The validity of drug users' self-reports in a non-treatment setting: prevalence and predictors of incorrect reporting methadone treatment modalities

Background Epidemiological studies among drug users are often based on retrospective self-reports. However, among others, memory failure, being under the influence of drugs, psychopathology, misunderstanding of questions and socially desirable answering may generate inaccurate reporting. Methods This study validated self-reported current (methadone dosage) and medium-term (main location of methadone dispensing and frequency of methadone programme attendance over the previous 4-6 months) aspects of methadone treatment in the Amsterdam AIDS cohort study among drug users, using data of the Central Methadone Register. In addition to descriptive measures, logistic regression analysis was used (adjusted for intra-individual correlation) to identify subgroups with incorrect reporting. Data collected at 4406 visits of SOS cohort participants were analysed. Results Current methadone dosage was accurately reported (unweighted kappa [kappa]: 0.94, weighted kappa [kappa(W)]: 0.97). A low methadone dosage; short duration of school education and depressive or euphoric mood during the interview were significant and independent predictors of incorrect reporting of methadone dosage. For main location of dispensing kappa was 0.82, for frequency of programme attendance kappa was 0.53 and kappa(W) 0.87. There was a tendency to reporting the extreme answering categories. Infrequent programme attendance was the only significant predictor of incorrectly reporting frequency of programme attendance. Conclusions Drug users are able to give valid self-reports in a setting where social desirability does not play an important role. The main reasons of incorrect reporting were impaired cognitive functioning, memory failure and misunderstanding of questions.