Search

HIV estimation using data from the demographic and health surveys (DHS) is limited by the presence of non-response and test refusals. Conventional adjustments such as imputation require the data to be missing at random. Methods that use instrumental variables allow the possibility that prevalence is different between the respondents and non-respondents, but their performance depends critically on the validity of the instrument. Using Manski's partial identification approach, we form instrumental variable bounds for HIV prevalence from a pool of candidate instruments. Our method does not require all candidate instruments to be valid. We use a simulation study to evaluate and compare our method against its competitors. We illustrate the proposed method using DHS data from Zambia, Malawi and Kenya. Our simulations show that imputation leads to seriously biased results even under mild violations of non-random missingness. Using worst case identification bounds that do not make assumptions about the non-response mechanism is robust but not informative. By taking the union of instrumental variable bounds balances informativeness of the bounds and robustness to inclusion of some invalid instruments. Non-response and refusals are ubiquitous in population based HIV data such as those collected under the DHS. Partial identification bounds provide a robust solution to HIV prevalence estimation without strong assumptions. Union bounds are significantly more informative than the worst case bounds without sacrificing credibility., (© 2024 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.)

Globally, there are over 1 billion people infected with soil-transmitted helminths (STHs), mostly living in marginalized settings with inadequate sanitation in sub-Saharan Africa and Southeast Asia. The World Health Organization recommends an integrated approach to STH morbidity control through improved access to sanitation and hygiene education and the delivery of preventive chemotherapy (PC) to school-age children delivered through schools. Progress of STH control programs is currently estimated using a baseline (pre-PC) school-based prevalence survey and then monitored using periodical school-based prevalence surveys, known as Impact Assessment Surveys (IAS). We investigated whether integrating geostatistical methods with a Markov model or a mechanistic transmission model for projecting prevalence forward in time from baseline can improve IAS design strategies. To do this, we applied these 2 methods to prevalence data collected in Kenya, before evaluating and comparing their performance in accurately informing optimal survey design for a range of IAS sampling designs. We found that, although both approaches performed well, the mechanistic method more accurately projected prevalence over time and provided more accurate information for guiding survey design. Both methods performed less well in areas with persistent STH hotspots where prevalence did not decrease despite multiple rounds of PC. Our findings show that these methods can be useful tools for more efficient and accurate targeting of PC. The general framework built in this paper can also be used for projecting prevalence and informing survey design for other neglected tropical diseases., Competing Interests: Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed., (© The Author(s) 2024. Published by Oxford University Press on behalf of Infectious Diseases Society of America.)

This study presents a comprehensive analysis of historical fire and climatic data to estimate the monthly frequency of vegetation fires in Kenya. This work introduces a statistical model that captures the behavior of fire count data, incorporating temporal explanatory factors and emphasizing the predictive significance of maximum temperature and rainfall. By employing Bayesian approaches, the paper integrates literature information, simulation studies, and real-world data to enhance model performance and generate more precise prediction intervals that encompass actual fire counts. To forecast monthly fire occurrences aggregated from the Moderate Resolution Imaging Spectroradiometer (MODIS) data in Kenya (2000-2018), the study utilizes maximum temperature and rainfall values derived from global GeoTiff (.tif) files sourced from the WorldClim database. The evaluation of the widely used Negative Binomial (NB) model and the proposed Bayesian Negative Binomial (BNB) model reveals the superiority of the latter in accounting for seasonal patterns and long-term trends. The simulation results demonstrate that the BNB model outperforms the NB model in terms of Root Mean Square Error (RMSE), and Mean Absolute Scaled Error (MASE) on both training and testing datasets. Furthermore, when applied to real data, the Bayesian Negative Binomial model exhibits better performance on the test dataset, showcasing lower RMSE (163.22 vs. 166.67), lower MASE (1.12 vs. 1.15), and reduced bias (-2.52% vs. -2.62%) compared to the NB model. The Bayesian model also offers prediction intervals that closely align with actual predictions, indicating its flexibility in forecasting the frequency of monthly fires. These findings underscore the importance of leveraging past data to forecast the future behavior of the fire regime, thus providing valuable insights for fire control strategies in Kenya. By integrating climatic factors and employing Bayesian modeling techniques, the study contributes to the understanding and prediction of vegetation fires, ultimately supporting proactive measures in mitigating their impact., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2024 Orero et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Background: Statistical modeling suggests that decreasing diarrhea-associated mortality rates in recent decades are largely attributed to improved case management, rotavirus vaccine, and economic development., Methods: We examined data collected in 2 multisite population-based diarrhea case-control studies, both conducted in The Gambia, Kenya, and Mali: the Global Enteric Multicenter Study (GEMS; 2008-2011) and Vaccine Impact on Diarrhea in Africa (VIDA; 2015-2018). Population-level diarrhea mortality and risk factor prevalence, estimated using these study data, were used to calculate the attribution of risk factors and interventions for diarrhea mortality using a counterfactual framework. We performed a decomposition of the effects of the changes in exposure to each risk factor between GEMS and VIDA on diarrhea mortality for each site., Results: Diarrhea mortality among children under 5 in our African sites decreased by 65.3% (95% confidence interval [CI]: -80.0%, -45.0%) from GEMS to VIDA. Kenya and Mali had large relative declines in diarrhea mortality between the 2 periods with 85.9% (95% CI: -95.1%, -71.5%) and 78.0% (95% CI: -96.0%, 36.3%) reductions, respectively. Among the risk factors considered, the largest declines in diarrhea mortality between the 2 study periods were attributed to reduction in childhood wasting (27.2%; 95% CI: -39.3%, -16.8%) and an increased rotavirus vaccine coverage (23.1%; 95% CI: -28.4%, -19.4%), zinc for diarrhea treatment (12.1%; 95% CI: -16.0%, -8.9%), and oral rehydration salts (ORS) for diarrhea treatment (10.2%)., Conclusions: The VIDA study sites demonstrated exceptional reduction in diarrhea mortality over the last decade. Site-specific differences highlight an opportunity for implementation science in collaboration with policymakers to improve the equitable coverage of these interventions globally., Competing Interests: Potential conflicts of interest. K. L. K. reports consultation fees and travel support from PATH and the University of Washington related to diarrheal diseases and grant support to her institution from the National Institutes of Health, Institut Pasteur, and the Bill & Melinda Gates Foundation. E. L. D. reports a T32 grant from the National Institutes of Health for training through the University of Maryland School of Medicine Center for Vaccine Development and Global Health (T32AI007524). M. D. T. reports support for attending an investigator meeting from the Bill & Melinda Gates Foundation. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed., (© The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America.)

It is common in longitudinal studies for scheduled visits to be accompanied by as-needed visits due to medical events occurring between scheduled visits. If the timing of these as-needed visits is related to factors that are associated with the outcome but are not among the regression model covariates, naively including these as-needed visits in the model yields biased estimates. In this paper, the authors illustrate and discuss the key issues pertaining to inverse intensity rate ratio (IIRR)-weighted generalized estimating equations (GEE) methods in the context of a study of Kenyan mothers infected with human immunodeficiency virus type 1 (1999–2005). The authors estimated prevalences and prevalence ratios for morbid conditions affecting the women during a 1-year postpartum follow-up period. Of the 484 women under study, 62% had at least 1 as-needed visit. Use of a standard GEE model including both scheduled and unscheduled visits predicted a pneumonia prevalence of 2.9% (95% confidence interval: 2.3%, 3.5%), while use of the IIRR-weighted GEE predicted a prevalence of 1.5% (95% confidence interval: 1.2%, 1.8%). The estimate obtained using the IIRR-weighted GEE approach was compatible with estimates derived using scheduled visits only. These results highlight the importance of properly accounting for informative follow-up in these studies. [ABSTRACT FROM PUBLISHER]

As countries decide on vaccination strategies and how to ease movement restrictions, estimating the proportion of the population previously infected with SARS-CoV-2 is important for predicting the future burden of COVID-19. This proportion is usually estimated from serosurvey data in two steps: first the proportion above a threshold antibody level is calculated, then the crude estimate is adjusted using external estimates of sensitivity and specificity. A drawback of this approach is that the PCR-confirmed cases used to estimate the sensitivity of the threshold may not be representative of cases in the wider population-e.g., they may be more recently infected and more severely symptomatic. Mixture modelling offers an alternative approach that does not require external data from PCR-confirmed cases. Here we illustrate the bias in the standard threshold-based approach by comparing both approaches using data from several Kenyan serosurveys. We show that the mixture model analysis produces estimates of previous infection that are often substantially higher than the standard threshold analysis., (© 2021. The Author(s).)

