A REVIEW OF DATA-DERIVED METHODS FOR ASSIGNING CAUSES OF DEATH FROM VERBAL AUTOPSY DATA

Background. Verbal autopsy (VA) is an indirect method for estimating c ause-specific mortality. In most previous studies, cause of death has been assigned from verbal autopsy data using expert algorithms or by p hysician review. Both of these methods may have poor validity. In addi tion, physician review is time consuming and has to be carried out by doctors, A range of methods exist for deriving classification rules fr om data. Such rules are quick and simple to apply and in many situatio ns perform as well as experts. Methods. This paper has two aims. First , it considers the advantages and disadvantages of the three main meth ods for deriving classification rules empirically; (a) linear and othe r discriminant techniques, (b) probability density estimation and (c) decision trees and rule-based methods. Second, it reviews the factors which need to be taken into account when choosing a classification met hod for assigning cause of death from VA data. Results. Four main fact ors influence the choice of classification method: (a) the purpose for which a classifier is being developed, (b) the number of validated ca uses of death assigned to each case, (c) the characteristics of the VA data and (d) the need for a classifier to be comprehensible. When the objective is to estimate mortality from a single cause of death, logi stic regression should be used. When the objective is to determine pat terns of mortality, the choice of method will depend on the above fact ors in ways which are elaborated in the paper. Conclusion. Choice of c lassification method for assigning cause of death needs to be consider ed when designing a VA validation study. Comparison of the performance of classifiers derived using different methods requires a large VA da taset, which is not currently available.