.2-2.1) -1.8 (-3.4-0.2) -2.1 (-3.6-0.7) -3.2 (-5.0-1.3) 0.001 0.001 <0.0001 0.017 <0.0001 0.609 0.036 0.012 0.002 Total 4.0 (3.2-4.8) 4,658 4.0 (3.1-5.0) 4,152 2.2 (1.6-3.0) 3,439 -1.8 (-2.5-1.0) <0.0001 Note: n=weighted number of children (denominator); *baseline for the chi-square test is 2007 compared with results from 2015; data presented for baseline 2007, midline 2010 and endline 2015. Parasite prevalence data not available for Kenya demographic and health surveys due to lack of biomarker testing. Source: Kenya Malaria Indicator Survey (KMIS) 4.3 Health facility malaria morbidity trends In Kenya, routine malaria indicator data was reported to the HIS (i.e., DHIS2 and integrated disease surveillance and response [IDSR]) platform. Malaria indicator data from DHIS2 includes total malaria cases (i.e., clinical and confirmed cases) and malaria commodity consumption reported monthly in the logistic management information system integrated into DHIS2. The malaria indicators reported to IDSR include incidence and malaria test positivity rate (TPR) reported weekly. In 2013, IDSR reporting changed to electronic IDSR (i.e., eIDSR). The primary change was at the county level. Prior to 2013, health facilities transmitted reports via hard copies, emails and short message service (SMS or “text message”) to the district level for aggregation and data entry into Excel spreadsheets, which were sent to the provincial and then national levels. After 2013, the data that health facilities sent to the county was directly entered into the web-based eIDSR, which was then available to the national level. The changes in the systems and reporting structures have led to fluctuations in reporting rates over the evaluation period. The percentage of health facilities reporting any health indicator data via DHIS2 has been consistently at or above 90 percent since 2012 (Figure 4.3). The percentage of health facilities reporting any health indicator data via IDSR and LMIS has been less consistent at approximately 70 percent since 2013. 52 Figure 4.3: Monthly percentage of health facilities reporting data to the routine health information system by reporting platform in Kenya, 2012–2016 Note: IDSR – integrated disease surveillance and response; LMIS – logistic management information system; DHIS2 –District Health Information Software 2 Source: Kenya Health Information System, accessed August 2016 Due to reporting rates for eIDSR, which were below the benchmark of 85%, this evaluation did not analyze TPR data from the routine system. Figure 4.4 is an example of the routine malaria indicator data from DHIS2 showing the total malaria cases (i.e., confirmed plus clinical cases) since 2012. The apparent increase in confirmed malaria cases might be a reflection of the increased capacity to test and provide a proper malaria diagnosis as per national treatment guidelines (add reference). Overall, there is a declining trend in combined malaria cases, with clinical cases decreasing as confirmed cases increased. 53 Figure 4.4: Malaria cases reported via the routine health information system in Kenya, 2011– 2015 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Total_Malaria_Cases (Clinical + Confirmed Malaria) Confirmed_Malaria Clinical Malaria Linear (Total_Malaria_Cases (Clinical + Confirmed Malaria)) Source: Kenya Health Information System, accessed August 2016 Cases (Millions) Jan-11 Jul-11 Jan-12 Jul-12 Jan-13 Jul-13 Jan-14 Jul-14 Jan-15 Jul-15 4.3.1 Summary of malaria morbidity • Malaria parasitemia prevalence increased from 3 to 5 percent nationally among children ages 6 to 59 months from 2007 to 2015. • The prevalence of severe anemia (i.e., hemoglobin <8 g/dL) among children ages 6 to 59 months declined nationally from 4 to 2 percent from 2007 to 2015. • In the lake- and coastal-endemic zones, the prevalence of severe anemia among children 6 to 59 months declined from 5 to 3 percent and 8 to 2 percent, respectively, from 2007 to 2015. • Overall, combined malaria cases (i.e., confirmed plus clinical) reported monthly to the routine health information system decreased from 2012 to 2015 with over 90 percent of health facilities reporting. 5 Trends in All-Cause Child Mortality 5.1 All-Cause Child Mortality Trends in ACCM rates for the period 2003–2014 were estimated using the Kenya Demographic and Health Surveys. The surveys represent direct estimates during the period 0 to 4 years before each survey. The ACCM declined from 115 deaths per 1,000 live births in 2003 to 52 deaths per 1,000 live births in 2014, a 54 percent reduction. The ACCM estimates show a sustained decline (Figure 5.1). 54 Figure 5.1: Trends in all-cause child mortality in Kenya, 2003–2014 115 74 52 0 20 40 60 80 100 120 140 2003 2008/9 2014 Deaths per 1000 live births Source: Kenya Demographic and Health Surveys (KDHS) The magnitude of change in the ACCM rate varied by malaria-endemicity zone, with the largest reduction observed in the lake-endemic zone. In the lake-endemic zone, the ACCM rate decreased from 213 deaths per 1,000 live births in 2003 to 64 deaths per 1,000 live births in the period 2014. (Figure 5.2). The largest reduction in ACCM occurred in the lake-endemic zone. Because the lake-endemic zone had highest malaria transmission, the zone also had the greatest potential to benefit from malaria prevention and control interventions. Figure 5.2: Trends in all-cause child mortality by malaria-endemicity zone in Kenya, 2003– 2014 55 71 100 90 93 213 48 71 46 65 130 55 42 43 54 64 0 50 100 150 200 250 300 Low Risk Seasonal/Semi-arid Highland epidemic Coastal endemic Lake endemic Deaths per 1000 live births KDHS 2003 KDHS 2008/9 KDHS 2014 Source: Kenya Demographic and Health Surveys (KDHS) 5.2 Age-Specific Mortality Significant reductions in ACCM occurred in all age groups (i.e., neonatal, postneonatal, infant, and child) from 1999 to 2003 and 2010 to 2014 (Figure 5.3). The largest decline in mortality was among children ages 1–4 years of age (65 percent) (Figure 5.4). Figure 5.3: Trends in age-specific all-cause mortality in Kenya, 2003–2014 Deaths per 1000 live births 140 120 100 80 60 40 20 0 115 77 74 52 52 33 31 44 39 41 22 21 16 23 14 N eo n a t a l P o s t n eo n a t a l I nf a nt C h ild Under f ive 2003 2008/9 2014 Source: Kenya Demographic and Health Surveys (KDHS) Figure 5.4: Relative percentage change in age-specific all-cause mortality rates in Kenya, 2003–2014 56 Table 5.1: All-cause child mortality, by demographic characteristic, in Kenya, 2003–2014 Characteristic KDHS 2003 ACCM per 1,000 live births KDHS 2008/9 ACCM per 1,000 live births KDHS 2014 ACCM per 1,000 live births Relative percent change (KDHS 2003 and KDHS 2014) Sex Male Female Residence Urban Rural Wealth Lowest Second Middle Fourth Highest Mother’s education None Primary incomplete Primary complete Secondary or higher 126.5 102.0 93.4 119.3 145.7 113.2 120.8 81.5 98.3 122.2 143.4 98.9 69.1 81.8 64.9 64.6 75.5 88.7 78.3 81.0 44.6 68.4 83.1 90.4 58.0 60.2 54.4 50.2 57.0 49.7 52.8 55.6 50.0 56.5 46.6 46.9 58.0 50.9 49.9 -57.0 -50.8 -39.0 -58.3 -63.8 -50.9 -58.6 -30.7 -52.6 -61.6 -59.6 -48.5 -27.8 Total 114.5 73.6 52.3 -54.3 Source: Kenya Demographic and Health Surveys (KDHS) Source: Kenya Demographic and Health Surveys (KDHS) -33 -63 -50 -65 -54 -70 -60 -50 -40 -30 -20 -10 0 10 N eo n a t a l P o s t n eo n a t a l Inf ant C h ild Unde r f i v e Percentage change 5.3 Equity in Change in Mortality In rural areas, ACCM declined from 119.3 deaths per 1,000 live births in 1999–2003 to 49.7 deaths per 1,000 live births in 2010–2014, a reduction of 58 percent. A decline of 39 percent in ACCM was observed in urban areas over the same period. The largest declines in ACCM during the evaluation period were observed among children in the poorest wealth quintile (64 percent) and among children whose mothers had no formal education (62 percent) (Table 5.1). 57 5.3.1 Summary of all-cause child mortality • The ACCM declined nationally by 54 percent from 115 deaths per 1,000 live births in 2003 to 52 deaths per 1,000 live births in 2014, with reductions observed in all malaria epidemiological zones. • The lake-endemic zone had the largest mortality reduction at 70 percent from 213 deaths per 1,000 live births in 2003 to 64 deaths per 1,000 in 2014. • Reductions in mortality occurred in all age groups (i.e., neonatal, postneonatal, infants, and children) from 2003 to 2014; the largest decline of 65 percent was observed among children ages 1–4 years of age. • Mortality reductions were equitable from 2003 to 2014; mortality declined most in rural areas (58 percent), among children in the poorest wealth quintile (64 percent) and among children whose mothers had no formal education (62 percent). 58 6 Trends in Contextual Factors 6.1 Background Impact evaluations based on plausibility inferences require collecting data on non￾malaria programs and other factors, collectively referred to as contextual factors, which might offer alternate explanations for the observed changes in malaria transmission, morbidity, and ACCM. Appropriate consideration of contextual factors is essential for ensuring the internal and external validity of evaluations of large­scale health programs (Victora et al., 2005), particularly for evaluations that are conducted when rapid changes are under way in many other aspects of health services (Bryce et al., 2004). Contextual factors are also important to consider when the associations of interest are based on ecological data, which describes this evaluation. Contextual factors associated with childhood mortality and morbidity, including malaria, can be broadly categorized into the fundamental and proximate determinants of disease ( Link and Phelan, 1996, Jones et al., 2003, Mosley and Chen, 2003, Stratton et al., 2008,). Fundamental determinants are the social and economic conditions under which people live while proximate determinants are biological risks. The conceptual framework (Bryce et al., 2005, Rowe et al., 2007, Rowe et al., 2011,) for the evaluation design (Figure 6.1) incorporates numerous contextual factors within subcategories of the fundamental and proximate determinants of disease. A review of relevant information on the levels and trends of contextual determinants, both fundamental and proximate, of childhood mortality and morbidity follows. Contextual factors data were obtained from national household surveys such as the KDHS and KMIS as well as other sources such as the World Bank, WHO and UNICEF, as indicated. Figure 6.1: Conceptual framework for the impact evaluation of the Kenya national malaria control program, 2003–2014 Note: LLIN – long-lasting insecticidal bed net; IRS – indoor residual spraying; IPTp – intermittent preventive treatment of malaria in pregnancy; GDP – gross domestic product; EPI – expanded program on immunization 59 6.2 Fundamental Determinants 6.2.1 Socioeconomic factors A range of socioeconomic determinants at the community, household, and individual level are associated with child survival (Mosley and Chen, 2003, Wang, 2003, Boyle et al., 2006) as shown in the impact model in Figure 6.1. 