Introduction: To make valid comparisons across groups, a measurement instrument needs to be measurement invariant across those groups. The present study evaluates measurement invariance for experience of violence among adolescent girls and young women (AGYW) in two informal settlements in Nairobi, Kenya., Methods: We used survey data collected from 1,081 AGYW aged 15-22 years from two Nairobi's informal settlements of Korogocho (n = 617) and Viwandani (n = 464) in 2017 through DREAMS (an initiative aimed at reducing HIV incidence among AGYW with a core package of evidence-based interventions) impact evaluation project. Experience of violence was measured using the 15-item WHO's violence against women instrument, and factorial (non)invariance assessed within exploratory structural equation modeling (ESEM) framework. Cross-group measurement invariance was assessed using Bayesian Multiple Indicator Multiple Causes (MIMIC) model across site, age groups, self-reported invitation to participate in DREAMS, marital status, currently in school, education level, religion, ethnic groups, ever had sex, slept hungry at night past 4 weeks, and wealth index., Results: The mean and median ages of the AGYW were 17.9 years and 17 years, respectively. About 59% reported having had sex and 58% of AGYW were in school. The percentage reporting each act of violence varied from 1.6% ("attacked you with a weapon") to 26.5% ("insult you or make you feel bad about yourself"). About 44% (n = 474) of participants experienced ≥1 acts of violence, and 2.7% (n = 29) experienced at least half of the 15 acts. The structure underlying the 15 items was configurally similar to that proposed by WHO, with three factors reflecting either psychological, physical, or sexual violence. Noninvariance was detected for five items-spread across the three domains. Three of five items showed noninvariance only for sleeping hungry at night in the past 4 weeks. As the majority of items did not show evidence of noninvariance, differences in latent mean scores likely reflect actual differences and may not be attributable to measurement artifacts., Conclusions: Using state-of-the-art statistical techniques on a widely used instrument for measuring exposure to violence among women, this study provides support for the subscales of psychological, physical and sexual violence in a Kenyan AGYW population. The instrument supports comparisons across groups within this population. This is crucial when comparing violence against girls/women prevalence rates and to understand challenges and exchange strategies to reduce abuse or violence experienced by AGYW, or women in general., Competing Interests: The authors have declared that no competing interests exist.

This study assessed the performance of modeling approaches to estimate personal exposure in Kenyan homes where cooking fuel combustion contributes substantially to household air pollution (HAP). We measured emissions (PM 2.5 , black carbon, CO); household air pollution (PM 2.5 , CO); personal exposure (PM 2.5 , CO); stove use; and behavioral, socioeconomic, and household environmental characteristics (eg, ventilation and kitchen volume). We then applied various modeling approaches: a single-zone model; indirect exposure models, which combine person-location and area-level measurements; and predictive statistical models, including standard linear regression and ensemble machine learning approaches based on a set of predictors such as fuel type, room volume, and others. The single-zone model was reasonably well-correlated with measured kitchen concentrations of PM 2.5 (R 2  = 0.45) and CO (R 2  = 0.45), but lacked precision. The best performing regression model used a combination of survey-based data and physical measurements (R 2  = 0.76) and a root mean-squared error of 85 µg/m 3 , and the survey-only-based regression model was able to predict PM2.5 exposures with an R 2 of 0.51. Of the machine learning algorithms evaluated, extreme gradient boosting performed best, with an R 2 of 0.57 and RMSE of 98 µg/m 3 ., (© 2021 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Understanding seasonal human mobility at subnational scales has important implications across sciences, from urban planning efforts to disease modelling and control. Assessing how, when, and where populations move over the course of the year, however, requires spatially and temporally resolved datasets spanning large periods of time, which can be rare, contain sensitive information, or may be proprietary. Here, we aim to explore how a set of broadly available covariates can describe typical seasonal subnational mobility in Kenya pre-COVID-19, therefore enabling better modelling of seasonal mobility across low- and middle-income country (LMIC) settings in non-pandemic settings. To do this, we used the Google Aggregated Mobility Research Dataset, containing anonymized mobility flows aggregated over users who have turned on the Location History setting, which is off by default. We combined this with socioeconomic and geospatial covariates from 2018 to 2019 to quantify seasonal changes in domestic and international mobility patterns across years. We undertook a spatiotemporal analysis within a Bayesian framework to identify relevant geospatial and socioeconomic covariates explaining human movement patterns, while accounting for spatial and temporal autocorrelations. Typical pre-pandemic mobility patterns in Kenya mostly consisted of shorter, within-county trips, followed by longer domestic travel between counties and international travel, which is important in establishing how mobility patterns changed post-pandemic. Mobility peaked in August and December, closely corresponding to school holiday seasons, which was found to be an important predictor in our model. We further found that socioeconomic variables including urbanicity, poverty, and female education strongly explained mobility patterns, in addition to geospatial covariates such as accessibility to major population centres and temperature. These findings derived from novel data sources elucidate broad spatiotemporal patterns of how populations move within and beyond Kenya, and can be easily generalized to other LMIC settings before the COVID-19 pandemic. Understanding such pre-pandemic mobility patterns provides a crucial baseline to interpret both how these patterns have changed as a result of the pandemic, as well as whether human mobility patterns have been permanently altered once the pandemic subsides. Our findings outline key correlates of mobility using broadly available covariates, alleviating the data bottlenecks of highly sensitive and proprietary mobile phone datasets, which many researchers do not have access to. These results further provide novel insight on monitoring mobility proxies in the context of disease surveillance and control efforts through LMIC settings., (© 2021. The Author(s).)

According to the Kenya National School-Based Deworming program launched in 2012 and implemented for the first 5 years (2012-2017), the prevalence of soil-transmitted helminths (STH) and schistosomiasis substantially reduced over the mentioned period among the surveyed schools. However, this reduction is heterogeneous. In this study, we aimed to determine the factors associated with the 5-year school-level infection prevalence and relative reduction (RR) in prevalence in Kenya following the implementation of the program. Multiple variables related to treatment, water, sanitation, and hygiene (WASH) and environmental factors were assembled and included in mixed-effects linear regression models to identify key determinants of the school location STH and schistosomiasis prevalence and RR. Reduced prevalence of Ascaris lumbricoides was associated with low (< 1%) baseline prevalence, seven rounds of treatment, high (50-75%) self-reported coverage of household handwashing facility equipped with water and soap, high (20-25°C) land surface temperature, and community population density of 5-10 people per 100 m2. Reduced hookworm prevalence was associated with low (< 1%) baseline prevalence and the presence of a school feeding program. Reduced Trichuris trichiura prevalence was associated with low (< 1%) baseline prevalence. Reduced Schistosoma mansoni prevalence was associated with low (< 1%) baseline prevalence, three treatment rounds, and high (> 75%) reported coverage of a household improved water source. Reduced Schistosoma haematobium was associated with high aridity index. Analysis indicated that a combination of factors, including the number of treatment rounds, multiple related program interventions, community- and school-level WASH, and several environmental factors had a major influence on the school-level infection transmission and reduction.

Background: Anthrax is an important zoonotic disease in Kenya associated with high animal and public health burden and widespread socio-economic impacts. The disease occurs in sporadic outbreaks that involve livestock, wildlife, and humans, but knowledge on factors that affect the geographic distribution of these outbreaks is limited, challenging public health intervention planning., Methods: Anthrax surveillance data reported in southern Kenya from 2011 to 2017 were modeled using a boosted regression trees (BRT) framework. An ensemble of 100 BRT experiments was developed using a variable set of 18 environmental covariates and 69 unique anthrax locations. Model performance was evaluated using AUC (area under the curve) ROC (receiver operating characteristics) curves., Results: Cattle density, rainfall of wettest month, soil clay content, soil pH, soil organic carbon, length of longest dry season, vegetation index, temperature seasonality, in order, were identified as key variables for predicting environmental suitability for anthrax in the region. BRTs performed well with a mean AUC of 0.8. Areas highly suitable for anthrax were predicted predominantly in the southwestern region around the shared Kenya-Tanzania border and a belt through the regions and highlands in central Kenya. These suitable regions extend westwards to cover large areas in western highlands and the western regions around Lake Victoria and bordering Uganda. The entire eastern and lower-eastern regions towards the coastal region were predicted to have lower suitability for anthrax., Conclusion: These modeling efforts identified areas of anthrax suitability across southern Kenya, including high and medium agricultural potential regions and wildlife parks, important for tourism and foreign exchange. These predictions are useful for policy makers in designing targeted surveillance and/or control interventions in Kenya. We thank the staff of Directorate of Veterinary Services under the Ministry of Agriculture, Livestock and Fisheries, for collecting and providing the anthrax historical occurrence data., Competing Interests: The authors have declared that no competing interests exist.

The spatial dynamics of epidemics are fundamentally affected by patterns of human mobility. Mobile phone call detail records (CDRs) are a rich source of mobility data, and allow semi-mechanistic models of movement to be parameterised even for resource-poor settings. While the gravity model typically reproduces human movement reasonably well at the administrative level spatial scale, past studies suggest that parameter estimates vary with the level of spatial discretisation at which models are fitted. Given that privacy concerns usually preclude public release of very fine-scale movement data, such variation would be problematic for individual-based simulations of epidemic spread parametrised at a fine spatial scale. We therefore present new methods to fit fine-scale mathematical mobility models (here we implement variants of the gravity and radiation models) to spatially aggregated movement data and investigate how model parameter estimates vary with spatial resolution. We use gridded population data at 1km resolution to derive population counts at different spatial scales (down to ∼ 5km grids) and implement mobility models at each scale. Parameters are estimated from administrative-level flow data between overnight locations in Kenya and Namibia derived from CDRs: where the model spatial resolution exceeds that of the mobility data, we compare the flow data between a particular origin and destination with the sum of all model flows between cells that lie within those particular origin and destination administrative units. Clear evidence of over-dispersion supports the use of negative binomial instead of Poisson likelihood for count data with high values. Radiation models use fewer parameters than the gravity model and better predict trips between overnight locations for both considered countries. Results show that estimates for some parameters change between countries and with spatial resolution and highlight how imperfect flow data and spatial population distribution can influence model fit., Competing Interests: The authors have declared that no competing interests exist.

