Differential profiles of crack users in respondent-driven and institutional samples: a three-site comparison.

BACKGROUND/AIM Respondent-driven sampling (RDS) is increasingly applied in social epidemiological surveys among 'hidden populations' of hard drug users. The objective of the present study was to assess whether the profile of frequent crack users recruited through RDS differed from those surveyed in two random institutional samples, i.e. low-threshold opiate substitution treatment (ST) and user rooms (URs). METHODS A total of 1,039 crack users (mean age 45.1 ± 9.1 years; 81.5% males; 49.5% non-Western ethnicity) were interviewed in three Dutch cities, using each sampling strategy in each city. Characteristics of respondents in the three samples (per city and aggregated) were compared. RESULTS Crack users in the RDS sample were more likely to be younger and less likely to be poorly educated, have a long crack-using career (≥10 years), use opiates and attend ST than those in both of the other samples. The RDS and UR samples showed the most differences, with more female and Western crack users in the RDS group, and UR participants were more likely to have been arrested and to have entered inpatient drug treatment in the last year. CONCLUSION RDS resulted in a different profile of crack users. It is a valuable methodology to achieve a more diverse representation of crack-using populations than institutional random samples.