6.2.1.1 Changes in gross domestic product Economic poverty, at both the country and individual levels, strongly correlates with poorer health outcomes (Subramanian et al., 2002). The GDP per capita, a measure of population wealth in a country, is considered to be a typical macroeconomic determinant of health and has an inverse and significant relationship between income and child mortality (O'Hare et al., 2013).Studies exploring the effect of GDP per capita purchasing power parity (GDP-PPP) on childhood mortality have found that every unit increase in GDP-PPP is associated with a 27–29 percent proportionate decline in child mortality ( Imam and Koch, 2004, Omariba et al., 2007, O'Hare et al., 2013). Previous national malaria impact evaluations have observed trends similar to and consistent with studies of GDP per capita and childhood mortality (add references here). Trends in GDP-PPP in US dollars and child mortality in Kenya are shown in Figure 6.2. Kenya’s GDP-PPP was US$2,146 in 2003 and US$2,818 in 2014, a 31 percent increase in constant international dollars to control for inflation (World Bank, 2016) (Figure 6.2). Figure 6.2: Trends in gross domestic product per capita purchasing power parity and all-cause child mortality in Kenya, 2000–2014 Note: GDP – gross domestic product Source: GDP data from World Bank, 2016; Mortality data from the Kenya Demographic Health Survey (KDHS) 6.2.1.2 Maternal education and marital status Female literacy and education significantly influence under-5 mortality (Imam and Koch, 2004, Omariba et al., 2007, Nattey et al., 2013). The proportion of females ages 15–49 years with at least a primary school education in Kenya increased by 16.4 percentage points between 2003 and 2014, while the proportion of literate females ages 15–49 years also significantly increased by 9.5 percentage points over the same period. Marital status remained constant, at 54 percent throughout the evaluation period (Table 6.1). 60 Indicator KDHS 2003 KDHS2014 Percentage point change (95% CI) p-value % (95% CI) n % (95% CI) n % women with at least a primary school 49.4 (47.2–51.5) 8,195 65.8 (64.6–66.9) 31,079 16.4 (3.6–19.2) <0.001 education % women literate 78.5 (76.5–80.4) 8,195 88.0 (87.3–88.7) 31,079 9.5 (7.3–11.7) <0.001 % women married 54.5 (53.0–55.9) 8,195 54.6 (53.6–55.5) 31,079 0.1 (-1.7–1.9) 0.891 Note: CI – confidence interval; p-values <0.05 are considered significantly different; percentage point change calculated as difference between percentage in 2014 (endline) and percentage in 2003 (baseline). Source: Kenya Demographic and Health Survey (KDHS) Indicator KDHS 2003 KDHS 2014 Percentage point change (95% CI) p-value % (95% CI) n % (95% CI) n Access to improved drinking water source 40.5 (37.4–43.7) 8,175 69.4 (67.8– 70.9) 36,423 28.9 (25.1– 32.8) <0.001 Access to improved sanitation 19.4 (17.4–21.6) 8,561 22.7 (21.1– 24.3) 36,421 3.3 (0.3, 6.2) 0.031 Household has improved house flooring 37.8 (35.1–40.5) 8,548 52.7 (51.1– 54.3) 36,423 14.9 (11.3– 18.5) <0.001 Household has electricity 16.0 (14.0– ,18.3) 8,548 36.0 (34.3– 37.7) 36,409 20.0 (16.8– 23.1) <0.001 Household has telephone 12.9 (11.4–14.5) 8,543 86.1 (85.4– 86.7) 36,408 73.2 (71.3– 75.1) <0.001 Note: CI – confidence interval; p-values <0.05 are considered significantly different; percentage point change calculated as difference between percentage in 2014 (endline) and percentage in 2003 (baseline). Source: Kenya Demographic and Health Surveys (KDHS) Table 6.1: Education and marital status of women ages 15-49 years in Kenya, 2003–2014 6.2.1.3 Household factors Household and microeconomic factors are important determinants of child health and malaria risk (Wang, 2003). Socioeconomic differentials at the household level are associated with access to malaria interventions ( Yé et al., 2006, Mmbando et al., 2011, Günther and Fink, 2011), thereby increasing the vulnerability of the poorest to malaria (Worrall et al., 2005). Households with access to safe water and proper sanitation facilities have lower childhood mortality rates (Van Bodegom et al., 2012, Mesike and Mojekwu, 2012, Kayode et al., 2012). There were changes in household-level fundamental determinants of mortality during the evaluation period. The proportion of households with access to improved drinking water sources increased from 40.5 percent in 2003 to 69.4 percent in 2014, and the proportion of households with access to improved sanitation facilities increased from 19.4 percent in 2003 to 22.7 percent in 2014. Households with improved flooring material (i.e., not earth, sand, or dung) increased by 14.9 percentage points to 52.7 percent, and households with electricity increased by 20.0 percentage points to 36.0 percent between 2003 and 2014. Households with telephones, mostly mobile phones, increased significantly from 12.9 percent in 2003 to 86.1 percent in 2014 (Table 6.2.). Table 6.2: Change in household attributes and asset ownership in Kenya, 2003–2014 61 6.2.2 Climatic variability 6.2.2.1 Climate and malaria in Kenya Geography and climate variability are key determinants of malaria transmission in Kenya. Rainfall and temperature influence transmission in the four eco￾epidemiological zones. In the arid and semi-arid low-transmission zone, higher than normal monthly rainfall and flooding are associated with outbreaks of vector-borne diseases, including malaria epidemics if vector control measures are not instituted in a timely manner (Maes et al., 2014). In the epidemic-prone western highlands, climate variability, particularly related to rainfall during El Niño cycles, has resulted in the occurrence of malaria epidemics (Githeko and Ndegwa, 2001, Hay et al., 2003, Zhou et al., 2004). For the first time in two decades, the coastal-endemic zone saw a rise in malaria prevalence beginning in 2011, paradoxically during a regional drought and well before the higher-than-average rainfall experienced in 2014 (Snow et al., 2015). In the lake-endemic zone of western Kenya, malaria transmission fluctuates less than other zones with seasonal variations in rainfall (Mutuku et al., 2009, Sewe et al., 2016). 6.2.2.2 Rainfall Figure 6.3 shows the national annual rainfall compared with long-term average, computed from 1970–1999, for each year. The long-term average is shown on the graph as the “0” line. A period of drought followed the 1997–1998 El Niño cycle, which extended well into 2001. In2006, there was unseasonable heavy rainfall during the short rains that resulted in severe flooding and malaria epidemics, particularly in the arid and semi-arid lands. From 2007 to 2009, every year was drier compared to the long-term average.. From 2012 to 2014, Kenya has experienced normal to above￾normal precipitation patterns. Additional information on rainfall anomalies by malaria-endemic zone is shown in Annex 3. Figure 6.3: Long-term anomalies and actual yearly rainfall in Kenya, 2000–2014 -350 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300 350 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total Rainfall Anomalies (mm) Year Source: Kenya Meteorological Department, 2015. 62 6.2.2.3 Temperature The maximum, minimum, and mean annual national temperatures between 2000 and 2014 are shown in Figure 6.4. The data show that the national-level annual temperature did not vary significantly from the long-term mean temperatures computed from 1970–1999. However, nine of the 15 years during the period from 2000 to 2014 had mean annual temperatures 0.1°C–0.6°C warmer than the long-term annual average. Detailed annual temperatures by endemicity zone are shown in Annex 4. 63 Figure 6.4: Annual national mean, minimum, and maximum temperatures in Kenya, 2000– 2014 30.0 Temperature (oC) 25.0 20.0 15.0 10.0 5.0 0.0 2000 2001 2002 2003 2004 2005 MAX Temp MIN Temp MEAN Temp Note: MAX- maximum; MIN – minimum; Temp - temperature 2006 Source: Kenya Meteorological Department 2007 2008 2009 2010 2011 2012 2013 2014 In conclusion, climate variability is an important factor in the different eco￾epidemiological zones, particularly in the highland epidemic-prone and arid and semi￾arid seasonal-transmission zones. Overall, neither rainfall nor temperature patterns suggest climate differences or anomalies existed during the period of rapid malaria intervention expansion that would have independently resulted in substantially different patterns of malaria morbidity and mortality during the latter years versus the earlier years of the evaluation period. Inter-annual weather patterns may have influenced annual variation of malaria transmission; however, there is no evidence that the weather patterns changed the longer-term trends of malaria transmission during the evaluation period. 6.3 Proximate Determinants 6.3.1 Maternal and reproductive health indicators Visits to ANC are an important entry point to health services and care for women during and after pregnancy and thereafter for their infants (Lawn and Kerber, 2006). The ANC visits provide an opportunity for the identification and management of obstetric complications such as preeclampsia, delivery of tetanus toxoid immunization for the prevention of neonatal tetanus, IPTp, and identification and management of infections including HIV, syphilis, and other sexually-transmitted infections. Visits to ANC are also an opportunity to promote skilled birth attendance and healthy behaviours, such as breastfeeding, early postnatal care, and family planning. Focused ANC is a goal-oriented approach that categorizes pregnant women into those eligible to receive routine ANC (i.e, the basic universal component) in four focused visits during pregnancy and those who need specialised care for specific health conditions or risk factors (Lincetto et al., 2006). During the evaluation period, pregnant women were encouraged to attend at least four ANC visits. The proportion of women making at least four ANC visits increased by 5.3 percentage points, health-facility births increased by 21.4 percentage points and post-partum vitamin A supplementation increased by 39.8 percentage points from 2003 to 2014 (Table 6.3). During the evaluation period, there was also significant decrease in the proportion of 64 births with avoidable fertility risks. A summary of maternal and reproductive health indicators is shown in Table 6.3. 