While the international pet trade and habitat destruction have been extensively discussed as major threats to the survival of the pancake tortoise (Malacochersus tornieri), the impact of climate change on the species remains unknown. In this study, we used species distribution modelling to predict the current and future distribution of pancake tortoises in Zambezian and Somalian biogeographical regions. We used 224 pancake tortoise occurrences obtained from Tanzania, Kenya and Zambia to estimate suitable and stable areas for the pancake tortoise in all countries present in these regions. We also used a protected area network to assess how many of the suitable and stable areas are protected for the conservation of this critically endangered species. Our model predicted the expansion of climatically suitable habitats for pancake tortoises from four countries and a total area of 90,668.75 km2 to ten countries in the future and an area of 343,459.60-401,179.70 km2. The model also showed that a more significant area of climatically suitable habitat for the species lies outside of the wildlife protected areas. Based on our results, we can predict that pancake tortoises may not suffer from habitat constriction. However, the species will continue to be at risk from the international pet trade, as most of the identified suitable habitats remain outside of protected areas. We suggest that efforts to conserve the pancake tortoise should not only focus on protected areas but also areas that are unprotected, as these comprise a large proportion of the suitable and stable habitats available following predicted future climate change., Competing Interests: The authors have declared that no competing interests exist.

Substantial progress has been made globally to control malaria, however there is a growing need for innovative new tools to ensure continued progress. One approach is to harness genetic sequencing and accompanying methodological approaches as have been used in the control of other infectious diseases. However, to utilize these methodologies for malaria, we first need to extend the methods to capture the complex interactions between parasites, human and vector hosts, and environment, which all impact the level of genetic diversity and relatedness of malaria parasites. We develop an individual-based transmission model to simulate malaria parasite genetics parameterized using estimated relationships between complexity of infection and age from five regions in Uganda and Kenya. We predict that cotransmission and superinfection contribute equally to within-host parasite genetic diversity at 11.5% PCR prevalence, above which superinfections dominate. Finally, we characterize the predictive power of six metrics of parasite genetics for detecting changes in transmission intensity, before grouping them in an ensemble statistical model. The model predicted malaria prevalence with a mean absolute error of 0.055. Different assumptions about the availability of sample metadata were considered, with the most accurate predictions of malaria prevalence made when the clinical status and age of sampled individuals is known. Parasite genetics may provide a novel surveillance tool for estimating the prevalence of malaria in areas in which prevalence surveys are not feasible. However, the findings presented here reinforce the need for patient metadata to be recorded and made available within all future attempts to use parasite genetics for surveillance., (© The Author(s) 2020. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution.)

Prevalence and levels of antibodies to multiple Plasmodium falciparum antigens show promise as tools for estimating malaria exposure. In a highland area of Kenya with unstable transmission, we assessed the presence and levels of antibodies to 12 pre-erythrocytic and blood-stage P. falciparum antigens by multiplex cytometric bead assay or ELISA in 604 individuals in August 2007, with follow-up testing in this cohort in April 2008, April 2009, and May 2010. Four hundred individuals were tested at all four time points. During this period, the only substantial malaria incidence occurred from April to August 2009. Antibody prevalence in adults was high at all time points (> 70%) for apical membrane antigen 1, erythrocyte-binding antigen 175, erythrocyte-binding protein-2, glutamate rich protein (GLURP)-R2, merozoite surface protein (MSP) 1 (19), MSP-1 (42), and liver-stage antigen-1; moderate (30-70%) for GLURP-R0, MSP-3, and thrombospondin-related adhesive protein; and low (< 30%) for SE and circumsporozoite protein (CSP). Changes in community-wide malaria exposure were best reflected in decreasing antibody levels overtime for highly immunogenic antigens, and in antibody seroprevalence overtime for the less-immunogenic antigens. Over the 3 years, antibody levels to all antigens except CSP and schizont extract (SE) decreased in an age-dependent manner. Prevalence and levels of antibodies to all antigens except CSP and SE increased with age. Increases in antibody prevalence and levels to CSP and SE coincided with increases in community-wide malaria incidence. Antibody levels to multiple P. falciparum antigens decrease in the absence of consistent transmission. Multiplex assays that assess both the presence and level of antibodies to multiple pre-erythrocytic and blood-stage P. falciparum antigens may provide the most useful estimates of past and recent malaria transmission in areas of unstable transmission and could be useful tools in malaria control and elimination campaigns.

Background: Reducing the burden of anaemia is a critical global health priority that could improve maternal outcomes amongst pregnant women and their neonates. As more counties in Kenya commit to universal health coverage, there is a growing need for optimal allocation of the limited resources to sustain the gains achieved with the devolution of healthcare services. This study aimed to describe the spatio-temporal patterns of maternal anaemia prevalence in Kenya from 2016 to 2019., Methods: Quarterly reported sub-county level maternal anaemia cases from January 2016 - December 2019 were obtained from the Kenyan District Health Information System. A Bayesian hierarchical negative binomial spatio-temporal conditional autoregressive (CAR) model was used to estimate maternal anaemia prevalence by sub-county and quarter. Spatial and temporal correlations were considered by assuming a conditional autoregressive and a first-order autoregressive process on sub-county and seasonal specific random effects, respectively., Results: The overall estimated number of pregnant women with anaemia increased by 90.1% (95% uncertainty interval [95% UI], 89.9-90.2) from 155,539 cases in 2016 to 295,642 cases 2019. Based on the WHO classification criteria, the proportion of sub-counties with normal prevalence decreased from 28.0% (95% UI, 25.4-30.7) in 2016 to 5.4% (95% UI, 4.1-6.7) in 2019, whereas moderate anaemia prevalence increased from 16.8% (95% UI, 14.7-19.1) in 2016 to 30.1% (95% UI, 27.5-32.8) in 2019 and severe anaemia prevalence increased from 7.0% (95% UI, 5.6-8.6) in 2016 to 16.6% (95% UI, 14.5-18.9) in 2019. Overall, 45.1% (95% UI: 45.0-45.2) of the estimated cases were in malaria-endemic sub-counties, with the coastal endemic zone having the highest proportion 72.8% (95% UI: 68.3-77.4) of sub-counties with severe prevalence., Conclusion: As the number of women of reproductive age continues to grow in Kenya, the use of routinely collected data for accurate mapping of poor maternal outcomes remains an integral component of a functional maternal health strategy. By unmasking the sub-county disparities often concealed by national and county estimates, our study findings reiterate the importance of maternal anaemia prevalence as a metric for estimating malaria burden and offers compelling policy implications for achieving national nutritional targets.

Introduction: Malaria surveillance is a key pillar in the control of malaria in Africa. The value of using routinely collected data from health facilities to define malaria risk at community levels remains poorly defined., Methods: Four cross-sectional parasite prevalence surveys were undertaken among residents at 36 enumeration zones in Kilifi county on the Kenyan coast and temporally and spatially matched to fever surveillance at 6 health facilities serving the same communities over 12 months. The age-structured functional form of the relationship between test positivity rate (TPR) and community-based parasite prevalence (PR) was explored through the development of regression models fitted by alternating the linear, exponential and polynomial terms for PR. The predictive ranges of TPR were explored for PR endemicity risk groups of control programmatic value using cut-offs of low (PR <5%) and high (PR ≥ 30%) transmission intensity., Results: Among 28,134 febrile patients encountered for malaria diagnostic testing in the health facilities, 12,143 (43.2%: 95% CI: 42.6%, 43.7%) were positive. The overall community PR was 9.9% (95% CI: 9.2%, 10.7%) among 6,479 participants tested for malaria. The polynomial model was the best fitting model for the data that described the algebraic relationship between TPR and PR. In this setting, a TPR of ≥ 49% in all age groups corresponded to an age-standardized PR of ≥ 30%, while a TPR of < 40% corresponded to an age-standardized PR of < 5%., Conclusion: A non-linear relationship was observed between the relative change in TPR and changes in the PR, which is likely to have important implications for malaria surveillance programs, especially at the extremes of transmission. However, larger, more spatially diverse data series using routinely collected TPR data matched to community-based infection prevalence data are required to explore the more practical implications of using TPR as a replacement for community PR., Competing Interests: The authors have declared that no competing interests exist.

