Snow, B.

Verbal autopsies are widely used to describe the causes of death in individuals who die outside of hospital or clinic settings, but are seldom validated. The technique assumes that disease which cause death can be easily distinguished from one another by distinct syndromes and that these are reported accurately by lay respondents. The article describes the potential problems of syndrome definition and the likely biases from poor recognition and recall by bereaved relatives; how these might be tested and what can be done when verbal autopsy cannot identify the cause of death.

Effective health planning requires good quality data, but many health facilities lack the ability to provide this. Health questions often have to be answered within specific research studies. Microcomputers are now generally recommended and used by researchers for data analysis at the end of projects. The article reviews the use of microcomputer based management of data collection during a study. A selection of pojects are described, all of which have used microcomputers in a decentralised fachio, closer to the point of data collection.