Monitoring for outbreak associated excess mortality in an African city: Detection limits in Antananarivo, Madagascar

Abstract

Quantitative estimates of the impact of infectious disease outbreaks are required to develop measured policy responses. In many low- and middle-income countries, inadequate surveillance and incompleteness of death registration are important barriers. Here, we characterize how large an impact on mortality would have to be to be detectable using the uniquely detailed mortality notification data from the city of Antananarivo in Madagascar, with application to a recent measles outbreak. The weekly mortality rate of children during the 2018-2019 measles outbreak was 154% above the expected value at its peak, and the signal can be detected earlier in children than in the general population. This approach to detecting anomalies from expected baseline mortality allows us to delineate the prevalence of COVID-19 at which excess mortality would be detectable with the existing death notification system in the capital of Madagascar. Given current age-specific estimates of the COVID-19 fatality ratio and the age structure of the population in Antananarivo, we estimate that as few as 11 deaths per week in the 60-70 years age group (corresponding to an infection rate of approximately 1%) would detectably exceed the baseline. Data from 2020 will undergo necessary processing and quality control in the coming months. Our results provide a baseline for interpreting this information.