Response errors and other problems of health interview surveys in developing countries.

The conceptual framework for health care use analysis may serve as an example of an attempt to modify the models from industrialized countries and adapt them to the specific requirements of developing countries. This article presents some methodological achievements of health interview surveys in the control of response errors in industrial countries and addresses the question of their limitations when applied to developing countries. This is examined in relation to illness reporting in the reporting on health services use and in some special problem areas of health interview surveys. Other health-related surveys such as nutrition fertility or mortality surveys are not considered. Validity and consistency tests of illness reporting are problematic. However it has been argued that a clinical examination is a sufficiently accurate check on the validity of illness reporting and this approach has been attempted in rural Africa and in a rural part of the Republic of Korea. The Cornell Medical Index consisting of Yes-No questions is better for measuring the general health status particularly emotional health rather than the presence of specific diseases in individuals or populations. Validity of illness reporting can be improved by using a list of tracer conditions by reinterviewing to evaluate the interviewees recall period and compare answers by using individual rather than proxy reporting avoiding or giving special treatment to delicate questions by avoiding difficult questions and by using professional rather than lay interviewers. Since severe diseases are more easily remembered than mild ones the analysis of morbidity data must consider different levels of severity. This is particularly important when relating health services use self-care and drug-use to perceived illness. Validity testing of health services use seems to be comparatively easy because peoples real utilization of a particular health facility may be used as an absolute yardstick. However problems still arise. 2 main types of response errors exist in the field of reporting on health services use: recall errors and barriers towards reporting on use of the informal health system. An indirect validity test of a reported consultation of medical facilities has been attempted in rural India; the number of consultations was checked against that elicited in a cross-sectional study. Obtaining valid answers is especially pronounced in the self-care area the most frequent form of health care. Quantification of multiple use of services is also problematic. Many errors can be avoided by using interviewers of ethnic and social backgrounds similar to the study population and by limiting the number of interviews per interviewer.