Measuring Health Care Needs and Coverage on a Probabilistic Scale in Population Surveys

Ajay Tandon, Harvard University
Christopher J.L. Murray, World Health Organization (WHO)
Bakuti Shenghelia, World Health Organization (WHO)

Health system coverage is defined as the likelihood that an individual receives an appropriate intervention given that he/she has a need for it. It is usually not easy to estimate health care needs in survey settings, making it difficult to compute coverage estimates. This paper proposes a method to estimate disease prevalence--and an individual's likelihood of having a given disease or condition--based on their responses to a series of categorical-response symptomatic questions. We build on earlier work on this area by incorporating information on the sensitivity and specifity of symptomatic questions derived from a separate sample of respondents with clinical diagnosed conditions (matched with those that did not have the condition). We are able to demonstrate that this method outperforms standard algorithms that are used to classify respondents as diseased or not. Plus it provides an easy way to compute coverage estimates from population surveys.

  See paper

Presented in Session 152: Health Care Policy and Access