65 Table 6.3: Trends in maternal and reproductive health indicators in Kenya, 2003–2014 Indicator KDHS 2003 KDHS 2014 Percentage point change (95% CI) p-value % (95% CI) n % (95% CI) n Four or more ANC visits At least two doses of tetanus toxoid Birth with an avoidable fertility risk* Birth interval >24 months Birth order >3 Mother age <18 or >34 years Delivery at health facility Post-partum vitamin A 52.3 (50.2–54.4) 51.9 (49.9–53.9) 56.0 (54.2–57.8) 17.4 (16.3–18.6) 69.4 (67.7–70.9) 19.1 (17.8–20.4) 40.1 (37.7–42.6) 14.2 (12.8–15.7) 3,972 3,972 5,949 5,949 11,611 5,949 5,949 3,972 57.6 (56.4–58.8) 51.1 (49.4–52.7) 48.7 (47.5–49.9) 13.1 (12.5–13.8) 64.3 (63.3–65.4) 17.9 (17.1–18.7) 61.5 (60.0–62.9) 54.0 (52.1–55.8) 14,945 7,161 20,964 20,964 39,997 20,964 20,850 7,160 5.3 (2.8–7.8) -0.8 (-3.5–1.8) -7.3 (-9.6– -5.0) -4.2 (-5.6– -2.9) -5.0 (-7.1– -2.9) -1.2 (-2.7–0.4) 21.4 (18.1–24.6) 39.8 (37.2–42.1) <0.001 0.552 0.001 <0.001 <0.001 0.138 <0.001 <0.001 Note: ANC – antenatal care; CI – confidence interval; p-values <0.05 are considered significantly different; percentage point change calculated as difference between percentage in 2014 (endline) and percentage in 2003 (baseline). *Births with avoidable fertility risk include children born to mothers <18 years or >34 years, birth order >3 or born <24 months after the preceding birth. Source: Kenya Demographic and Health Surveys (KDHS) 6.3.2 Child health and Immunization Childhood immunizations and maternal tetanus toxoid immunization are associated with substantial reductions in childhood mortality; hence, great emphasis has been placed on increasing their coverage (McGovern and Canning, 2015). Kenya adopted pentavalent vaccine (i.e., hepatitis B, H. influenzae type B (Hib), diphtheria, pertussis, and tetanus) in 2001. Pneumococcal vaccine was added to the expanded program on immunization schedule in 2012, and rotavirus vaccine was added in 2014 (Ministry of Health, 2013b). The basic childhood immunization coverage indicators improved significantly by between 9 to 18 percentage points between 2003 and 2014, as did the proportion of fully-immunized children, which increased by 22 percentage points in the same period (Table 6.4). Maternal and child undernutrition resulting in intrauterine growth restriction, stunting, and severe wasting is an important underlying cause of child morbidity and mortality (Black et al., 2008). Micronutrient deficiencies, in particular vitamin A and zinc deficiencies, significantly contribute to disease burden in childhood. Early initiation and exclusive breastfeeding are important and cost￾effective child survival interventions that reduce neonatal, infant, and child mortality (Bhutta et al., 2008, Bhutta and Labbok, 2011). The proportion of exclusively breastfed children under 6 months increased almost five-fold from 13 percent in 2003 to 62 percent in 2014. Among anthropometric indices, stunting and underweight declined by four and three percentage points, respectively, from 2003 to 2014. Vitamin A supplementation for children ages 6 to 59 months increased by 38 percentage points from 33 percent in 2003 to 71 percent in 2014 (Table 6.4). 66 Table 6.4: Summary of child health indicators in Kenya, 2003–2014 Indicator KDHS 2003 KDHS 2014 Percentage point change (95% CI) p-value %(95% CI) n %(95% CI) n Bacille Calmette-Guerin vaccine 87.4 (84.3–90.0) 1,003 97.1 (96.3–97.7) 3,762 9.7 (6.5–12.8) <0.001 Pentavalent vaccine, dose 3* 72.8 (68.9–76.3) 1,003 90.6 (89.2–91.9) 3,762 17.9 (13.6–22.1) <0.001 Polio vaccine, dose 3* 72.7 (69.0–76.4) 1,003 90.6 (89.2–91.9) 3,762 17.8 (13.5–22.1) <0.001 Measles vaccine, dose 1* 72.4 (68.5–75.9) 1,003 87.0 (85.6–88.3) 3,762 14.6 (10.5–18.8) <0.001 Fully-immunized child* 59.9 (56.1–63.6) 1,003 81.7 (80.0–83.3) 3,762 21.8 (17.3–26.3) <0.001 Vitamin A 33.3 (31.0–35.7) 3,972 71.4 (70.2–72.5) 18,221 38.0 (35.3–40.8) <0.001 Stunting 29.4 (27.7–31.1) 5,949 25.8 (24.8–26.8) 18,656 -3.6 (-5.5– -1.6) <0.001 Underweight 18.4 (12.4–15.1) 5,949 10.4 (10.0–11.4) 18,656 -3.1 (-4.6– -1.5) <0.001 Wasting 5.1 (4.3–6.0) 5,949 4.1 (3.6–4.5) 18,656 -1.0 (-2.0–0.0) 0.047 Early initiation of breastfeeding 51.3 (48.3–54.2) 2,195 63.1 (60.9–65.4) 3,652 11.9 (8.1–15.6) <0.001 Exclusive breastfeeding of infants <6 months 12.7 (9.5–16.8) 597 61.7 (57.5–65.8) 852 49.0 (43.4–54.6) <0.001 Complementary feeding of infants 6-9 months 87.2 (81.4–91.4) 393 82.1 (78.4–85.3) 639 -5.1 (-10.8–0.68) 0.084 Note: CI – confidence interval; p-values <0.05 are considered significantly different; *immunization coverage for children 12–23 months; percentage point change calculated as difference between percentage in 2014 (endline) and percentage in 2003 (baseline). Source: Kenya Demographic and Health Surveys (KDHS) 6.4 Summary of Contextual Factors Kenya experienced many positive cross-sectoral developments associated with improved child survival during the evaluation period as summarized in Table 6.5. Between 2003 and 2015, the most important indicators of socio-economic, maternal and child health improved significantly in Kenya. Many of the fundamental and contextual factors associated with child survival are as important as malaria prevention and control interventions, including GDP growth, access to drinking water, exclusive breastfeeding, vitamin A supplementation, full vaccination, and deliveries in a health facility (Table 6.5). 67 Table 6.5: Summary of evidence associated with all-cause child mortality in Kenya, 2003 – 2015 Evidence supporting reduction in mortality Evidence supporting no change in mortality Evidence supporting increase in mortality Other contextual determinants Fundamental • • • Improved GDP per capita PPP ($2,146 in 2003 to $2,818 in 2014, a 31 percent increase) Improved maternal education and literacy (16 and 10 percentage points, respectively) Improved household attributes and asset ownership (29, 20 and 73 percentage point increases in access to improved drinking water source, electricity and telephone, respectively) • • • Cyclical rainfall and temperature variation No change in proportion of married women Limited improvement in household access to improved sanitation (3 percentage point increase) • None Proximate • • • • • • Increased births at health facilities (21 percentage points) Improved full vaccination (22 percentage point increase) Improved pentavalent and polio vaccination (18 percentage point increase each) Improved measles vaccination (15 percentage point increase) Improved vitamin A supplementation (38 percentage point increase) Early initiation and exclusive breastfeeding for infants <6 months of age(12 and 49 percentage point increases, respectively) • • • • • Limited improvement in four or more ANC visits (5 percentage point increase) Limited improvement in births with avoidable fertility risks (7 percentage point decrease) Limited improvements in stunting, underweight and wasting (4, 3 and 1 percentage point increases, respectively) No change in two doses of tetanus toxoid No change in complementary feeding of children ages 6–9 months • Negative trend in birth interval >24 months (4 percentage point decrease) Note: ANC – antenatal care; GDP – gross domestic product; IPTp – intermittent preventive treatment in pregnancy; ITN – insecticide-treated net; PPP –purchasing power parity 68 7 Additional Analyses to Bolster the Evaluation Design 7.1 Case Studies 7.1.1 Kilifi Health and Demographic Surveillance System data 7.1.1.1 Brief description of the data The Kilifi HDSS study area covers a population of approximately 260,000 people living in an area of 891 square km (Scott et al., 2012). A detailed description of the morbidity data has recently been published (Mogeni et al., 2016). Demographic data and clinical history for all children <15 years of age admitted to the pediatric ward at the Kilifi County Hospital between 2000 and 2015 were assembled directly from hospital registers. Screening for malaria parasites using microscopy has been continuous since the establishment of the inpatient pediatric surveillance system at the hospital. 7.1.1.2 Analytical approach Descriptive data are presented for total inpatient pediatric malaria admissions and total under all-cause child mortality over time. Malaria slide-positive deaths were analysed per 1,000 person years as estimated from the longitudinal follow up of cases in the HDSS. 7.1.1.3 Results Morbidity The data from Kilifi County Hospital, in the coastal-endemic zone, show an almost threefold reduction in inpatient malaria cases in 2015 compared to the 2000 baseline. The data show a rise in inpatient malaria cases in 2014 and 2015, but the number of cases is substantially less than the number reported annually in the period of 2000 to 2004 (Figure 7.1). Figure 7.1: Trends in inpatient malaria cases among children under 15 years of age at Kilifi County Hospital, in the coastal-endemic zone of Kenya, 2000–2015 69 Source: Kenya Medical Research Institute-Wellcome Trust Health and Demographic Surveillance System, Kilifi County, Kenya. Mortality Data from the Kilifi County HDSS was based on inpatient records and showed a declining trend in ACCM from 2003 to 2014 (Mogeni et al., 2016). Malaria-specific mortality among inpatients remained relatively stable throughout the evaluation period (Figure 7.2). Figure 7.2: Trends in all-cause and malaria-specific mortality among children under 5 years of age in Kilifi County Health and Demographic Surveillance System, 1990–2014 Note: PY – person years Source: Kenya Medical Research Institute-Wellcome Trust Health and Demographic Surveillance System, Kilifi County, Kenya. Mortality data is from inpatient hospital records only. 7.1.1.4 Conclusion • Overall, the annual number of inpatient malaria cases declined almost threefold between 2000 and 2015. • Overall, there was a declining trend in the absolute number of all-cause deaths annually from 2003 to 2015. • The annual malaria-specific mortality rate remained relatively stable from 2003–2014. 