Comparison of two methods for assessing child mortality in areas without comprehensive registration systems

We compared the effectiveness of 2 approaches to assessing child deaths in a rural area of Burkina Faso, West Africa. Censuses, repeated yearly, identified 410 child deaths in the age range 6-59 months. Surveillance using community informants identified only 319 deaths. The estimated sensitivities of the 2 systems were 97% and 76%, respectively. Both systems appeared less effective at detecting child deaths before 6 months of age (sensitivities 74% and 57%). Most of the deaths missed by the census were of children born since the previous census. The cost of one year's surveillance was twice that of a single census. The marginal cost of the surveillance system per additional child death identified between 6 and 59 months was about US$ 1500. Thorough annual censuses may be sufficient to ascertain almost all child deaths over 6 months of age. In studies wishing to identify child deaths before 6 months of age, such an approach is unlikely to be adequate. In such situations, our data indicated that the use of unpaid community informants can improve assessment of deaths. Where accurate assessment of early infant death rate is required, regular visits to each household by members of the study team are likely to be the only reliable approach.