BACKGROUND: Rates of self-reported sexually transmitted diseases (STD) among drug-using prostitutes in Amsterdam have been shown to be declining. However, self reports may provide biased estimates. METHODS: We determined the prevalence of risk indicators for and time trends in incorrect self-reported STD in a cohort of drug-using prostitutes. Rates of self-reported STD were validated with diagnosed STD from a special STD clinic for drug-using prostitutes. RESULTS: Between 1986 and 1992, 314 STD were diagnosed in 85 (65%) of 131 studied prostitutes. Overall, 34% of diagnosed STD was not reported in the cohort, referred to as underreporting. For 40% of reported STD, no diagnosis was obtained ('overreporting'). In spite of these misclassifications, absolute numbers of reported and diagnosed STD practically balanced. Underreporting was independently associated with longer residence in Amsterdam (OR = 2.34 per 5 years), more than six clinic visits between two cohort visits (OR = 0.31), daily use of non-parenteral drug (OR = 8.45) and mainly injecting heroin (OR = 6.93); overreporting was independently associated only with a lower frequency of clinic visits. Multivariate analyses showed that trends in underreporting and overreporting were in part explained by a higher number of visits in the cohort and by lower numbers of clinic visits between two cohort visits. CONCLUSIONS: Self-reporting of STD in our cohort is unreliable and misclassification prevalent; risk factors for and associates of self-reported STD are biased. Inferences drawn from self reports are potentially misleading and therefore such data need to be validated. Trends in misclassification can be interpreted as recall errors and as a study participation effect. For monitoring of STD prevalences in relation to trends in sexual risk behaviours, diagnosed STD should be used in preference to self reports.
BACKGROUND: Rates of self-reported sexually transmitted diseases (STD) among drug-using prostitutes in Amsterdam have been shown to be declining. However, self reports may provide biased estimates. METHODS: We determined the prevalence of risk indicators for and time trends in incorrect self-reported STD in a cohort of drug-using prostitutes. Rates of self-reported STD were validated with diagnosed STD from a special STD clinic for drug-using prostitutes. RESULTS: Between 1986 and 1992, 314 STD were diagnosed in 85 (65%) of 131 studied prostitutes. Overall, 34% of diagnosed STD was not reported in the cohort, referred to as underreporting. For 40% of reported STD, no diagnosis was obtained ('overreporting'). In spite of these misclassifications, absolute numbers of reported and diagnosed STD practically balanced. Underreporting was independently associated with longer residence in Amsterdam (OR = 2.34 per 5 years), more than six clinic visits between two cohort visits (OR = 0.31), daily use of non-parenteral drug (OR = 8.45) and mainly injecting heroin (OR = 6.93); overreporting was independently associated only with a lower frequency of clinic visits. Multivariate analyses showed that trends in underreporting and overreporting were in part explained by a higher number of visits in the cohort and by lower numbers of clinic visits between two cohort visits. CONCLUSIONS: Self-reporting of STD in our cohort is unreliable and misclassification prevalent; risk factors for and associates of self-reported STD are biased. Inferences drawn from self reports are potentially misleading and therefore such data need to be validated. Trends in misclassification can be interpreted as recall errors and as a study participation effect. For monitoring of STD prevalences in relation to trends in sexual risk behaviours, diagnosed STD should be used in preference to self reports.