7.1.2 Siaya HDSS data 7.1.2.1 Brief description of the data Data were collected within the Kenya Medical Research Institute (KEMRI) and Centers for Disease Control and Prevention (CDC) HDSS in Siaya County, western Kenya. Briefly, the HDSS area covers a population of approximately 223,000 people residing in three sub-counties of Siaya County, spread over approximately 700 square km along the shores of Lake Victoria Adazu et al., 2005; Odhiambo et al., 2012). Mortality data were collected from the community and at two inpatient health 70 1500 malaria parasitemia 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2000 60 Number admitted with 50 40 % admitted with malaria parasitemia 1000 30 20 500 10 0 0 facilities, Siaya County Hospital and Lwak Mission Hospital. From 2003 to 2008, the cause of death was assigned by physician review of verbal autopsy data, and from 2009 onwards, a computer algorithm, interpretation of verbal autopsy (inter-VA), assigned cause of death from verbal autopsy data. Malaria morbidity was collected from annual community-based cross-sectional surveys during peak transmission periods. The presence of parasitemia was confirmed via microscopy for all participants. Data through 2012 were available for analysis. 7.1.2.2 Analytical approach For morbidity, the prevalence of malaria parasitemia confirmed through microscopy was stratified by the following age groups: <5 years of age, ages 5–14 years, and ages ≥15 years. All-cause and malaria-specific deaths were summarized per year and by age groups and incidence rates were calculated per 1,000 person years. 7.1.2.3 Results Morbidity In the two Siaya (Figure 7.3) hospitals, there was an overall decrease in cases among children <5 years of age from 2003, with a rise in 2008 followed by a slow decline annually to the lowest number reported in 2012. Figure 7.3: Trends in inpatient malaria cases among children <5 years of age in two hospitals in Siaya County Health and Demographic Surveillance System in Kenya, 2003– 2012 Note: HDSS – Health and Demographic Surveillance System Source: Kenya Medical Research Institute and U.S. Centers for Disease Control and Prevention Health and Demographic Surveillance System, Siaya County, Kenya. Mortality The Siaya County HDSS site in the lake-endemic zone showed a declining trend in all￾cause and malaria-specific mortality in children under 5 years from 2004 to 2012 (Figure 7.4). From 2003–2012 in children under 5 years, the malaria-specific mortality rate decreased from 13.2 to 3.7 per 1,000 person-years; the declines were greatest in the first 3 years of life (Desai et al., 2014). There was an increase in ACCM in 2008 subsequent to social and healthcare system disruptions following the 2007 post￾71 election violence and population displacement, which remained elevated during the first 8 months of 2009 (Hamel et al., 2011). 72 Figure 7.4: Trends in all-cause and malaria-specific mortality in children under 5 years of age in Siaya County Health and Demographic Surveillance System, 2003–2012 Note: PY – person years Source: Kenya Medical Research Institute and U.S. Centers for Disease Control and Prevention Health and Demographic Surveillance System, Siaya County, Kenya. Mortality results based on verbal autopsy data. 7.1.2.4 Conclusion • In two Siaya County hospitals, the overall absolute number of inpatients under 5 years of age with malaria parasitemia declined from 2003 to 2012. • In the Siaya County HDSS site in the lake-endemic zone, both all-cause and malaria-specific mortality in children under 5 years of age declined from 2003 to 2012. 7.2 Multivariable Analysis 7.2.1 Kaplan-Meier survival analysis 7.2.1.1 Data The full birth history data from the KDHS 2014 was transformed into 10-year retrospective longitudinal data reflecting individual child observations from birth until the date of the survey or, in the unfortunate event, the death of the child. 7.2.1.2 Analytical approach Kaplan-Meier survival estimates were calculated and comparisons made for survival probability of children ages 0 to 59 months before and after the expansion of malaria control interventions. The outcome variable was defined as the age at which a child dies or the age at interview for those who survived. A dichotomous variable (coded 1 if the child died and 0 if not) was used to define the censoring status. 7.2.1.3 Results Child survival improved during the evaluation period. At each age interval up to 5 years, the probability of surviving was higher during the period 2010–2014, after the 73 increased coverage of malaria interventions, compared with the period 2005–2009, during intervention expansion, and 2000–2004, before the malaria intervention expansion (Figure 7.5). Similar trends were observed in the lake-endemic, coastal￾endemic, and seasonal-transmission zones. In the highland epidemic-prone zone, survival was lower in the period 2000–2004 but not significantly different in periods 2005–2009 and 2010–2014. In the low-risk zone, survival is higher in the period 2005– 2009 compared to 2000–2004 and 2010–2014 (Annex 5). Figure 7.5: Kaplan-Meier survival curves for children under 5 years of age by malaria intervention expansion period in Kenya, 2000–2014 Source: Kenya Demographic and Health Survey, 2014. 7.2.1.4 Conclusion There were notable improvements in child survival, especially during the 2010–2014 period following the expansion of malaria prevention and control interventions nationally. Based on the observations across malaria-endemicity zones, particularly in the lake- and coastal-endemic zones which benefited the most from expansion, malaria control interventions might have contributed to these gains. 7.2.2 Regression analysis: Cox model 7.2.2.1 Data The full birth history data from the KDHS 2014 was transformed into 10-year retrospective longitudinal data reflecting individual child observations from birth until the date of the survey or, in the unfortunate event, the death of the child. 7.2.2.2 Analytical approach Cox’s proportional hazards models were used for the regression analysis, since they do not need specification of the form of the distribution of the baseline hazard rate (Cox, 1972; Cox & Oakes, 1984; Blossfeld et al., 1989). The models also allow for use of time-varying covariates: that is, characteristics whose status may change over time. When the hazard ratio is greater than one, there is a higher risk of mortality in the corresponding category as compared with the reference category. Conversely, the 74 risk of dying is lower when the hazard ratio is less than one. The hazard rate in the Cox model is computed as: h(t / zj) = h0(t).exp(ßj zj(t)) where the regression coefficients are to be estimated from the data. The term h0(t) is the baseline hazard function (the hazard when z=0), zj(t) is the individual covariates vector and ßj is a vector of the regression parameters that indicates the effects of these covariates, some of them varying with t (hence, the term time-varying covariate). The relative hazards are given by exp(zj(t)ß). For each malaria-endemicity zone, a model was developed that included all children ages 0 to 59 months for the 2-year period 2013–2014 of exposure to ITN ownership to account for recall bias related to duration of ITN ownership. During the survey, household heads were asked if they owned a net, and “How many months ago did your household get the mosquito net?” Respondents were supposed to indicate the precise number of months, if less the 36 months. To identify an individual child’s exposure to an ITN, data on the duration of ownership of ITNs and the time of retreatment of nets (if any) was used to construct a time-varying variable of ITN ownership for up to 2 years before the survey. For each malaria-endemicity zone, the analysis time is measured in months from the beginning of the period (i.e., 2013 for the 2-year period) allowing the introduction of age as a covariate in the analysis. Each child is observed from the beginning of the observation period until right truncation by loss to follow up, death, or date of the survey. A dichotomous variable (coded 1 if the child died and 0 if the child is alive) is used to define the censoring status. The Cox proportional hazards model assessed the relationship between household ITN ownership and child mortality (deaths of children age 1 to 59 months) over the 24 months preceding the survey in each malaria-endemicity zone. The model was adjusted for several covariates including child’s age (month), child’s sex, mother’s age (year) at birth, mother’s education level, parity, household wealth quintiles, and place of residence because these co-variates are likely to be associated with both mortality and household ownership of ITN. 7.2.2.3 Results Table 7.1 presents the results of the Cox proportional hazard model for Kenya and by malaria-endemicity zone. Nationally, household ownership of at least one ITN significantly reduces the risk of mortality among children aged 1 to 59 months by 56% (hazard ratio [HR]=0.44, 95% CI: 0.30–0.63) during the 24-month period before the survey. The lake-endemic zone contributes significantly to the national reduction in mortality risk associated with ITN ownership. In the lake-endemic zone, in households with at least one ITN, the mortality risk for children aged 1 to 59 months was reduced by 72% (HR=0.28, 95% CI: 0.14–0.55). Similar protective effects of ITN ownership were observed in the other malaria-endemicity zones; however, the results were not statically significant. 75 Among children ages 1 to 59 months, mortality risk decreased with age, with statistically significant mortality reductions observed for the 12–23 months and ≥24 month age groups. In all epidemiological zones, mortality risk among children aged >6 months was lower compared to children aged 1 to 5 months, but the hazard ratios and statistical significance varied by zone. In each epidemiological zone, mortality risk was also higher among children born to mothers aged ≥30 years. There were significant associations between mortality risk in children ages 1 to 59 months and child’s sex, mother’s age at birth, mother’s education, wealth quintile or place of residence. 