Reliability and validity of the Spanish RBA: Translation issues, protocol drift, and drug distribution patterns

BACKGROUND: The NIDA RBA English interview was found to be reliable and valid; however, the Spanish version had not been tested.

METHOD: Participants were 22,755 English-speaking and 1,946 Spanish-speaking active drug users who completed the RBA drug-use items. Internal consistency of self-reported items was assessed to evaluate reliability. To evaluate validity, self-reported 48-hour use was compared to urine test results.

RESULTS: For both English and Spanish RBAs, internal consistency indicated almost perfect reliability for crack (kappa=.997, .998), cocaine (kappa=.968, .990), and heroin (kappa=.995, .992) use. Self-reported 48-hour use demonstrated substantial validity for the English RBA, with 88.5% agreement (kappa=.617) for cocaine, and 91.4% agreement (kappa= .808) for opiates. For the Spanish RBA, validity of self-reported use was only slight for cocaine, but substantial for opiates, with 65.5% and 85.7% agreement, respectively (kappa=.189, .676).

DISCUSSION: The Spanish RBA was found to be as reliable as the English RBA; however, the Spanish instrument was insufficiently valid for self-reported cocaine use, possibly contributing to null findings. This study highlights the importance of testing translated versions of instruments with known psychometric properties, examining multi-site data by site, establishing procedures to minimize “protocol drift”, and also the influence of potential geographic differences in drug distribution and the effect of rapidly changing, region-specific drug terms on drug use self-report. Funding should be made available to adequately test instruments in all languages into which they have been translated. The psychometric properties of translated instruments should be documented and/or published prior to their broader use.