Relationship between follow-up rates and treatment outcomes in substance abuse research: more is better but when is "enough" enough?

Abstract

To examine the effects of different follow-up rates on estimates of treatment outcome and predictive models thereof, and to specify participant characteristics associated with tracking difficulty. An observational study using data collected for a randomized, experimental design. The King County Assessment Center in Seattle, Washington, an organization responsible for referral to publicly funded substance abuse treatment. Substance-addicted individuals referred to publicly funded inpatient or outpatient treatment. Standardized self-report instruments measuring substance use, substance use consequences and general functioning. Chart review was used to measure treatment entry and completion. There was a significant association between follow-up difficulty and outcomes related to addiction treatment and later substance use. However, outcome estimates based on 60% of the sample who were easiest to locate were only minimally different from those based on the 90-100% ultimately captured, and predictive models of outcome based on the 60% group were reasonably similar to those based on the final sample. Of baseline characteristics examined, only age was associated with later tracking difficulty. Studies reporting follow-up rates below 70% may produce valid findings and study attrition may be largely unpredictable from participant characteristics at baseline. However, a number of factors such as type of population studied, geographical location of the sample, reasons for loss to follow-up and sample size must be considered when attempting to generalize the findings of this study.