76 77 Table 7.1. Effect of household ownership of at least one insecticidal bed net nationally and by malaria-endemicity zone in Kenya, 2012–2014 on mortality risk among children aged 1 to 59 months Malaria-endemicity Zone Variable National Low risk Seasonal risk Epidemic-prone Coast endemic Lake endemic Predictor Household ownership of ITN HH = no ITN 1 HH ≥1 ITN 0.44 (0.30 - Covariate Child’s age (months) 1–5 (reference) 1 6–11 0.70 (0.42 - 12–23 0.54 (0.34 - ≥24 0.28 (0.18 - Sex of Child Male (reference) 1 Female 0.87 (0.65 - Mother’s age at birth (years) <20 (reference) 1 20–24 0.82 (0.50 - 25–29 1.40 (0.86 - 30–49 1.63 (1.02 - Mother’s education None (reference) 1 Primary 1.17 (0.73 - Secondary or higher 0.93 (0.54 - Wealth Quintiles Lowest (reference) 1 Second 1.04 (0.67 - Third 1.24 (0.78 - Fourth 0.89 (0.53 - Highest 0.84 (0.49 - Place of residence Urban (reference) 1 Rural 0.77 (0.56 - 0.63) 1.18) 0.86) 0.45) 1.17) 1.37) 2.26) 2.58) 1.87) 1.63) 1.63) 1.97) 1.50) 1.45) 1.05) 1 0.37 (0.13 - 1 0.73 (0.18 - 0.44 (0.13 - 0.19 (0.05 - 1 0.72 (0.29 - 1 0.84 (0.17 - 1.41 (0.29 - 1.46 (0.29 - 1 - - 1 0.27 (0.02 - 0.90 (0.12 - 0.38 (0.04 - 0.45 (0.06 - 1 0.34 (0.11 - 1.01) 3.01) 1.53) 0.65) 1.78) 4.26) 6.86) 7.50) 3.48) 6.52) 3.37) 3.16) 1.06) 1 0.54 (0.26 - 1 0.84 (0.32 - 0.61 (0.25 - 0.20 (0.08 - 1 0.94 (0.50 - 1 1.07 (0.32 - 1.65 (0.53 - 2.01 (0.66 - 1 1.14 (0.55 - 0.64 (0.21 - 1 0.86 (0.32 - 0.84 (0.31 - 1.24 (0.50 - 0.12 (0.02 - 1 0.46 (0.24 - 1.12) 2.22) 1.47) 0.51) 1.76) 3.51) 5.11) 6.19) 2.40) 1.98) 2.29) 2.29) 3.10) 0.81) 0.90) 1 0.80 (0.30 1 0.29 (0.05 0.47 (0.13 0.55 (0.18 1 0.70 (0.34 1 1.83 (0.50 1.77 (0.44 2.88 (0.78 - 1 2.07(0.26 - 1.92(0.22 - 1 0.84 (0.32 0.73 (0.24 0.98 (0.32 0.32 (0.06 1 0.52 (0.25 - 2.15) - 1.57) - 1.71) - 1.73) - 1.42) - 6.72) - 7.15) 10.58) 16.35) 17.11) - 2.20) - 2.20) - 2.98) - 1.70) - 1.10) 1 0.44 (0.18 - 1 0.86 (0.31 - 0.28 (0.09 - 0.18 (0.07 - 1 0.70 (0.35 - 1 0.57 (0.20 - 0.58 (0.21 - 0.61 (0.23 - 1 0.77 (0.33 - 1.00 (0.29 - 1 2.49 (0.82 - 4.15 (1.57 - 1.77 (0.45 - 2.79 (0.84 - 1 1.94 (0.93 - 1.07) 2.35) 0.85) 0.47) 1.43) 1.57) 1.63) 1.58) 1.81) 3.43) 7.52) 10.98) 6.90) 9.21) 4.01) 1 0.28 (0.14 - 1 0.77 (0.26 - 0.89 (0.35 - 0.34 (0.13 - 1 1.06 (0.62 - 1 0.52 (0.19 - 1.87 (0.85 - 1.88 (0.88 - 1 0.45 (0.10 - 0.26 (0.05 - 1 0.99 (0.47 - 1.11 (0.49 - 0.42 (0.13 - 2.03 (0.70 - 1 1.38 (0.74 - 0.55) 2.30) 2.23) 0.86) 1.82) 1.43) 4.10) 4.01) 1.94) 1.33) 2.07) 2.51) 1.38) 5.89) 2.58) Person-months 26,4184.70 32,933.22 72,775.24 64,148.31417 30,661.71253 62,470.30 N (deaths) 15,612 (175) 1,910 (19) 4,372 (39) 3,776 (32) 1,822 (31) 3,662 (53) Note: ITN – insecticide-treated net; HH – household; bold indicates significance at 5% level; for low-risk zone, calculated because no deaths were recorded for the reference category (i.e., no education). hazard ratio for mother’s education were not 7.2.2.4 Conclusion The Cox proportional hazard model indicates a strong association between a household’s ownership of at least one ITN and mortality risk reduction (56 percent) among children aged 1 to 59 months at the national level. The protective effect of ITN ownership on mortality risk for children ages 1 to 59 months was most pronounced in the lake-endemic zone (72 percent reduction), where the expansion of malaria prevention and control interventions was greatest during the evaluation period. The lake-endemic zone had the largest potential to benefit from increased ITN ownership and utilization because the population had the highest burden of malaria and highest ACCM rate throughout the evaluation period. 78 8 Plausibility Analysis and Conclusions The section below summarizes the changes in malaria intervention coverage, malaria￾related morbidity and ACCM in the period from 2003 to 2015 (Figure 8.1). An assessment of the plausible link between improvements in malaria intervention coverage and changes in malaria-related morbidity and ACCM while accounting for other contextual determinants of child survival in the causal pathway is described. Figure 8.1: Summary of trends in coverage of malaria control interventions, malaria￾specific morbidity, and all-cause child mortality in Kenya, 2003–2015 Note: ITN – insecticide-treated net; ACT- artemisinin-based combination therapy Source: Data from the Kenya Demographic Health Survey (KDHS) and Kenya Malaria Indicator Survey (KMIS) 8.1 Increased Access to and Use of Malaria Prevention and Control Interventions Key changes in malaria control interventions include, at the national level, significant increases in household ownership of at least one ITN from 6 percent in 2003 to 63 percent in 2015. In the lake- and coastal-endemic zones, ITN ownership increased from 12 percent and 13 percent in 2003 to 73 percent and 87 percent in 2015, respectively. Similar changes were observed for use of ITNs, with an increase of use from 5 percent in 2003 to 48 percent in 2015 nationally. In households with at least one ITN, use among children under 5 years and pregnant women increased from 66 to 79 percent and 70 to 82 percent, respectively, from 2007 to 2015. (Add the UC data here for ownership and use and include lake- and coastal-endemic zone data if available.) Although there have been no major disparities in ITN ownership and use between households in urban and rural areas, notable gaps in ITN ownership have 79 been observed between households in the lowest and highest wealth quintiles throughout the evaluation period. IRS was initially implemented as an epidemic prevention and response tool in the highland epidemic-prone zone from 2002–2009. In 2010, IRS was expanded to include parts of three lake-endemic counties for malaria burden reduction. From 2005 to 2012, IRS coverage in the targeted areas expanded from less than 100,000 housing units with less than 300,000 people protected to over 600,000 housing units with 2.4 million people protected. At the height of the IRS program in 2010, over 1.5 million housing units and 4.7 million people were protected. From 2013 to 2015, IRS was not implemented due to emerging resistance to pyrethroids, resource constraints, and program priorities. Although IRS was expected to have a substantial impact on malaria morbidity and mortality locally, national-level impact was not expected given the geographically restricted use. Since 2009, IPTp implementation has been restricted to only malaria-endemic counties (i.e., the lake- and coastal-endemic zones). In the malaria-endemic zones, IPTp2 coverage increased from 14 percent in 2007 to 56 percent in 2015. No disparities in IPTp coverage by socioeconomic status or education level were observed by 2015. 8.2 Improved Malaria Case Management The availability of malaria diagnostics increased from 55 to 97 percent in public sector health facilities from 2010 to 2015, with all of the gains observed resulting from increased availability of malaria RDTs. Concurrently, the recommended first-line medication, AL, was available in 90 percent or more of public sector health facilities from 2010 to 2015, except for during 2014. Increased availability of malaria diagnostics and medications had a positive effect on case management; healthcare worker adherence to national treatment guidelines more than doubled from 28 to 61 percent nationally from 2010 to 2015 in public sector health facilities. Nationally, the proportion of children under 5 years of age with fever who received malaria diagnostic testing was 12 percent in 2010 and increased to 39 percent in 2015. The lake- and coastal-endemic zones had the greatest improvement in malaria diagnostic testing from 11 and 18 percent in 2010 to 59 and 44 percent in 2015, respectively. Nationally, children under 5 years of age treated with an antimalarial who received the recommended first-line medication increased from 42 to 92 percent from 2003 to 2015. In the lake- and coastal-endemic zones, children under 5 years of age treated with an antimalarial who received the recommended first-line medication increased from 40 to 94 percent and from 34 to 95 percent, respectively, from 2003 to 2015. 8.3 Decline in Malaria-Related Morbidity Malaria parasitemia prevalence increased from 3 to 5 percent among children ages 6 to 59 months from 2007 to 2015. However, over the same period, the prevalence of severe anemia among children ages 6 to 59 months declined nationally from 4 to 2 percent, and in the lake- and coastal-endemic zones, the prevalence of severe anemia 80 declined from 5 to 3 percent and 8 to 2 percent, respectively. During the period of greatest expansion of malaria prevention and control interventions from 2010 to 2015, malaria parasitemia prevalence decreased from 9 to 5 percent nationally and from 33 to 17 percent in the lake-endemic zone and severe anemia decreased from 4 to 2 percent nationally and from 7 to 3 percent in the lake-endemic zone. Households in lower wealth quintiles and located in rural areas had higher prevalences of both malaria parasitemia and severe anemia. At the facility level in two health and demographic surveillance sites located in the malaria-endemic zones, there has been a decline in inpatient malaria cases during the evaluation period. At two hospitals in Siaya County, in the lake-endemic zone, the number of children under age 5 years admitted annually with malaria parasitemia decreased by 71 percent between 2003 and 2015. At Kilifi County Hospital in the coastal-endemic zone, there was an almost threefold reduction in inpatient pediatric malaria cases observed between 2003 and 2015. 8.4 Declining all-cause child mortality A 54 percent reduction in all-cause mortality among children under the age of 5 years was observed from 115 deaths per 1,000 live births in 1999–2003 to 59 deaths per 1,000 live births in 2010–2014. The highest ACCM reductions were observed in the lake-endemic (70 percent) and seasonal-transmission zones (58 percent). Because the lake-endemic zone had the highest burden of malaria and all-cause mortality, the zone also had the greatest potential to benefit from malaria prevention and control interventions. Data from the Siaya County HDSS located in the lake-endemic zone show a declining trend in all-cause and malaria-specific mortality in children under 5 years of age during the period from 2004 to 2012. Although reductions in mortality occurred in all age groups from 2003 to 2014; the largest decline was observed among children ages 1–4 years of age (65 percent). The ACCM reductions were equitable across the evaluation period. Mortality declined more in rural (58 percent) areas, among children in the poorest wealth quintile (64 percent) and among children whose mothers had no formal education (62 percent). . 