Improving estimation of vaccine effectiveness during outbreaks in low-resource settings: A case study of oral cholera vaccination during the 2022-2023 cholera outbreak in Malawi

Background Use of oral cholera vaccine (OCV) is globally recommended as a public health response to cholera outbreaks, alongside water, sanitation and hygiene (WASH) interventions. Estimating vaccine effectiveness during emergencies in low–and middle–income countries is challenging because vaccination campaigns are often implemented over short time frames, while individual–level data are frequently incomplete due to constraints in infrastructure, resources and data systems. There is a need for pragmatic approaches that can generate timely, policy–relevant evidence using routinely collected data. Methods We analysed routine surveillance data from a large 2022–2023 cholera outbreak in Blantyre District, Malawi. The EpiEstim framework was used to generate estimates of the time-varying reproduction number (Rt) from line–listed case data. We modelled changes in Rt as a function of cumulative OCV coverage using a log–linear framework and propagated uncertainty through posterior sampling. Lagged WASH exposure variables were incorporated in the model to generate adjusted vaccine effectiveness estimates and to explore potential interaction effects. Sensitivity analyses assessed robustness to alternative lag structures. Findings The Blantyre outbreak was characterised by an initial period of low–level transmission followed by a sharp increase in cases from late November 2022, after which transmission declined steadily through April 2023. This decline coincided with the implementation of a reactive OCV campaign. The majority of the cases were among middle-aged men living in urban Blantyre. The unadjusted vaccine-associated reduction in transmission was estimated at 53.52% (95% credible interval (CrI):42.5–64.1%). After adjusting for a 7–day rolling average WASH activity, total vaccine effectiveness increased to 62.1% (95% CrI: 49.3–74.9%). Sensitivity analyses using alternative lag structures for WASH exposure produced comparable adjusted estimates. Interpretation Implementation of OCV contributed to a substantial reduction in cholera transmission during the outbreak. This study demonstrates a feasible approach for estimating vaccine-attributable impact whilst accounting for public health and social measures, such as WASH interventions. The methods described will be useful in outbreaks where classical observational designs are not possible, providing actionable evidence to policy makers for outbreak response in resource-limited settings.