Missing information is a major drawback in analyzing data collected in many routine health care settings. Multiple imputation assuming a missing at random mechanism is a popular method to handle missing data. The missing at random assumption cannot be confirmed from the observed data alone, hence the need for sensitivity analysis to assess robustness of inference. However, sensitivity analysis is rarely conducted and reported in practice. We analyzed routine paediatric data collected during a cluster randomized trial conducted in Kenyan hospitals. We imputed missing patient and clinician-level variables assuming the missing at random mechanism. We also imputed missing clinician-level variables assuming a missing not at random mechanism. We incorporated opinions from 15 clinical experts in the form of prior distributions and shift parameters in the delta adjustment method. An interaction between trial intervention arm and follow-up time, hospital, clinician and patient-level factors were included in a proportional odds random-effects analysis model. We performed these analyses using R functions derived from the jomo package. Parameter estimates from multiple imputation under the missing at random mechanism were similar to multiple imputation estimates assuming the missing not at random mechanism. Our inferences were insensitive to departures from the missing at random assumption using either the prior distributions or shift parameters sensitivity analysis approach.

The analysis of area-level aggregated summary data is common in many disciplines including epidemiology and the social sciences. Typically, Markov random field spatial models have been employed to acknowledge spatial dependence and allow data-driven smoothing. In the context of an irregular set of areas, these models always have an ad hoc element with respect to the definition of a neighborhood scheme. In this article, we exploit recent theoretical and computational advances to carry out modeling at the continuous spatial level, which induces a spatial model for the discrete areas. This approach also allows reconstruction of the continuous underlying surface, but the interpretation of such surfaces is delicate since it depends on the quality, extent and configuration of the observed data. We focus on models based on stochastic partial differential equations. We also consider the interesting case in which the aggregate data are supplemented with point data. We carry out Bayesian inference and, in the language of generalized linear mixed models, if the link is linear, an efficient implementation of the model is available via integrated nested Laplace approximations. For nonlinear links, we present two approaches: a fully Bayesian implementation using a Hamiltonian Monte Carlo algorithm and an empirical Bayes implementation, that is much faster and is based on Laplace approximations. We examine the properties of the approach using simulation, and then apply the model to the classic Scottish lip cancer data., (© The Author 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)

Multiple diagnostic tests are often used due to limited resources or because they provide complementary information on the epidemiology of a disease under investigation. Existing statistical methods to combine prevalence data from multiple diagnostics ignore the potential overdispersion induced by the spatial correlations in the data. To address this issue, we develop a geostatistical framework that allows for joint modelling of data from multiple diagnostics by considering two main classes of inferential problems: (a) to predict prevalence for a gold-standard diagnostic using low-cost and potentially biased alternative tests; (b) to carry out joint prediction of prevalence from multiple tests. We apply the proposed framework to two case studies: mapping Loa loa prevalence in Central and West Africa, using miscroscopy, and a questionnaire-based test called RAPLOA; mapping Plasmodium falciparum malaria prevalence in the highlands of Western Kenya using polymerase chain reaction and a rapid diagnostic test. We also develop a Monte Carlo procedure based on the variogram in order to identify parsimonious geostatistical models that are compatible with the data. Our study highlights (a) the importance of accounting for diagnostic-specific residual spatial variation and (b) the benefits accrued from joint geostatistical modelling so as to deliver more reliable and precise inferences on disease prevalence., (© 2019 The International Biometric Society.)

Background: Some villages, labeled "persistent hotspots (PHS)," fail to respond adequately in regard to prevalence and intensity of infection to mass drug administration (MDA) for schistosomiasis. Early identification of PHS, for example, before initiating or after 1 or 2 years of MDA could help guide programmatic decision making., Methods: In a study with multiple rounds of MDA, data collected before the third MDA were used to predict PHS. We assessed 6 predictive approaches using data from before MDA and after 2 rounds of annual MDA from Kenya and Tanzania., Results: Generalized linear models with variable selection possessed relatively stable performance compared with tree-based methods. Models applied to Kenya data alone or combined data from Kenya and Tanzania could reach over 80% predictive accuracy, whereas predicting PHS for Tanzania was challenging. Models developed from one country and validated in another failed to achieve satisfactory performance. Several Year-3 variables were identified as key predictors., Conclusions: Statistical models applied to Year-3 data could help predict PHS and guide program decisions, with infection intensity, prevalence of heavy infections (≥400 eggs/gram of feces), and total prevalence being particularly important factors. Additional studies including more variables and locations could help in developing generalizable models., (© The Author(s) 2019. Published by Oxford University Press for the Infectious Diseases Society of America.)

Misclassification of outcomes or event types is common in health sciences research and can lead to serious bias when estimating the cumulative incidence functions in settings with competing risks. Recent work has shown how to estimate nonparametric cumulative incidence functions in the presence of nondifferential outcome misclassification when the misclassification probabilities are known. Here, we extend this approach to account for misclassification that is differential with respect to important predictors of the outcome using misclassification probabilities estimated from external validation data. Moreover, we propose a bootstrap approach in which the observations from both the main study data and the external validation study are resampled to allow the uncertainty in the misclassification probabilities to propagate through the analysis into the final confidence intervals, ensuring appropriate confidence interval coverage probabilities. The proposed estimator is shown to be uniformly consistent and simulation studies indicate that both the estimator and the standard error estimation approach perform well in finite samples. The methodology is applied to estimate the cumulative incidence of death and disengagement from HIV care in a large cohort of HIV infected individuals in sub-Saharan Africa, where a significant death underreporting issue leads to outcome misclassification. This analysis uses external validation data from a separate study conducted in the same country., (© 2019 John Wiley & Sons, Ltd.)

Background: As many countries with endemic soil-transmitted helminth (STH) burdens achieve high coverage levels of mass drug administration (MDA) to treat school-aged and pre-school-aged children, understanding the detailed effects of MDA on the epidemiology of STH infections is desirable in formulating future policies for morbidity and/or transmission control. Prevalence and mean intensity of infection are characterized by heterogeneity across a region, leading to uncertainty in the impact of MDA strategies. In this paper, we analyze this heterogeneity in terms of factors that govern the transmission dynamics of the parasite in the host population., Results: Using data from the TUMIKIA study in Kenya (cluster STH prevalence range at baseline: 0-63%), we estimated these parameters and their variability across 120 population clusters in the study region, using a simple parasite transmission model and Gibbs-sampling Monte Carlo Markov chain techniques. We observed great heterogeneity in R 0 values, with estimates ranging from 1.23 to 3.27, while k-values (which vary inversely with the degree of parasite aggregation within the human host population) range from 0.007 to 0.29 in a positive association with increasing prevalence. The main finding of this study is the increasing trend for greater parasite aggregation as prevalence declines to low levels, reflected in the low values of the negative binomial parameter k in clusters with low hookworm prevalence. Localized climatic and socioeconomic factors are investigated as potential drivers of these observed epidemiological patterns., Conclusions: Our results show that lower prevalence is associated with higher degrees of aggregation and hence prevalence alone is not a good indicator of transmission intensity. As a consequence, approaches to MDA and monitoring and evaluation of community infection status may need to be adapted as transmission elimination is aimed for by targeted treatment approaches.

Background: Human metapneumovirus (HMPV) have similar symptoms to those caused by the respiratory syncytial virus (RSV). The modes of transmission and dynamics of time series data still remain poorly understood. Climatic factors have long been suspected to be implicated in impacting on the number of cases for these epidemics. Currently, only a few models satisfactorily capture the dynamics of time series data of these two viruses. Our objective was to assess the presence of influence of high incidences between the viruses and to ascertain whether higher incidences of one virus are influenced by the other., Methods: In this study, we used a negative binomial model to investigate the relationship between RSV and HMPV while adjusting for climatic factors. We specifically aimed at establishing the heterogeneity in the autoregressive effect to account for the influence between these viruses., Results: In this study, our findings showed that RSV incidence contributed to the severity of HMPV incidence. This was achieved through comparison of 12 models with different structures, including those with and without interaction between climatic factors. The models with climatic factors out-performed those without., Conclusions: The study has improved our understanding of the dynamics of RSV and HMPV in relation to climatic cofactors thereby setting a platform to devise better intervention measures to combat the epidemics. We conclude that preventing and controlling RSV infection subsequently reduces the incidence of HMPV.

