Adult mortality: time for a reappraisal.

BACKGROUND In many countries, little is known about adult mortality rates. New innovations are necessary to develop reasonable estimates from available information. One readily available resource is household survey data. While birth histories collected in surveys have produced reasonable estimates of child mortality, the use of sibling survival data collected in similar household surveys has not been comprehensively analysed, largely because of concerns of underreporting. METHODS This paper uses sibling survival schedules from 29 Demographic and Health Surveys (DHS) to generate estimates of under-5 mortality and of the summary measure of adult mortality 45q15-the probability of dying between ages 15 and 59. These are then compared with UN child and adult mortality estimates. RESULTS Sibling history data collected in these household surveys seems to contain adequate information to estimate adult mortality rates, though there are problems with underreporting. The correlation coefficient between UN estimates and DHS estimates is 0.74 for adult mortality, indicating a strong relationship between the two but suggesting there may be underreporting of adult deaths in the survey data. CONCLUSIONS Further investigation is necessary to determine the usefulness of household survey data for the estimation of adult mortality. New survey instruments like the World Health Survey have incorporated questions to help correct for underreporting in sibling histories. Further analyses need to be carried out in countries where vital registration data are also available, to determine how well household survey data do in estimating adult mortality and whether improvements in the survey instrument adequately correct for underreporting of deaths.