Process and Methodological Considerations for Observational Analyses of Vector Control Interventions in Sub-Saharan Africa Using Routine Malaria Data.

Progress in malaria control has stalled in recent years. With growing resistance to existing malaria vector control insecticides and the introduction of new vector control products, national malaria control programs (NMCPs) increasingly need to make data-driven, subnational decisions to inform vector control deployment. As NMCPs are increasingly conducting subnational stratification of malaria control interventions, including malaria vector control, country-specific frameworks and platforms are increasingly needed to guide data use for vector control deployment. Integration of routine health systems data, entomological data, and vector control program data in observational longitudinal analyses offers an opportunity for NMCPs and research institutions to conduct evaluations of existing and novel vector control interventions. Drawing on the experience of implementing 22 vector control evaluations across 14 countries in sub-Saharan Africa, as well as published and gray literature on vector control impact evaluations using routine health information system data, this article provides practical guidance on the design of these evaluations, makes recommendations for key variables and data sources, and proposes methods to address challenges in data quality. Key recommendations include appropriate parameterization of impact and coverage indicators, incorporating explanatory covariates and contextual factors from multiple sources (including rapid diagnostic testing stockouts; insecticide susceptibility; vector density measures; vector control coverage, use, and durability; climate and other malaria and non-malaria health programs), and assessing data quality before the evaluation through either on-the-ground or remote data quality assessments. These recommendations may increase the frequency, rigor, and utilization of routine data sources to inform national program decision-making for vector control.

Using routine health data to evaluate the impact of indoor residual spraying on malaria transmission in Madagascar.

INTRODUCTION: Indoor residual spraying (IRS) and insecticide-treated bed nets (ITNs) are cornerstone malaria prevention methods in Madagascar. This retrospective observational study uses routine data to evaluate the impacts of IRS overall, sustained IRS exposure over multiple years and level of spray coverage (structures sprayed/found) in nine districts where non-pyrethroid IRS was deployed to complement standard pyrethroid ITNs from 2017 to 2020. METHODS: Multilevel negative-binomial generalised linear models were fit to estimate the effects of IRS exposure overall, consecutive years of IRS exposure and spray coverage level on monthly all-ages population-adjusted malaria cases confirmed by rapid diagnostic test at the health facility level. The study period extended from July 2016 to June 2021. Facilities with missing data and non-geolocated communes were excluded. Facilities in IRS districts were matched with control facilities by propensity score analysis. Models were controlled for ITN survivorship, mass drug administration coverage, precipitation, enhanced vegetation index, seasonal effects and district. Predicted cases under a counterfactual no IRS scenario and number of cases averted by IRS were estimated using the fitted models. RESULTS: Exposure to IRS overall reduced case incidence by an estimated 30.3% from 165.8 cases per 1000 population (95% CI=139.7 to 196.7) under a counterfactual no IRS scenario, to 114.3 (95% CI=96.5 to 135.3) over 12 months post-IRS campaign in nine districts. A third year of IRS reduced malaria cases 30.9% more than a first year (incidence rate ratio (IRR)=0.578, 95% CI=0.578 to 0.825, p<0.001) and 26.7% more than a second year (IRR=0.733, 95% CI=0.611 to 0.878, p=0.001). There was no significant difference between the first and second year (p>0.05). Coverage of 86%-90% was associated with a 19.7% reduction in incidence (IRR=0.803, 95% CI=0.690 to 0.934, p=0.005) compared with coverage ≤85%, although these results were not robust to sensitivity analysis. CONCLUSION: This study demonstrates that non-pyrethroid IRS appears to substantially reduce malaria incidence in Madagascar and that sustained implementation of IRS over three years confers additional benefits.

Spatiotemporal mapping of malaria prevalence in Madagascar using routine surveillance and health survey data.

Malaria transmission in Madagascar is highly heterogeneous, exhibiting spatial, seasonal and long-term trends. Previous efforts to map malaria risk in Madagascar used prevalence data from Malaria Indicator Surveys. These cross-sectional surveys, conducted during the high transmission season most recently in 2013 and 2016, provide nationally representative prevalence data but cover relatively short time frames. Conversely, monthly case data are collected at health facilities but suffer from biases, including incomplete reporting and low rates of treatment seeking. We combined survey and case data to make monthly maps of prevalence between 2013 and 2016. Health facility catchment populations were estimated to produce incidence rates from the case data. Smoothed incidence surfaces, environmental and socioeconomic covariates, and survey data informed a Bayesian prevalence model, in which a flexible incidence-to-prevalence relationship was learned. Modelled spatial trends were consistent over time, with highest prevalence in the coastal regions and low prevalence in the highlands and desert south. Prevalence was lowest in 2014 and peaked in 2015 and seasonality was widely observed, including in some lower transmission regions. These trends highlight the utility of monthly prevalence estimates over the four year period. By combining survey and case data using this two-step modelling approach, we were able to take advantage of the relative strengths of each metric while accounting for potential bias in the case data. Similar modelling approaches combining large datasets of different malaria metrics may be applicable across sub-Saharan Africa.