Background: In the context of WHO's End TB strategy, there is a need to focus future control efforts on those interventions and innovations that would be most effective in accelerating declines in tuberculosis burden. Using a modelling approach to link the tuberculosis care cascade to transmission, we aimed to identify which improvements in the cascade would yield the greatest effect on incidence and mortality., Methods: We engaged with national tuberculosis programmes in three country settings (India, Kenya, and Moldova) as illustrative examples of settings with a large private sector (India), a high HIV burden (Kenya), and a high burden of multidrug resistance (Moldova). We collated WHO country burden estimates, routine surveillance data, and tuberculosis prevalence surveys from 2011 (for India) and 2016 (for Kenya). Linking the tuberculosis care cascade to tuberculosis transmission using a mathematical model with Bayesian melding in each setting, we examined which cascade shortfalls would have the greatest effect on incidence and mortality, and how the cascade could be used to monitor future control efforts., Findings: Modelling suggests that combined measures to strengthen the care cascade could reduce cumulative tuberculosis incidence by 38% (95% Bayesian credible intervals 27-43) in India, 31% (25-41) in Kenya, and 27% (17-41) in Moldova between 2018 and 2035. For both incidence and mortality, modelling suggests that the most important cascade losses are the proportion of patients visiting the private health-care sector in India, missed diagnosis in health-care settings in Kenya, and drug sensitivity testing in Moldova. In all settings, the most influential delay is the interval before a patient's first presentation for care. In future interventions, the proportion of individuals with tuberculosis who are on high-quality treatment could offer a more robust monitoring tool than routine notifications of tuberculosis., Interpretation: Linked to transmission, the care cascade can be valuable, not only for improving patient outcomes but also in identifying and monitoring programmatic priorities to reduce tuberculosis incidence and mortality., Funding: US Agency for International Development, Stop TB Partnership, UK Medical Research Council, and Department for International Development., (Copyright © 2019 The Authors. Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND 4.0 license. Published by Elsevier Ltd.. All rights reserved.)

Young children are infected by a diverse variety of enteric pathogens in low-income, high-burden countries. Little is known about which conditions pose the greatest risk for enteric pathogen exposure and infection. Young children frequently play in residential public areas around their household, including areas contaminated by human and animal feces, suggesting these exposures are particularly hazardous. The objective of this study was to examine how the dose of six types of common enteric pathogens, and the probability of exposure to one or multiple enteric pathogens for young children playing at public play areas in Kisumu, Kenya is influenced by the type and frequency of child play behaviors that result in ingestion of soil or surface water. Additionally, we examine how pathogen doses and multi-pathogen exposure are modified by spatial variability in the number of public areas children are exposed to in their neighborhood. A Bayesian framework was employed to obtain the posterior distribution of pathogen doses for a certain number of contacts. First, a multivariate mixed effects tobit model was used to obtain the posterior distribution of pathogen concentrations, and their interdependencies, in soil and surface water, based upon empirical data of enteric pathogen contamination in three neighborhoods of Kisumu. Then, exposure doses were estimated using behavioral contact parameters from previous studies and contrasted under different exposure conditions. Pathogen presence and concentration in soil varied widely across local (< 25 meter radius area) and neighborhood-level scales, but pathogens were correlated among distinct surface water samples collected near to each other. Multi-pathogen exposure of children at public play areas was common. Pathogen doses and the probability of multi-pathogen ingestion increased with: higher frequency of environmental contact, especially for surface water; larger volume of soil or water ingested; and with play at multiple sites in the neighborhood versus single site play. Child contact with surface water and soil at public play areas in their neighborhood is an important cause of exposure to enteric pathogens in Kisumu, and behavioral, environmental, and spatial conditions are determinants of exposure., Competing Interests: The authors have declared that no competing interests exist.

Background: Kenya is 1 of 14 priority countries in Africa scaling up voluntary medical male circumcision (VMMC) for HIV prevention following the recommendations of the World Health Organization and the Joint United Nations Programme on HIV/AIDS. To inform VMMC target setting, we modeled the impact of circumcising specific client age groups across several Kenyan geographic areas., Methods: The Decision Makers' Program Planning Tool, Version 2 (DMPPT 2) was applied in Kisumu, Siaya, Homa Bay, and Migori counties. Initial modeling done in mid-2016 showed coverage estimates above 100% in age groups and geographic areas where demand for VMMC continued to be high. On the basis of information obtained from country policy makers and VMMC program implementers, we adjusted circumcision coverage for duplicate reporting, county-level population estimates, migration across county boundaries for VMMC services, and replacement of traditional circumcision with circumcisions in the VMMC program. To address residual inflated coverage following these adjustments we applied county-specific correction factors computed by triangulating model results with coverage estimates from population surveys., Results: A program record review identified duplicate reporting in Homa Bay, Kisumu, and Siaya. Using county population estimates from the Kenya National Bureau of Statistics, we found that adjusting for migration and correcting for replacement of traditional circumcision with VMMC led to lower estimates of 2016 male circumcision coverage especially for Kisumu, Migori, and Siaya. Even after addressing these issues, overestimation of 2016 male circumcision coverage persisted, especially in Homa Bay. We estimated male circumcision coverage in 2016 by applying correction factors. Modeled estimates for 2016 circumcision coverage for the 10- to 14-year age group ranged from 50% in Homa Bay to approximately 90% in Kisumu. Results for the 15- to 19-year age group suggest almost complete coverage in Kisumu, Migori, and Siaya. Coverage for the 20- to 24-year age group ranged from about 80% in Siaya to about 90% in Homa Bay, coverage for those aged 25-29 years ranged from about 60% in Siaya to 80% in Migori, and coverage in those aged 30-34 years ranged from about 50% in Siaya to about 70% in Migori., Conclusions: Our analysis points to solutions for some of the data issues encountered in Kenya. Kenya is the first country in which these data issues have been encountered because baseline circumcision rates were high. We anticipate that some of the modeling methods we developed for Kenya will be applicable in other countries., Competing Interests: The authors have declared that no competing interests exist.

This study examined men's participation and performance in the Boston Marathon. Data from all 368,940 official male finishers in the Boston Marathon from 1897 through 2017 were analyzed using different analyses and regression models for all runners, annual top hundred finishers, annual top ten finishers and annual winners. Effect of calendar year on race time was examined alone, in a uni-variable model, and together with country of origin and weather conditions (average air temperature and precipitation) in a multi-variable model. The mean race time of all finishers increased across calendar years, in line with the participation, but it decreased when we considered the annual winners, 10 and 100 fastest. Kenyans and Ethiopians were the fastest nationalities (p<0.001) when we considered all finishers and the annual top 100 fastest but not when the annual 10 fastest were considered. Air temperature≤8 °C improved (p<0.001) race times compared to air temperature>8°C. Precipitation (>0 mm) improved performance for the annual 100 fastest and annual 10 fastest (p=0.013-0.031) but not for all finishers. Our findings improved the knowledge about the evolution of male marathoners across calendar years, considering as main effects country of origin and particular weather conditions., Competing Interests: No conflict of interest has been declared by the authors., (© Georg Thieme Verlag KG Stuttgart · New York.)

Precision public health approaches are crucial for targeting health policies to regions most affected by disease. We present the first sub-national and spatially explicit burden of disease study in Africa. We used a cross-sectional study design and assessed data from the Kenya population and housing census of 2009 for calculating YLLs (years of life lost) due to premature mortality at the division level (N = 612). We conducted spatial autocorrelation analysis to identify spatial clusters of YLLs and applied boosted regression trees to find statistical associations between locational risk factors and YLLs. We found statistically significant spatial clusters of high numbers of YLLs at the division level in western, northwestern, and northeastern areas of Kenya. Ethnicity and household crowding were the most important and significant risk factors for YLL. Further positive and significantly associated variables were malaria endemicity, northern geographic location, and higher YLL in neighboring divisions. In contrast, higher rates of married people and more precipitation in a division were significantly associated with less YLL. We provide an evidence base and a transferable approach that can guide health policy and intervention in sub-national regions afflicted by disease burden in Kenya and other areas of comparable settings.

The timing of antiretroviral therapy (ART) initiation for HIV and tuberculosis (TB) co-infected patients needs to be considered carefully. CD4 cell count can be used to guide decision making about when to initiate ART. Evidence from recent randomized trials and observational studies generally supports early initiation but does not provide information about effects of initiation time on a continuous scale. In this article, we develop and apply a highly flexible structural proportional hazards model for characterizing the effect of treatment initiation time on a survival distribution. The model can be fitted using a weighted partial likelihood score function. Construction of both the score function and the weights must accommodate censoring of the treatment initiation time, the outcome, or both. The methods are applied to data on 4903 individuals with HIV/TB co-infection, derived from electronic health records in a large HIV care program in Kenya. We use a model formulation that flexibly captures the joint effects of ART initiation time and ART duration using natural cubic splines. The model is used to generate survival curves corresponding to specific treatment initiation times; and to identify optimal times for ART initiation for subgroups defined by CD4 count at time of TB diagnosis. Our findings potentially provide 'higher resolution' information about the relationship between ART timing and mortality, and about the differential effect of ART timing within CD4 subgroups., (© 2017, The International Biometric Society.)

Malaria has been a major driving force in the evolution of the human genome. In sub-Saharan African populations, two neighbouring polymorphisms in the Complement Receptor One ( CR1 ) gene, named Sl2 and McC b , occur at high frequencies, consistent with selection by malaria. Previous studies have been inconclusive. Using a large case-control study of severe malaria in Kenyan children and statistical models adjusted for confounders, we estimate the relationship between Sl2 and McC b and malaria phenotypes, and find they have opposing associations. The Sl2 polymorphism is associated with markedly reduced odds of cerebral malaria and death, while the McC b polymorphism is associated with increased odds of cerebral malaria. We also identify an apparent interaction between Sl2 and α + thalassaemia, with the protective association of Sl2 greatest in children with normal α-globin. The complex relationship between these three mutations may explain previous conflicting findings, highlighting the importance of considering genetic interactions in disease-association studies., Competing Interests: DO, OS, AM, SU, GB, CN, EH, MT, AK, DD, OD, KL, CP, JM, MS, NM, NP, KM, AR, DK, KR, TW, JR No competing interests declared, (© 2018, Opi et al.)

