Assessing the accuracy of adherence and sexual behaviour data in the MDP301 vaginal microbicides trial using a mixed methods and triangulation model.

BACKGROUND: Accurate data on adherence and sexual behaviour are crucial in microbicide (and other HIV-related) research. In the absence of a "gold standard" the collection of such data relies largely on participant self-reporting. The Microbicides Development Programme has developed a mixed method/triangulation model for generating more accurate data on adherence and sexual behaviour. METHODOLOGY/PRINCIPAL FINDINGS: Data were collected from a random subsample of 725 women using structured case record form (CRF) interviews, coital diaries (CD) and in-depth interviews (IDI). Returned used and unused gel applicators were counted and additional data collected through focus group discussions and ethnography. The model is described in detail in a companion paper [1]. When CRF, CD and IDI are compared there is some inconsistency with regard to reporting of sexual behaviour, gel or condom use in more than half. Inaccuracies are least prevalent in the IDI and most prevalent in the CRF, where participants tend to under-report frequency of sex and gel and condom use. Women reported more sex, gel and condom use than their partners. IDI data on adherence match the applicator-return data more closely than the CRF. The main reasons for inaccuracies are participants forgetting, interviewer error, desirability bias, problems with the definition and delineation of key concepts (e.g. "sex act"). Most inaccuracies were unintentional and could be rectified during data collection. CONCLUSIONS/SIGNIFICANCE: The CRF--the main source of self-report data on behaviour and adherence in many studies--was the least accurate with regard to measuring sexual behaviour, gel and condom use. This has important implications for the use of structured questionnaires for the collection of data on sexual behaviour and adherence. Integrating in-depth interviews and triangulation into clinical trials could increase the richness and accuracy of behavioural and adherence data