Standard care quality determines treatment outcomes in control groups of HAART-adherence intervention studies: implications for the interpretation and comparison of intervention effects

Abstract

Objective: Clinical trials of behavioral interventions seek to enhance evidence-based health care. However, in case the quality of standard care provided to control conditions varies between studies and affects outcomes, intervention effects cannot be directly interpreted or compared. The objective of the present study was to examine whether standard care quality (SCQ) could be reliably assessed, varies between studies of highly active antiretroviral HIV-adherence interventions, and is related to the proportion of patients achieving an undetectable viral load ("success rate").

Design: Databases were searched for relevant articles. Authors of selected studies retrospectively completed a checklist with standard care activities, which were coded to compute SCQ scores. The relationship between SCQ and the success rates was examined using meta-regression.

Main outcome measures: Cronbach's alpha, variability in SCQ, and relation between SCQ and success rate.

Results: Reliability of the SCQ instrument was high (Cronbach's alpha = .91). SCQ scores ranged from 3.7 to 27.8 (total range = 0-30) and were highly predictive of success rate (p = .002).

Conclusions: Variation in SCQ provided to control groups may substantially influence effect sizes of behavior change interventions. Future trials should therefore assess and report SCQ, and meta-analyses should control for variability in SCQ, thereby producing more accurate estimates of the effectiveness of behavior change interventions.