Background: Kato-Katz technique has been the mainstay test in Schistosoma mansoni diagnosis in endemic areas. However, recent studies have documented its poor sensitivity in evaluating Schistosoma mansoni infection especially in areas with lower rates of transmission. It's the primary diagnostic tool in monitoring impact of the Kenya national school based deworming program on infection transmission, but there is need to consider a more sensitive technique as the prevalence reduces. Therefore, this study explored the relationship between results of the stool-based Kato-Katz technique with urine-based point-of-care circulating cathodic antigen (POC-CCA) test in view to inform decision-making by the program in changing from Kato-Katz to POC-CCA test., Methods: We used two cross-sectional surveys conducted pre- and post- mass drug administration (MDA) using praziquantel in a representative random sample of children from 18 schools across 11 counties. A total of 1944 children were randomly sampled for the study. Stool and urine samples were tested for S. mansoni infection using Kato-Katz and POC-CCA methods, respectively. S. mansoni prevalence using each technique was calculated and 95% confidence intervals obtained using binomial regression model. Specificity (Sp) and sensitivity (Sn) were determined using 2 × 2 contingency tables and compared using the McNemar's chi-square test., Results: A total of 1899 and 1878 children were surveyed at pre- and post-treatment respectively. S. mansoni infection prevalence was 26.5 and 21.4% during pre- and post-treatment respectively using POC-CCA test, and 4.9 and 1.5% for pre- and post-treatment respectively using Kato-Katz technique. Taking POC-CCA as the gold standard, Kato-Katz was found to have significantly lower sensitivity both at pre- and post-treatment, Sn = 12.5% and Sn = 5.2% respectively, McNemar test χ 2 m  = 782.0, p < 0.001. In overall, the results showed a slight/poor agreement between the two methods, kappa index (k) = 0.11, p < 0.001, inter-rater agreement = 77.1%., Conclusions: Results showed POC-CCA technique as an effective, sensitive and accurate screening tool for Schistosoma mansoni infection in areas of low prevalence. It was up to 14-fold accurate than Kato-Katz which had extremely inadequate sensitivity. We recommend usage of POC-CCA alongside Kato-Katz examinations by Schistosomiasis control programs in low prevalence areas.

Malaria hotspots, defined as areas where transmission intensity exceeds the average level, become more pronounced as transmission declines. Targeting hotspots may accelerate reductions in transmission and could be pivotal for malaria elimination. Determinants of hotspot location, particularly of their movement, are poorly understood. We used spatial statistical methods to identify foci of incidence of self-reported malaria in a large census population of 64,000 people, in 8,290 compounds over a 2.5-year study period. Regression models examine stability of hotspots and identify static and dynamic correlates with their location. Hotspot location changed over short time-periods, rarely recurring in the same area. Hotspots identified in spring versus fall season differed in their stability. Households located in a hotspot in the fall were more likely to be located in a hotspot the following fall (RR = 1.77, 95% CI: 1.66-1.89), but the opposite was true for compounds in spring hotspots (RR = 0.15, 95% CI: 0.08-0.28). Location within a hotspot was related to environmental and static household characteristics such as distance to roads or rivers. Human migration into a household was correlated with risk of hotspot membership, but the direction of the association differed based on the origin of the migration event.

Malaria surveillance data provide opportunity to develop forecasting models. Seasonal variability in environmental factors correlate with malaria transmission, thus the identification of transmission patterns is useful in developing prediction models. However, with changing seasonal transmission patterns, either due to interventions or shifting weather seasons, traditional modelling approaches may not yield adequate predictive skill. Two statistical models,a general additive model (GAM) and GAMBOOST model with boosted regression were contrasted by assessing their predictive accuracy in forecasting malaria admissions at lead times of one to three months. Monthly admission data for children under five years with confirmed malaria at the Siaya district hospital in Western Kenya for the period 2003 to 2013 were used together with satellite derived data on rainfall, average temperature and evapotranspiration(ET). There was a total of 8,476 confirmed malaria admissions. The peak of malaria season changed and malaria admissions reduced overtime. The GAMBOOST model at 1-month lead time had the highest predictive skill during both the training and test periods and thus can be utilized in a malaria early warning system.

Most agricultural pests are poikilothermic species expected to respond to climate change. Currently, they are a tremendous burden because of the high losses they inflict on crops and livestock. Smallholder farmers in developing countries of Africa are likely to suffer more under these changes than farmers in the developed world because more severe climatic changes are projected in these areas. African countries further have a lower ability to cope with impacts of climate change through the lack of suitable adapted management strategies and financial constraints. In this study we are predicting current and future habitat suitability under changing climatic conditions for Tuta absoluta, Ceratitis cosyra, and Bactrocera invadens, three important insect pests that are common across some parts of Africa and responsible for immense agricultural losses. We use presence records from different sources and bioclimatic variables to predict their habitat suitability using the maximum entropy modelling approach. We find that habitat suitability for B. invadens, C. cosyra and T. absoluta is partially increasing across the continent, especially in those areas already overlapping with or close to most suitable sites under current climate conditions. Assuming a habitat suitability at three different threshold levels we assessed where each species is likely to be present under future climatic conditions and if this is likely to have an impact on productive agricultural areas. Our results can be used by African policy makers, extensionists and farmers for agricultural adaptation measures to cope with the impacts of climate change.

Background: Malaria is a vector-borne disease which, despite recent scaled-up efforts to achieve control in Africa, continues to pose a major threat to child survival. The disease is caused by the protozoan parasite Plasmodium and requires mosquitoes and humans for transmission. Rainfall is a major factor in seasonal and secular patterns of malaria transmission along the East African coast., Objective: The goal of the study was to develop a model to reliably forecast incidences of paediatric malaria admissions to Kilifi District Hospital (KDH)., Design: In this article, we apply several statistical models to look at the temporal association between monthly paediatric malaria hospital admissions, rainfall, and Indian Ocean sea surface temperatures. Trend and seasonally adjusted, marginal and multivariate, time-series models for hospital admissions were applied to a unique data set to examine the role of climate, seasonality, and long-term anomalies in predicting malaria hospital admission rates and whether these might become more or less predictable with increasing vector control., Results: The proportion of paediatric admissions to KDH that have malaria as a cause of admission can be forecast by a model which depends on the proportion of malaria admissions in the previous 2 months. This model is improved by incorporating either the previous month's Indian Ocean Dipole information or the previous 2 months' rainfall., Conclusions: Surveillance data can help build time-series prediction models which can be used to anticipate seasonal variations in clinical burdens of malaria in stable transmission areas and aid the timing of malaria vector control.

Background: The body of knowledge on evaluating complex interventions for integrated healthcare lacks both common definitions of 'integrated service delivery' and standard measures of impact. Using multiple data sources in combination with statistical modelling the aim of this study is to develop a measure of HIV-reproductive health (HIV-RH) service integration that can be used to assess the degree of service integration, and the degree to which integration may have health benefits to clients, or reduce service costs., Methods and Findings: Data were drawn from the Integra Initiative's client flow (8,263 clients in Swaziland and 25,539 in Kenya) and costing tools implemented between 2008-2012 in 40 clinics providing RH services in Kenya and Swaziland. We used latent variable measurement models to derive dimensions of HIV-RH integration using these data, which quantified the extent and type of integration between HIV and RH services in Kenya and Swaziland. The modelling produced two clear and uncorrelated dimensions of integration at facility level leading to the development of two sub-indexes: a Structural Integration Index (integrated physical and human resource infrastructure) and a Functional Integration Index (integrated delivery of services to clients). The findings highlight the importance of multi-dimensional assessments of integration, suggesting that structural integration is not sufficient to achieve the integrated delivery of care to clients--i.e. "functional integration"., Conclusions: These Indexes are an important methodological contribution for evaluating complex multi-service interventions. They help address the need to broaden traditional evaluations of integrated HIV-RH care through the incorporation of a functional integration measure, to avoid misleading conclusions on its 'impact' on health outcomes. This is particularly important for decision-makers seeking to promote integration in resource constrained environments.