8.5 Contextual Factors and the Plausibility Argument Review of fundamental and proximate contextual factors reveal a number of positive changes during the evaluation period, some of which would lead to improved child survival. During this period GDP per capita increased from $2,146 in 2003 to $2,818 in 2014. In addition, there was a notable increasing trend in GDP-PPP and an associated declining trend in mortality among children under 5 years. Significant improvements in household-level fundamental determinants of mortality were observed, including a 28.9 percentage point increase in the proportion of households with access to improved water sources and a 3.3 percentage point increase in households with access to improved sanitation. Increased access to clean water and sanitation potentially contributed to the reduction of mortality in children under 5 years of age related to diarrheal disease. Maternal health indicators were less likely to be drivers of the reduction in mortality among children under 5 years in Kenya during the evaluation period, given that there 81 were only slight changes in ANC attendance from 52 percent in 2003 to 58 percent in 2014, and the static uptake of tetanus toxoid vaccine by pregnant women. On the other hand, child health indicators have shown improvements: there was increased immunization coverage for all the key antigens; the proportion of exclusively breastfed children under 6 months rose to 62 percent in 2014 from 13 percent in 2003; vitamin A supplementation for children ages 6 to 59 months increased by 38 percentage points, from 33 percent in 2003 to 71 percent in 2014; and there was a decline in stunting and underweight by 4 and 3 percentage points, respectively. Climate variability does not appear to have substantially influenced malaria transmission over the evaluation period of 2003–2014. 8.6 Conclusion The findings demonstrate substantial progress towards expanding coverage of malaria prevention and control interventions for populations at risk of malaria. Household ownership and use of ITNs among pregnant women, children under the age of 5 years, and general household members increased nationally and particularly in targeted malaria-endemic zones. Prevention of malaria in pregnancy significantly increased. Effective case management also improved nationally and in targeted malaria-endemic zones, particularly. Nationally, the overall number of malaria cases reported via the routine health information system declined as did inpatient malaria cases at sentinel sites in both the lake-and coast-endemic zones over the evaluation period. The national prevalence of severe anemia deceased with the largest declines Estimated deaths prevented due to expansion of malaria control interventions The Lives Saved Tool (LiST) analysis modelled the effect of expanded malaria control interventions on reduction in malaria-specific mortality during the period of 2000–2015. The model is not used to provide evidence of a decline in malaria-specific mortality but rather examines the probable impact of intervention coverage. The model estimated that 37,637 deaths among children ages 0 to 59 months were prevented due to expansion of ITN ownership at the household level compared to the number of deaths prevented if no expansion had occurred since the 2000 coverage levels. The number of deaths averted per year increased steadily from 2003 to 2008, then dropped slightly in 2009 and 2010, which was the period corresponding to post-election violence that resulted in internally displaced populations and disruption of the h lth i d li t i il i observed in the malaria-endemic zones. Consequently, given the expansion of malaria prevention and control interventions and declining trends in malaria morbidity, malaria interventions very likely contributed substantially to the 54 percent reduction in ACCM measured between 2003 and 2015. Kaplan-Meier survival curves demonstrating that child survival was higher after the malaria intervention expansion and Cox proportional hazard regression analysis showing a strong protective effect of household ITN ownership on reduction of ACCM, particularly in the lake-endemic zone, provide additional evidence to support the plausibility argument that malaria-intervention expansion contributed to ACCM reductions in Kenya. 82 9 References Adazu, K., Lindblade, K., Rosen, D.H., Odhiambo, F.O., Ofware, P., Kwach, J., Van Eijk, A.M., DeCock, K., Amornkul, P., Karanja, D., Vulule, J.M., and Slutsker, L. (2005) ‘Health and demographic surveillance in rural western kenya: A platform for evaluating interventions to reduce morbidity and mortality from infectious disease’ Am. J. Trop. Med. Hyg., 73(6), pp. 1151–1158Bayoh, M. N., Mathias, D. K., Odiere, M. R., Mutuku, F. M., Kamau, L., Gimnig, J. E., Vulule, J. M., Hawley, W. A., Hamel, M. J. and Walker, E. D. (2010) 'Anopheles gambiae: historical population decline associated with regional distribution of insecticide-treated bed nets in western Nyanza Province, Kenya', Malaria journal, 9(1), pp. 1. Bayoh, M. N., Walker, E. D., Kosgei, J., Ombok, M., Olang, G. B., Githeko, A. K., Killeen, G. F., Otieno, P., Desai, M. and Lobo, N. F. (2014) 'Persistently high estimates of late night, indoor exposure to malaria vectors despite high coverage of insecticide treated nets', Parasites & vectors, 7(1), pp. 1. Bhutta, Z. A., Ahmed, T., Black, R. E., Cousens, S., Dewey, K., Giugliani, E., Haider, B. A., Kirkwood, B., Morris, S. S., Sachdev, H. P. and Shekar, M. (2008) 'What works? Interventions for maternal and child under nutrition and survival', Lancet, 371. Bhutta, Z. A. and Labbok, M. (2011) 'Scaling up breastfeeding in developing countries', The Lancet, 378(9789), pp. 378-380. Black, R. E., Allen, L. H., Bhutta, Z. A., Caulfield, L. E., De Onis, M., Ezzati, M., Mathers, C., Rivera, J., Maternal and Group, C. U. S. (2008) 'Maternal and child undernutrition: global and regional exposures and health consequences', The lancet, 371(9608), pp. 243-260. Bousema, J., Gouagna, L., Meutstege, A., Okech, B., Akim, N., Githure, J., Beier, J. C. and Sauerwein, R. (2003) 'Treatment failure of pyrimethamine‐sulphadoxine and induction of Plasmodium falciparum gametocytaemia in children in western Kenya', Tropical Medicine & International Health, 8(5), pp. 427-430. Boyle, M. H., Racine, Y., Georgiades, K., Snelling, D., Hong, S., Omariba, W., Hurley, P. and Rao-Melacini, P. (2006) 'The influence of economic development level, household wealth and maternal education on child health in the developing world', Soc Sci Med, 63(8), pp. 2242-54. Brandling-Bennett, A. D., Oloo, A. J., Watkins, W. M., Boriga, D. A., Kariuki, D. M. and Collins, W. E. (1988) 'Chloroquine treatment of falciparum malaria in an area of Kenya of intermediate chloroquine resistance', Transactions of the Royal Society of Tropical Medicine and Hygiene, 82(6), pp. 833-837. Bryce, J., Victora, C. G., Habicht, J.-P., Black, R. E., Scherpbier, R. W. and Advisors, o. b. o. t. M.-I. T. (2005) 'Programmatic pathways to child survival: results of a multi-country evaluation of Integrated Management of Childhood Illness', Health Policy and Planning, 20(suppl 1), pp. i5-i17. Bryce, J., Victora, C. G., Habicht, J.-P., Vaughan, J. P. and Black, R. E. (2004) 'The multi￾country evaluation of the integrated management of childhood illness strategy: lessons for the evaluation of public health interventions', American journal of public health, 94(3), pp. 406-415. Carneiro, I., Roca-Feltrer, A., Griffin, J. T., Smith, L., Tanner, M., Schellenberg, J. A., Greenwood, B. and Schellenberg, D. (2010) 'Age-patterns of malaria vary with severity, transmission intensity and seasonality in sub-Saharan Africa: a systematic review and pooled analysis', PLoS One, 5(2), pp. e8988. CBS, MOH and ORC Macro (1999) Kenya Demographic and Health Survey, 1998, Calverton, Maryland: Central Bureau of Statistics, Kenya Ministry of Health and ORC Macro. Desai, M., Buff, A. M., Khagayi, S., Byass, P., Amek, N., van Eijk, A., Slutsker, L., Vulule, J., Odhiambo, F. O. and Phillips-Howard, P. A. (2014) 'Age-specific malaria mortality 83 rates in the KEMRI/CDC Health and Demographic Surveillance System in Western Kenya, 2003–2010', PloS one, 9(9), pp. e106197. Fegan, G. W., Noor, A. M., Akhwale, W. S., Cousens, S. and Snow, R. W. (2007) 'Effect of expanded insecticide-treated bednet coverage on child survival in rural Kenya: a longitudinal study', The Lancet, 370(9592), pp. 1035-1039. Garner, P. and Gülmezoglu, A. M. (2006) 'Drugs for preventing malaria in pregnant women', The Cochrane Library. Githeko, A. K. and Ndegwa, W. (2001) 'Predicting malaria epidemics in the Kenyan highlands using climate data: a tool for decision makers', Global Change and Human Health, 2(1), pp. 54-63. Githinji, S., Oyando, R., Malinga, J., Ejersa, W., Soti, D., Rono, J., Snow, R.W., Buff, A.M., Noor, A.M. (2017) 'Completeness of malaria indicator data reporting via the District Health Information Software 2 in Kenya, 2011–2015'. Malaria J. 2017 Günther, I. and Fink, G. (2011) Water and sanitation to reduce child mortality: The impact and cost of water and sanitation infrastructure, Zurich, Switzerland: World Bank. Available at: http://elibrary.worldbank.org/doi/abs/10.1596/1813-9450-5618 (Accessed: May 20, 2016). Habicht, J.-P., Victora, C. and Vaughan, J. P. (1999) 'Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact', International journal of epidemiology, 28(1), pp. 10-18. Haldar, K. and Mohandas, N. (2009) 'Malaria, erythrocytic infection, and anemia', ASH Education Program Book, 2009(1), pp. 87-93. Hamel, M. J., Adazu, K., Obor, D., Sewe, M., Vulule, J., Williamson, J. M., Slutsker, L., Feikin, D. R. and Laserson, K. F. (2011) 'A Reversal in Reductions of Child Mortality in Western Kenya, 2003–2009', The American Journal of Tropical Medicine and Hygiene, 85(4), pp. 597-605. Hay, S. I., Smith, D. L. and Snow, R. W. (2008) 'Measuring malaria endemicity from intense to interrupted transmission', The Lancet infectious diseases, 8(6), pp. 369-378. Hay, S. I., Were, E. C., Renshaw, M., Noor, A. M., Ochola, S. A., Olusanmi, I., Alipui, N. and Snow, R. W. (2003) 'Forecasting, warning, and detection of malaria epidemics: a case study', The Lancet, 361(9370), pp. 1705-1706. Imam, B. M. and Koch, S. F. (2004) 'The Determinants of Infant, Child and Maternal Mortality in Sub-Saharan Africa', Journal of Studies in Economics and Econometrics, 28(2), pp. 23- 40. JHSPH (2014) Lives Saved Tool (LiST): Johns Hopkins Bloomberg School of Public Health Available at: http://livessavedtool.org/ (Accessed: October 25, 2016). Jones, G., Steketee, R. W., Black, R. E., Bhutta, Z. A. and Morris, S. S. (2003) 'The Bellagio Child Survival Study Group. How many child deaths can we prevent this year?', Lancet, 362. Kamau, L., Hunt, R. and Coetzee, M. (2002) 'Analysis of the population structure of Anopheles funestus (Diptera: Culicidae) from