Abstract

PIP: This analysis of the methodological problems in measuring access to health care in Africa is based on a survey of the literature and the author's personal experience in Tunisia, Burkina Faso, and Senegal. 3 data bases were consulted for the previous decade and authors' references were verified. 18 published works were identified. Salient data from them are summarized in 2 tables, one covering methods and the other present findings. Surveys conducted by ministries of health, research institutions, and international organizations not leading to publication in scientific reviews were excluded. The essential argument is that methodological problems preclude comparison of the results of different surveys. Surveys of access to health care have different major objectives, which are less similar than they appear. They are either morbidity surveys which study the frequency and distribution of illness and secondarily the therapeutic practices for different pathologies or they are behavior surveys attempting to study recourse to care and its conditioning factors without reference to epidemiology. Different methodological problems are specific to each type of study. For morbidity surveys, sociocultural factors intervene in recognition of illness, individuals vary in their declaration of symptoms, and the least troublesome episodes are forgotten. The choice of open-ended question or list of symptoms influences the outcome. Results of surveys of recourse to care are influenced by differences between potential accessibility and true access, discrepancies between stated preferences and actual use, and dissimulation about use of therapies considered less legitimate. Questions about a specific illness episode provide a concrete support and enable data on all care practices to be elicited, including self-medication and nontreatment. The social relationship between the interviewer and the respondent influences responses and thus quality of information. Social and cultural factors such as ethnic group and literacy status and even political pressures are especially important in Africa, but data on the interviewer and respondents are not always specified. The language used may be significant if respondents do not fully understand the questions. Few surveys concern themselves with the validity of responses despite the opportunities for inaccuracy. Only 1 of the African surveys contained a systematic analysis of biases in responses. The variability of sampling techniques and procedures also hampers comparison of results. Different systems of classifying types of care available or omission of some types of care further complicate efforts to compare results of different surveys. And surveys of access to care do not take into account the quality of care offered or the efficacy of treatments sought.