Background: Visceral leishmaniasis (VL) is a parasitic disease transmitted by sandflies and is fatal if left untreated. Phase II trials of new treatment regimens for VL are primarily carried out to evaluate safety and efficacy, while pharmacokinetic data are also important to inform future combination treatment regimens. The efficacy of VL treatments is evaluated at two time points, initial cure, when treatment is completed and definitive cure, commonly 6 months post end of treatment, to allow for slow response to treatment and detection of relapses. This paper investigates a generalization of the triangular design to impose a minimum sample size for pharmacokinetic or other analyses, and methods to estimate efficacy at extended follow-up accounting for the sequential design and changes in cure status during extended follow-up., Methods: We provided R functions that generalize the triangular design to impose a minimum sample size before allowing stopping for efficacy. For estimation of efficacy at a second, extended, follow-up time, the performance of a shrinkage estimator (SHE), a probability tree estimator (PTE) and the maximum likelihood estimator (MLE) for estimation was assessed by simulation., Results: The SHE and PTE are viable approaches to estimate an extended follow-up although the SHE performed better than the PTE: the bias and root mean square error were lower and coverage probabilities higher., Conclusions: Generalization of the triangular design is simple to implement for adaptations to meet requirements for pharmacokinetic analyses. Using the simple MLE approach to estimate efficacy at extended follow-up will lead to biased results, generally over-estimating treatment success. The SHE is recommended in trials of two or more treatments. The PTE is an acceptable alternative for one-arm trials or where use of the SHE is not possible due to computational complexity., Trial Registration: NCT01067443 , February 2010.

Antiretroviral preexposure prophylaxis (PrEP) for persons at high risk of human immunodeficiency virus infection is a promising new prevention strategy. Six randomized trials of oral PrEP were recently conducted and demonstrated efficacy estimates ranging from 75% to no effect, with nonadherence likely resulting in attenuated estimates of the protective effect of PrEP. In 1 of these trials, the Partners PrEP Study (Kenya and Uganda, 2008-2011), participants (4,747 serodiscordant heterosexual couples) were randomized to receipt of tenofovir (TDF), coformulated TDF/emtricitabine (FTC), or placebo. Intention-to-treat analyses found efficacy estimates of 67% for TDF and 75% for TDF/FTC. We applied multiple methods to data from that trial to estimate the efficacy of PrEP with high adherence, including principal stratification and inverse-probability-of-censoring (IPC) weights. Results were further from the null when correcting for nonadherence: 1) among the strata with an estimated 100% probability of high adherence (TDF hazard ratio (HR) = 0.19, 95% confidence interval (CI): 0.07, 0.56; TDF/FTC HR = 0.12, 95% CI: 0.03, 0.52); 2) with IPC weights used to approximate a continuously adherent population (TDF HR = 0.18, 95% CI: 0.06, 0.53; TDF/FTC HR = 0.15, 95% CI: 0.04, 0.52); and 3) in per-protocol analysis (TDF HR = 0.18, 95% CI: 0.06, 0.53; TDF/FTC HR = 0.16, 95% CI: 0.05, 0.53). Our results suggest that the efficacy of PrEP with high adherence is over 80%., (© The Author 2015. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)

Background: Respiratory syncytial virus (RSV) is globally ubiquitous, and infection during the first six months of life is a major risk for severe disease and hospital admission; consequently RSV is the most important viral cause of respiratory morbidity and mortality in young children. Development of vaccines for young infants is complicated by the presence of maternal antibodies and immunological immaturity, but vaccines targeted at older children avoid these problems. Vaccine development for young infants has been unsuccessful, but this is not the case for older children (> 6 m). Would vaccinating older children have a significant public health impact? We developed a mathematical model to explore the benefits of a vaccine against RSV., Methods and Findings: We have used a deterministic age structured model capturing the key epidemiological characteristics of RSV and performed a statistical maximum-likelihood fit to age-specific hospitalization data from a developing country setting. To explore the effects of vaccination under different mixing assumptions, we included two versions of contact matrices: one from a social contact diary study, and the second a synthesised construction based on demographic data. Vaccination is assumed to elicit an immune response equivalent to primary infection. Our results show that immunisation of young children (5-10 m) is likely to be a highly effective method of protection of infants (<6 m) against hospitalisation. The majority benefit is derived from indirect protection (herd immunity). A full sensitivity and uncertainty analysis using Latin Hypercube Sampling of the parameter space shows that our results are robust to model structure and model parameters., Conclusions: This result suggests that vaccinating older infants and children against RSV can have a major public health benefit.

Several diseases have common risk factors. The joint modeling of disease outcomes within a spatial statistical context may provide more insight on the interaction of diseases both at individual and at regional level. Spatial joint modeling allows for studying of the relationship between diseases and also between regions under study. One major approach for joint spatial modeling is the multivariate conditional autoregressive approach. In this approach, it is assumed that all the covariates in the study have linear effects on the multiple response variables. In this study, we relax this linearity assumption and allow some covariates to have nonlinear effects using the penalized regression splines. This model was used to jointly model the spatial variation of human immunodeficiency virus (HIV) and herpes simplex virus-type 2 (HSV-2) among women in Kenya. The model was applied to HIV and HSV-2 prevalence data among women aged 15-49 years in Kenya, derived from the 2007 Kenya AIDS indicator survey. A full Bayesian approach was used and the models were implemented in WinBUGS software. Both diseases showed significant spatial variation with highest disease burdens occurring around the Lake Victoria region. There was a nonlinear association between age of an individual and HIV and HSV-2 infection. The peak age for HIV was around 30 years while that of HSV-2 was about 40 years. A positive significant spatial correlation between HIV and HSV-2 was observed with a correlation of 0.6831(95% CI: 0.3859, 0.871).

Background: Mathematical models have played important roles in the understanding of epidemics and in the study of the impacts of various behavioral or medical measures. However, modeling accurately the future spread of an epidemic requires context-specific parameters that are difficult to estimate because of lack of data. Our objective is to propose a methodology to estimate context-specific parameters using Demographic and Health Survey (DHS)-like data that can be used in mathematical modeling of short-term HIV spreading., Methods and Findings: The model splits the population according to sex, age, HIV status, and antiretroviral treatment status. To estimate context-specific parameters, we used individuals' histories included in DHS-like data and a statistical analysis that used decomposition of the Poisson likelihood. To predict the course of the HIV epidemic, sex- and age-specific differential equations were used. This approach was applied to recent data from Kenya. The approach allowed the estimation of several key epidemiological parameters. Women had a higher infection rate than men and the highest infection rate in the youngest age groups (15-24 and 25-34 years) whereas men had the highest infection rate in age group 25-34 years. The immunosuppression rates were similar between age groups. The treatment rate was the highest in age group 35-59 years in both sexes. The results showed that, within the 15-24 year age group, increasing male circumcision coverage and antiretroviral therapy coverage at CD4 ≤ 350/mm3 over the current 70% could have short-term impacts., Conclusions: The study succeeded in estimating the model parameters using DHS-like data rather than literature data. The analysis provides a framework for using the same data for estimation and prediction, which can improve the validity of context-specific predictions and help designing HIV prevention campaigns.

Introduction: Rotavirus vaccines have the potential to prevent a substantial amount of life-threatening gastroenteritis in young African children. This paper presents the results of prospective cost-effectiveness analyses for rotavirus vaccine introduction for Kenya and Uganda., Methodology: In each country, a national consultant worked with a national technical working group to identify appropriate data and validate study results. Secondary data on demographics, disease burden, health utilization, and costs were used to populate the TRIVAC cost-effectiveness model. The baseline analysis assumed an initial vaccine price of $0.20 per dose, corresponding to Gavi, the Vaccine Alliance stipulated copay for low-income countries. The incremental cost-effectiveness of a 2-dose rotavirus vaccination schedule was evaluated for 20 successive birth cohorts from the government perspective in both countries, and from the societal perspective in Uganda., Results: Between 2014 and 2033, rotavirus vaccination can avert approximately 60,935 and 216,454 undiscounted deaths and hospital admissions respectively in children under 5 years in Kenya. In Uganda, the respective number of undiscounted deaths and hospital admission averted is 70,236 and 329,779 between 2016 and 2035. Over the 20-year period, the discounted vaccine program costs are around US$ 80 million in Kenya and US$ 60 million in Uganda. Discounted government health service costs avoided are US$ 30 million in Kenya and US$ 10 million in Uganda (or US$ 18 million including household costs). The cost per disability-adjusted life-year (DALY) averted from a government perspective is US$ 38 in Kenya and US$ 34 in Uganda (US$ 29 from a societal perspective)., Conclusions: Rotavirus vaccine introduction is highly cost-effective in both countries in a range of plausible 'what-if' scenarios. The involvement of national experts improves the quality of data used, is likely to increase acceptability of the results in decision-making, and can contribute to strengthened national capacity to undertake economic evaluations., (Copyright © 2015 Elsevier Ltd. All rights reserved.)

High resolution, contemporary data on human population distributions are vital for measuring impacts of population growth, monitoring human-environment interactions and for planning and policy development. Many methods are used to disaggregate census data and predict population densities for finer scale, gridded population data sets. We present a new semi-automated dasymetric modeling approach that incorporates detailed census and ancillary data in a flexible, "Random Forest" estimation technique. We outline the combination of widely available, remotely-sensed and geospatial data that contribute to the modeled dasymetric weights and then use the Random Forest model to generate a gridded prediction of population density at ~100 m spatial resolution. This prediction layer is then used as the weighting surface to perform dasymetric redistribution of the census counts at a country level. As a case study we compare the new algorithm and its products for three countries (Vietnam, Cambodia, and Kenya) with other common gridded population data production methodologies. We discuss the advantages of the new method and increases over the accuracy and flexibility of those previous approaches. Finally, we outline how this algorithm will be extended to provide freely-available gridded population data sets for Africa, Asia and Latin America.

