Attrition and related trends in scientific rigor: a score card for ART adherence intervention research and recommendations for future directions.

Scientific rigor in intervention trials is frequently used in systems that identify effective interventions for dissemination. In these systems, and in work that synthesizes bodies of research, percent attrition is often considered a fatal threat to validity. However, differential attrition, versus percent total, is of primary concern. Key methodologic, design, and analytic issues pertaining to scientific rigor in longitudinal designs were identified, and the current literature on antiretroviral therapy adherence interventions (k = 51) was evaluated in relation to these. Although results suggest that this body of literature has progressed in rigor, improvements are needed in transparency of reporting participant flow, retention strategies, handling of missing data, and characterization of retained and lost cohorts. Attrition averaged 30% total, and differential by study arm was estimated at 9%. Differential attrition continues to be underreported and is not well represented by the more frequently used, though arguably less appropriate, metric of overall percent attrition.