Abstract

An assessment of the impact of malaria intervention methodological issues and the use of verbal autopsies.

Snow RW
Evaluation of the impact of health interventions. 1995;

Permenent descriptor

This chapter reviews data on malaria-specific mortality in Africa in order to assess various approaches to estimating the contribution of malaria to childhood mortality. The review emphasizes the most frequently used methods: population laboratories and verbal autopsy surveys. Alternative methods which are cheaper yet still suited for national malaria monitoring are presented. After its introductory section, the chapter looks at hospital-based studies of malaria mortality and presents details about the data collection procedures used by the Kilifi District Hospital in Kenya. The next section concentrates on the identification of childhood mortality events by population laboratories and the use of verbal autopsies to attribute a cause of death to malaria. The practical issues surrounding the use of population laboratories are described, including delineating and describing a population and the use of follow-up surveys. The sensitivities, specificities, and predictive values of verbal autopsies as compared to hospital-based data are discussed, and the results of a study in Kilifi which gleaned the percentage of true positive and false positive recall rates in postneonatal deaths are presented. The proposed alternative designs for mortality measurement include 1) retrospective single round surveys, 2) hospital surveillance, 3) the use of civil registration systems, 4) upgraded civil registration, 5) the use of village health workers, 6) the use of school children to report on deaths, and 7) indirect demographic techniques such as those proposed by various researchers. It is concluded that whereas reductions in childhood mortality must be the goal of any new antimalarial intervention in Africa, surprisingly little is known about the true levels of childhood malaria mortality or the basic epidemiology of malaria. Use of total childhood mortality estimates may prove to be more indicative of malaria-specific interventions than verbal autopsy studies. Hospital data may be a useful surveillance tool in identifying regions where malaria control should be targeted. While carefully controlled randomized trials are only possible within the context of well-defined populations or population laboratories, cheaper methods can be developed to provide a means of continuously assessing intervention success after the implementation has been allowed to proceed in an uncontrolled manner.