Background: Access to safe drinking water has been on the global agenda for decades. The key to safe drinking water is found in household water treatment and safe storage systems., Objective: In this study, we assessed rural and urban household demand for a new gravity-driven membrane (GDM) drinking-water filter., Methods: A choice experiment (CE) was used to assess the value attached to the characteristics of a new GDM filter before marketing in urban and rural Kenya. The CE was followed by a contingent valuation (CV) question. Differences in willingness to pay (WTP) for the same filter design were tested between methods, as well as urban and rural samples., Results: The CV follow-up approach produces more conservative and statistically more efficient WTP values than the CE, with only limited indications of anchoring. The effect of the new filter technology on children with diarrhea is among the most important drivers behind choice behavior and WTP in both areas. The urban sample is willing to pay more in absolute terms than the rural sample irrespective of the valuation method. Rural households are more price sensitive, and willing to pay more in relative terms compared with disposable household income., Conclusion: A differentiated marketing strategy across rural and urban areas is expected to increase uptake and diffusion of the new filter technology.

Purpose: In Kenya, gaps exist in health service provision to slum residents, especially service availability and access to quality care. There is also little information on the health status of people living in slums other than in Nairobi. The purpose of this paper is to generate evidence for use in designing interventions to improve health services in four mid-sized slums in Embu, Nyeri and Thika, Kenya., Design/methodology/approach: A cross-sectional survey of clients receiving services in health facilities was conducted in the targeted slums. Data were collected through face-to-face interviews. Factor scores were generated using the Rasch model; simple and multivariate logistic regression analyses were done using the R statistical software., Findings: Overall, 81 per cent of the 203 participants reported being satisfied with health services. Most clients (89 per cent) reported that health facility staff greeted them warmly; 82 per cent said their consultation was private. The facility type, waiting time and client experience with service providers determined their satisfaction (p<0.05)., Practical Implications: Healthcare managers can improve client satisfaction levels by understanding the client flow in their facilities and addressing causes of client dissatisfaction, such as long waiting times, while at the same time promoting facilitating factors., Originality/value: The authors use latent variable modelling to compute client satisfaction scores, which were dichotomised into two categories and fitted into a logistic regression model to identify factors that influence client satisfaction. Health facility clients in the four slums are satisfied with services and have confidence the providers will serve them in a friendly and professional manner that promotes respect and quality care. The paper recommend healthcare managers in similar settings carry out client flow analysis and institute remedial measures to address long waiting times. Qualitative studies are recommended to determine the reasons behind the high satisfaction levels reported in this study.

Background: Assessing the progress in achieving the United Nation's Millennium Development Goals in terms of population health requires consistent and reliable information on cause-specific mortality, which is often rare in resource-constrained countries. Health and demographic surveillance systems (HDSS) have largely used medical personnel to review and assign likely causes of death based on the information gathered from standardized verbal autopsy (VA) forms. However, this approach is expensive and time consuming, and it may lead to biased results based on the knowledge and experience of individual clinicians. We assessed the cause-specific mortality for children under 5 years old (under-5 deaths) in Siaya County, obtained from a computer-based probabilistic model (InterVA-4)., Design: Successfully completed VA interviews for under-5 deaths conducted between January 2003 and December 2010 in the Kenya Medical Research Institute/US Centers for Disease Control and Prevention HDSS were extracted from the VA database and processed using the InterVA-4 (version 4.02) model for interpretation. Cause-specific mortality fractions were then generated from the causes of death produced by the model., Results: A total of 84.33% (6,621) childhood deaths had completed VA data during the study period. Children aged 1-4 years constituted 48.53% of all cases, and 42.50% were from infants. A single cause of death was assigned to 89.18% (5,940) of cases, 8.35% (556) of cases were assigned two causes, and 2.10% (140) were assigned 'indeterminate' as cause of death by the InterVA-4 model. Overall, malaria (28.20%) was the leading cause of death, followed by acute respiratory infection including pneumonia (25.10%), in under-5 children over the study period. But in the first 5 years of the study period, acute respiratory infection including pneumonia was the main cause of death, followed by malaria. Similar trends were also reported in infants (29 days-11 months) and children aged 1-4 years., Conclusions: Under-5 cause-specific mortality obtained using the InterVA-4 model is consistent with existing knowledge on the burden of childhood diseases in rural western Kenya.

Introduction: Tools that allow for in silico optimization of available malaria control strategies can assist the decision-making process for prioritizing interventions. The OpenMalaria stochastic simulation modeling platform can be applied to simulate the impact of interventions singly and in combination as implemented in Rachuonyo South District, western Kenya, to support this goal., Methods: Combinations of malaria interventions were simulated using a previously-published, validated model of malaria epidemiology and control in the study area. An economic model of the costs of case management and malaria control interventions in Kenya was applied to simulation results and cost-effectiveness of each intervention combination compared to the corresponding simulated outputs of a scenario without interventions. Uncertainty was evaluated by varying health system and intervention delivery parameters., Results: The intervention strategy with the greatest simulated health impact employed long lasting insecticide treated net (LLIN) use by 80% of the population, 90% of households covered by indoor residual spraying (IRS) with deployment starting in April, and intermittent screen and treat (IST) of school children using Artemether lumefantrine (AL) with 80% coverage twice per term. However, the current malaria control strategy in the study area including LLIN use of 56% and IRS coverage of 70% was the most cost effective at reducing disability-adjusted life years (DALYs) over a five year period., Conclusions: All the simulated intervention combinations can be considered cost effective in the context of available resources for health in Kenya. Increasing coverage of vector control interventions has a larger simulated impact compared to adding IST to the current implementation strategy, suggesting that transmission in the study area is not at a level to warrant replacing vector control to a school-based screen and treat program. These results have the potential to assist malaria control program managers in the study area in adding new or changing implementation of current interventions.

Background: Health-care financing should be equitable. In many developing countries such as Kenya, changes to health-care financing systems are being implemented as a means of providing equitable access to health care with the aim of attaining universal coverage. Vertical equity means that people of dissimilar ability to pay make dissimilar levels of contribution to the health-care financing system. Vertical equity can be analysed by measuring progressivity., Objectives: The aim of this study was to analyse progressivity by measuring deviations from proportionality in the relationship between sources of health-care financing and ability to pay using Kakwani indices applied to data from the Kenya Household Health Utilisation and Expenditure Survey 2007., Methods: Concentration indices and Kakwani indices were obtained for the sources of health-care financing: direct and indirect taxes, out of pocket (OOP) payments, private insurance contributions and contributions to the National Hospital Insurance Fund. The bootstrap method was used to analyse the sensitivity of the Kakwani index to changes in the equivalence scale or the use of an alternative measure of ability to pay., Results: The overall health-care financing system was regressive. Out of pocket payments were regressive with all other payments being proportional. Direct taxes, indirect taxes and private insurance premiums were sensitive to the use of income as an alternative measure of ability to pay. However, the overall finding of a regressive health-care system remained., Conclusion: Reforms to the Kenyan health-care financing system are required to reduce dependence on out of pocket payments. The bootstrap method can be used in determining the sensitivity of the Kakwani index to various assumptions made in the analysis. Further analyses are required to determine the equity of health-care utilization and the effect of proposed reforms on overall equity of the Kenyan health-care system., (Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine © The Author 2013; all rights reserved.)

Background: Malaria epidemics remain a serious threat to human populations living in the highlands of East Africa where transmission is unstable and climate sensitive. An existing early malaria epidemic prediction model required further development, validations and automation before its wide use and application in the region. The model has a lead-time of two to four months between the detection of the epidemic signal and the evolution of the epidemic. The validated models would be of great use in the early detection and prevention of malaria epidemics., Methods: Confirmed inpatient malaria data were collected from eight sites in Kenya, Tanzania and Uganda for the period 1995-2009. Temperature and rainfall data for the period 1960-2009 were collected from meteorological stations closest to the source of the malaria data. Process-based models were constructed for computing the risk of an epidemic in two general highland ecosystems using temperature and rainfall data. The sensitivity, specificity and positive predictive power were used to validate the models., Results: Depending on the availability and quality of the malaria and meteorological data, the models indicated good functionality at all sites. Only two sites in Kenya had data that met the criteria for the full validation of the models. The additive model was found most suited for the poorly drained U-shaped valley ecosystems while the multiplicative model was most suited for the well-drained V-shaped valley ecosystem. The +18°C model was adaptable to any of the ecosystems and was designed for conditions where climatology data were not available. The additive model scored 100% for sensitivity, specificity and positive predictive power. The multiplicative model had a sensitivity of 75% specificity of 99% and a positive predictive power of 86%., Conclusions: The additive and multiplicative models were validated and were shown to be robust and with high climate-based, early epidemic predictive power. They are designed for use in the common, well- and poorly drained valley ecosystems in the highlands of East Africa.