western and coastal Kenya using paracentric chromosomal inversion frequencies', Journal of medical entomology, 39(1), pp. 78-83. Kamau, L., Koekemoer, L. L., Hunt, R. H. and Coetzee, M. (2003) 'Anopheles parensis: the main member of the Anopheles funestus species group found resting inside human dwellings in Mwea area of central Kenya toward the end of the rainy season', Journal of the American Mosquito Control Association, 19(2), pp. 130-133. Kamau, L. and Vulule, J. M. (2006) 'Status of insecticide susceptibility in Anopheles arabiensis from Mwea rice irrigation scheme, Central Kenya', Malaria journal, 5(1), pp. 1. Kawada, H., Dida, G. O., Sonye, G., Njenga, S. M., Mwandawiro, C. and Minakawa, N. (2012) 'Reconsideration of Anopheles rivulorum as a vector of Plasmodium falciparum in 84 Western Kenya: some evidence from biting time, blood preference, sporozoite positive rate, and pyrethroid resistance', Parasites & vectors, 5(1), pp. 1. Kayode, G. A., Adekanmbi, V. T. and Uthman, O. A. (2012) 'Risk factors and a predictive model for under-5 mortality in Nigeria: evidence from Nigeria demographic and health survey', BMC Pregnancy Childbirth, 12, pp. 10. KNBS (2004) Economic Survey 2004. Nairobi: Kenya National Bureau of Statistics. KNBS (2015) Economic Survey 2015. Nairobi Kenya Naitional Bureau of Statistics. KNBS and ICF-Macro (2015) The 2014 Kenya Demographic and Health Survey Calverton, Maryland. Korenromp, E. L., Armstrong‐Schellenberg, J. R., Williams, B. G., Nahlen, B. L. and Snow, R. W. (2004) 'Impact of malaria control on childhood anaemia in Africa–a quantitative review', Tropical Medicine & International Health, 9(10), pp. 1050-1065. Lawn, J. and Kerber, K. (2006) 'Opportunities for Africas newborns: practical data policy and programmatic support for newborn care in Africa'. Lengeler, C. (2004) 'Insecticide-treated bed nets and curtains for preventing malaria', Cochrane Database Syst Rev, 2(2). Lincetto, O., Mothebesoane-Anoh, S., Gomez, P. and Munjanja, S. P. (2006) 'Antenatal care', Opportunities for Africa's newborns: Practical data, policy and programmatic support for newborn care in Africa. Geneva: The Partnership for Maternal, Newborn and Child Heatlh. Link, B. G. and Phelan, J. C. (1996) 'Understanding sociodemographic differences in health-- the role of fundamental social causes', American journal of public health, 86(4), pp. 471- 473. Machini, B., Memusi, D., Njiru, P., Kigen, S., Kimbui, R., Amboko, B., Zurovac, D., Kiptui, R., Waqo, E. Monitoring outpatient malaria case management under the 2010 diagnostic and treatment policy in Kenya: progress January 2010–December 2015. Nairobi, Kenya: Ministry of Health, February, 2016 Maes, P., Harries, A. D., Van den Bergh, R., Noor, A., Snow, R. W., Tayler-Smith, K., Hinderaker, S. G., Zachariah, R. and Allan, R. (2014) 'Can timely vector control interventions triggered by atypical environmental conditions prevent malaria epidemics? A case-study from Wajir county, Kenya', PloS one, 9(4), pp. e92386. McElroy, P. D., ter Kuile, F. O., Lal, A. A., Bloland, P. B., Hawley, W. A., Oloo, A. J., Monto, A. S., Meshnick, S. R. and Nahlen, B. L. (2000) 'Effect of Plasmodium falciparum parasitemia density on hemoglobin concentrations among full-term, normal birth weight children in western Kenya, IV. The Asembo Bay Cohort Project', The American journal of tropical medicine and hygiene, 62(4), pp. 504-512. McGovern, M. E. and Canning, D. (2015) 'Vaccination and all-cause child mortality from 1985 to 2011: Global evidence from the demographic and health surveys', American journal of epidemiology, pp. kwv125. Menendez, C., Kahigwa, E., Hirt, R., Vounatsou, P., Aponte, J. J., Font, F., Acosta, C. J., Schellenberg, D. M., Galindo, C. M. and Kimario, J. (1997) 'Randomised placebo￾controlled trial of iron supplementation and malaria chemoprophylaxis for prevention of severe anaemia and malaria in Tanzanian infants', The Lancet, 350(9081), pp. 844- 850. Mesike, G. C. and Mojekwu, J. N. (2012) 'Environmental Determinants of CHild Mortality in Nigeria', Journal of Sustainable Development, 5(1), pp. 65-75. Ministry of Health (1992) Kenya National Plan of Action for Malaria Control 1992-1997: 5 year plan and budget Nairobi: Ministry of Health Ministry of Health (1998) National Guidelines for Diagnosis and Treatment and Prevention of Malaria for Health Workers. Nairobi: Kenya: Ministry of Health Ministry of Health (2001a) Insecticide Treated Nets Strategy 2001–2006. Nairobi: Kenya: Division of Malaria Control, Ministry of Health. 85 Ministry of Health (2001b) National Malaria Strategy. Nairobi: Kenya: Ministry of Health, Division of Malaria Control. Ministry of Health (2006) National Guidelines for Diagnosis and Treatment and Prevention of Malaria for Health Workersin Kenya. Nairobi: Kenya: Ministry of Health. Ministry of Health (2013) Department of Health Information System Data. Online: Ministry of Health. Available at: https://hiskenya.org/dhis-web￾reporting/showDataSetReportForm.action (Accessed: December 2, 2015 2015). Ministry of Health (2013b) National Policy Guidelines on Immunization 2013. Nairobi: Ministry of Health. Ministry of Health (2014a) Kenya Malaria Strategy 2009-2018. Revised edn. Nairobi: Ministry of Health, National Malaria Control Programme. Ministry of Health (2014b) National Guidelines for Diagnosis and Treatment and Prevention of Malaria for Health Workers. Nairobi: Kenya: Ministry of Health. Ministry of Health (2015a) Indoor Residual Spraying Business Plan 2015-2018. Nairobi, Kenya: National Malaria Control Programme, Ministry of Health. Ministry of Health (2015b) Insecticide Resistance Management Strategy 2015-2018. Nairobi, Kenya: National Malaria Control Programme, Ministry of Health Ministry of Health (2016) The epidemiology and control profile of malaria in Kenya: reviewing the evidence to guide the future vector control. Nairobi, Kenya: National Malaria Control Programme, Ministry of Health. Mmbando, B. P., Kamugisha, M. L., Lusingu, J. P., Francis, F., Ishengoma, D. S., Theander, T. G., Lemnge, M. M. and Scheike, T. H. (2011) 'Spatial variation and socio-economic determinants of Plasmodium falciparum infection in northeastern Tanzania', Malaria Journal, 10(1), pp. 1-9. Mogeni, P., Williams, T. N., Fegan, G., Nyundo, C., Bauni, E., Mwai, K., Omedo, I., Njuguna, P., Newton, C. R. and Osier, F. (2016) 'Age, Spatial, and Temporal Variations in Hospital Admissions with Malaria in Kilifi County, Kenya: A 25-Year Longitudinal Observational Study', PLoS Med, 13(6), pp. e1002047. MOPHS (2009a) Integrated Vector Management Policy Guidelines for Kenya. Nairobi: Ministry of Public Health and Sanitation, Division of Malaria ControlMOPHS (2009b) National Malaria Strategy 2009-2017. Nairobi: Ministry of Public Health and Sanitation, Division of Malaria Control MOPHS (2010) National Guidelines for Diagnosis and Treatment and Prevention of Malaria for Health Workers. Nairobi: Kenya: Division of Malaria Control, Ministry of Public Health and Sanitation. MOPHS (2012) National Guidelines for Diagnosis and Treatment and Prevention of Malaria for Health Workers. Nairobi: Kenya: Division of Malaria Control, Ministry of Public Health and Sanitation. Mortality Task Force of Roll Back Malaria’s Monitoring & Evaluation Reference Group (2014) Guidance for Evaluating the Impact of National Malaria Control Programs in Highly Endemic Countries, Rockville, MD, USA: MEASURE Evaluation. Mosley, W. H. and Chen, L. C. (2003) 'An analythical framework for the study of child survival in developing countries', Bulletin of the World Health Organization, 81(2), pp. 140-145. Mulamba, C., Riveron, J. M., Ibrahim, S. S., Irving, H., Barnes, K. G., Mukwaya, L. G., Birungi, J. and Wondji, C. S. (2014) 'Widespread pyrethroid and DDT resistance in the major malaria vector Anopheles funestus in East Africa is driven by metabolic resistance mechanisms', PLoS One, 9(10), pp. e110058. Murphy, S. C. and Breman, J. G. (2001) 'Gaps in the childhood malaria burden in Africa: cerebral malaria, neurological sequelae, anemia, respiratory distress, hypoglycemia, and complications of pregnancy', The American journal of tropical medicine and hygiene, 64(1 suppl), pp. 57-67. 86 Mutuku, F., Bayoh, M., Hightower, A., Vulule, J., Gimnig, J., Mueke, J., Amimo, F. and Walker, E. (2009) 'A supervised land cover classification of a western Kenya lowland endemic for human malaria: associations of land cover with larval Anopheles habitats', International Journal of Health Geographics, 8(1), pp. 1-13. Nattey, C., Masanja, H. and Klipstein-Grobusch, K. (2013) 'Relationship between household socio-economic status and under-5 mortality in Rufiji DSS, Tanzania', Glob Health Action, 6, pp. 19278. Nevill, C., Some, E., Mung'Ala, V., Muterni, W., New, L., Marsh, K., Lengeler, C. and Snow, R. (1996) 'Insecticide‐treated bednets reduce mortality and severe morbidity from malaria among children on the Kenyan coast', Tropical Medicine & International Health, 1(2), pp. 139-146. NMCP 2016. Implications of Insecticide Resistance Among Malaria Vectors on Disease Burden & Transmission Intensity: KENYA: A Case Study of the Lake Victoria Malaria Endemic Region, National Malaria Control Programme, Ministry of Health NMCP, KNBS and ICF-International (2016) Kenya Malaria Inicator Survey 2015, Nairobi: Kenya and Rockville: Maryland National Malaria Control Programme (NMCP), Kenya National Bureau of Statistics (KNBS), and ICF International. Noor, A. M., Alegana, V. A., Patil, A. P. and Snow, R. W. (2010) 'Predicting the unmet need for biologically targeted coverage of insecticide-treated nets in Kenya', The American journal of tropical medicine and hygiene, 83(4), pp. 854-860. Noor, A. M., Amin, A. A., Akhwale, W. S. and Snow, R. W. (2007) 'Increasing coverage and decreasing inequity in insecticide-treated bed net use among rural Kenyan children', PLoS Med, 4(8), pp. e255. Noor, A. M., Gething, P. W., Alegana, V. A., Patil, A. P., Hay, S. I., Muchiri, E., Juma, E. and Snow, R. W. (2009) 'The risks of malaria infection in Kenya in 2009', BMC Infectious Diseases, 9(1), pp. 1. O'Hare, B., Makuta, I., Chiwaula, L. and Bar-Zeev, N. (2013) 'Income and child mortality in developing countries: a systematic review and meta-analysis', J R Soc Med, 106(10), pp. 408-14. Ochomo, E., Bayoh, N., Kamau, L., Atieli, F., Vulule, J., Ouma, C., Ombok, M., Njagi, K., Soti, D. and Mathenge, E. (2014) 'Pyrethroid susceptibility of malaria vectors in 4 districts of Western Kenya ', Parasites & Vectors 310(7). Ochomo, E. O., Bayoh, N. M., Walker, E. D., Abongo, B. O., Ombok, M. O., Ouma, C., Githeko, A. K., Vulule, J., Yan, G. and Gimnig, J. E. (2013) 'The efficacy of long-lasting nets with declining physical integrity may be compromised in areas with high levels of pyrethroid resistance', Malaria journal, 12(1), pp. 1. Odhiambo, C. O., Otieno, W., Adhiambo, C., Odera, M. M. and Stoute, J. A. (2008) 'Increased deposition of C3b on red cells with low CR1 and CD55 in a malaria-endemic region of western Kenya: implications for the development of severe anemia', BMC medicine, 6(1), pp. 1. Odhiambo, F.O., Laserson, K.F., Sewe, M., Hamel, M. J, Feikin, D.R.,Adazu, K., Ogwang, S., Obor,D., Amek, N., Bayoh, N., Ombok, M., Lindblade, K., Desai, M., Kuile,F., Phillips￾Howard, P., Van Eijk, A.M., Rosen, D., Hightower, A., Ofware, P., Muttai, H., Nahlen, B., DeCock, K., Slutsker, L., Breiman, R.F. and Vulule, J.M. (2012) ‘Profile: The KEMRI/CDC Health and Demographic Surveillance System—Western Kenya’ International Journal of Epidemiology, 41:977–987 Okiro, E. A., Al-Taiar, A., Reyburn, H., Idro, R., Berkley, J. A. and Snow, R. W. (2009) 'Age patterns of severe paediatric malaria and their relationship to Plasmodium falciparum transmission intensity', Malaria journal, 8(1), pp. 1. Okiro, E. A., Alegana, V. A., Noor, A. M. and Snow, R. W. (2010) 'Changing malaria intervention coverage, transmission and hospitalization in Kenya', Malaria journal, 9(1), pp. 1. 87 Okiro, E. A., Bitira, D., Mbabazi, G., Mpimbaza, A., Alegana, V. A., Talisuna, A. O. and Snow, R. W. (2011) 'Increasing malaria hospital admissions in Uganda between 1999 and 2009', BMC medicine, 9(1), pp. 1. Okiro, E. A., Kazembe, L. N., Kabaria, C. W., Ligomeka, J., Noor, A. M., Ali, D. and Snow, R. W. (2013) 'Childhood malaria admission rates to four hospitals in Malawi between 2000 and 2010', PloS one, 8(4), pp. e62214. Omariba, D. W. R., Beaujot, R. and Rajulton, F. (2007) 'Determinants of infant and child mortality in Kenya: An analysis controlling for frailty effects', Population Research and Policy Review, 26(3), pp. 299-321. Phillips-Howard, P. A., Ter Kuile, F. O., Nahlen, B. L., Alaii, J. A., Gimnig, J. E., Kolczak, M. S., Terlouw, D. J., Kariuki, S. K., Shi, Y. P. and Kachur, S. P. (2003) 'The efficacy of permethrin-treated bed nets on child mortality and morbidity in western Kenya II. Study design and methods', The American journal of tropical medicine and hygiene, 68(4 suppl), pp. 10-15. Rajaratnam, J. K., Marcus, J. R., Flaxman, A. D., Wang, H., Levin-Rector, A., Dwyer, L., Costa, M., Lopez, A. D. and Murray, C. J. L. (2010) 'Neonatal, postneonatal, childhood, and under-5 mortality for 187 countries, 1970?2010: a systematic analysis of progress towards Millennium Development Goal 4', The Lancet, 375(9730), pp. 1988-2008. RBM (2013) Household Survey Indicators for Malaria Control. Geneva, Switzerland: Roll Back Malaria Partnership. RBM and WHO (2000) The Abuja Declaration and the Plan of Action: Roll Back Malaria Partnership & World Health Organization. Roll Back Malaria (2005) Global Strategic Plan: Roll Back Malaria : 2005-2015. Geneva, Switzerland: RBM Partnership. Rowe, A. K., Kachur, S. P., Yoon, S. S., Lynch, M., Slutsker, L. and Steketee, R. W. (2009) 'Caution is required when using health facility-based data to evaluate the health impact of malaria control efforts in Africa', Malaria journal, 8(1), pp. 1. Rowe, A. K., Onikpo, F., Lama, M., Osterholt, D. M. and Deming, M. S. (2011) 'Impact of a malaria-control project in Benin that included the integrated management of childhood illness strategy', American journal of public health, 101(12), pp. 2333-2341. Rowe, A. K., Steketee, R. W., Arnold, F., Wardlaw, T., Basu, S., Bakyaita, N., Lama, M., Winston, C. A., Lynch, M. and Cibulskis, R. E. (2007) 'Viewpoint: Evaluating the impact of malaria control efforts on mortality in sub‐Saharan Africa', Tropical Medicine & International Health, 12(12), pp. 1524-1539. Scott, J. A. G., Bauni, E., Moisi, J. C., Ojal, J., Gatakaa, H., Nyundo, C., Molyneux, C. S., Kombe, F., Tsofa, B. and Marsh, K. (2012) 'Profile: the Kilifi health and demographic surveillance system (KHDSS)', International journal of epidemiology, 41(3), pp. 650-657. Sewe, M. O., Ahlm, C. and Rocklöv, J. (2016) 'Remotely Sensed Environmental Conditions and Malaria Mortality in Three Malaria Endemic Regions in Western Kenya', PloS one, 11(4), pp. e0154204. Shretta, R. (1999) NETMARK briefing book: Insecticide treated materials in Kenya. London: Malaria Consortium, UK, for Academy for Educational Development. Smith, D. L. and Hay, S. I. (2009) 'Endemicity response timelines for Plasmodium falciparum elimination', Malaria Journal, 8(1), pp. 1. Snow, R. W., Kibuchi, E., Karuri, S. W., Sang, G., Gitonga, C. W., Mwandawiro, C., Bejon, P. and Noor, A. M. (2015) 'Changing malaria prevalence on the Kenyan coast since 1974: climate, drugs and vector control', PLoS One, 10(6), pp. e0128792. Snow, R. W., Okiro, E. A., Gething, P. W., Atun, R. and Hay, S. I. (2010) 'Equity and adequacy of international donor assistance for global malaria control: an analysis of populations at risk and external funding commitments', The Lancet, 376(9750), pp. 1409-1416. Snow, R. W., Okiro, E. A., Noor, A. M., Munguti, K., Tetteh, G. and Juma, E. (2009) The coverage and impact of malaria intervention in Kenya 2007-2009, Nairobi: Kenya: Division of Malaria Control, Ministry of Public Health and Sanitation. 88 Snow, R. W., Omumbo, J. A., Lowe, B., Molyneux, C. S., Obiero, J.-O., Palmer, A., Weber, M. W., Pinder, M., Nahlen, B. and Obonyo, C. (1997) 'Relation between severe malaria morbidity in children and level of Plasmodium falciparum transmission in Africa', The Lancet, 349(9066), pp. 1650-1654. Spencer, H., Oloo, A. J., Watkins, W., Sixsmith, D., Churchill, F. and Koech, D. (1984) 'Amodiaquine more effective than chloroquine against Plasmodium falciparum malaria on Kenya coast', The Lancet, 323(8383), pp. 956-957. Stratton, L., O'Neill, M. S., Kruk, M. E. and Bell, M. L. (2008) 'The persistent problem of malaria: Addressing the fundamental causes of a global killer', Social Science & Medicine, 67(5), pp. 854-862. Subramanian, S., Belli, P. and Kawachi, I. (2002) 'The macroeconomic determinants of health', Annual review of public health, 23(1), pp. 287-302. UN IGME (2014) Levels & Trends and Child Mortality: United Nations Inter-agency Group for Child Mortality Estimation Available at: http://www.childmortality.org/files_v20/download/unicef-2013-child-mortality￾report-LR-10_31_14_195.pdf (Accessed: October 25, 2016). UNDP (2015) Human Development Report 2015: Work for Human Development: Briefing note for countries on the 2015 Human Development Report - Kenya, Online: UNDP. Available at: http://hdr.undp.org/sites/all/themes/hdr_theme/country-notes/KEN.pdf (Accessed: May 2, 2016). Van Bodegom, D., Eriksson, U. K., Houwing-Duistermaat, J. J. and Westendorp, R. G. (2012) 'Clustering of child mortality in a contemporary polygamous population in Africa', Biodemography Soc Biol, 58(2), pp. 162-72. Victora, C. G., Black, R. E., Boerma, J. T. and Bryce, J. (2011) 'Measuring impact in the Millennium Development Goal era and beyond: a new approach to large-scale effectiveness evaluations', The Lancet, 377(9759), pp. 85-95. Victora, C. G., Schellenberg, J. A., Huicho, L., Amaral, J., El Arifeen, S., Pariyo, G., Manzi, F., scherpbier, R. W., Bryce, J. and Habicht, J.-P. (2005) 'Context matters: interpreting impact findings in child survival evaluations', Health Policy and Planning, 20(suppl 1), pp. i18-i31. Wang, L. (2003) 'Determinants of child mortality in LDCs: empirical findings from demographic and health surveys', Health policy, 65(3), pp. 277-299. Watkins, W., Spencer, H., Kariuki, D., Sixsmith, D., Boriga, D., Kipingor, T. and Koech, D. (1984) 'Effectiveness of amodiaquine as treatment for chloroquine-resistant Plasmodium falciparum infections in Kenya', The Lancet, 323(8373), pp. 357-359. Webster, J., Lines, J., Bruce, J., Armstrong Schellenberg, J. R. M. and Hanson, K. (2005) 'Which delivery systems reach the poor? A review of equity of coverage of ever￾treated nets, never-treated nets, and immunisation to reduce child mortality in Africa', The Lancet infectious diseases, 5(11), pp. 709-717. WHO (2007a) INSECTICIDE-TREATED MOSQUITO NETS: a WHO Position Statement Geneva: Global Malaria Programme, World Health Organization. WHO (2007b) Kenya Country Office Annual Report, Nairobi: Kenya: World Health Organization, Kenya Country Office. WHO (2007c) Technical expert group meeting on intermittent preventive treatment in pregnancy (IPTp) Geneva, Switzerland: World Health Organization. WHO (2010) Guidelines for the Treatment of Malaria. Second edn. Geneva, Switzerland: World Health Organization. WHO (2012) Intermittent preventive treatment of malaria in pregnancy using sulfadoxine￾pyrimethamine (IPTp-SP), Geneva: World Health Organization. WHO (2014) 'Global Health Expenditure Database'. Available at: http://apps.who.int/nha/database/ViewData/Indicators/en (Accessed October 25, 2016). 89 WHO 2015. Fact Sheet: World Malaria Report 2015. Geneva, Switzerland: World Health Organization. World Bank (2014) World Development Indicators: Education completion and outcomes. Online: World Bank. Available at: http://wdi.worldbank.org/table/2.13 (Accessed: April 27, 2016 2016). World Bank (2016) Kenya Data: The World Bank Group. Available at: http://data.worldbank.org/country/kenya (Accessed: April 27, 2016. Worrall, E., Basu, S. and Hanson, K. (2005) 'Is malaria a disease of poverty? A review of the literature', Tropical Medicine & International Health, 10(10), pp. 1047-1059. Yé, Y., Hoshen, M., Louis, V., Séraphin, S., Traoré, I. and Sauerborn, R. (2006) 'Housing conditions and Plasmodium falciparum infection: protective effect of iron-sheet roofed houses', Malaria Journal, 5(1), pp. 1. Yé, Y., Eisele,T. P., Eckert, E., Korenromp, E., Shah, J. A., Hershey, C. L., Ivanovich, E., Newby, H., Carvajal-Velez, L., Lynch, M., Komatsu, R., Cibulskis, R. E., Moore, Z., and Bhattarai, A. (2017) 'Framework for evaluating the health impact of the scale-up of malaria control interventions on all-cause child mortality in Sub-Saharan Africa', Am. J. Trop. Med. Hyg., 97(Suppl 3), 2017, pp. 9–19. Zhou, G., Minakawa, N., Githeko, A. K. and Yan, G. (2004) 'Association between climate variability and malaria epidemics in the East African highlands', Proceedings of the National Academy of Sciences of the United States of America, 101(8), pp. 2375-2380. 90