Impact of participant and physician intervention preferences on randomized trials: a systematic review

Abstract

Context: Allocation on the basis of randomization rather than patient choice is the gold standard of unbiased estimates of efficacy in clinical medicine. However, randomly allocating patients to treatments that do not accord with their preferences may influence internal and external validity.

Objective: To determine whether preferences affect recruitment to trials (external validity) and outcomes in trials (internal validity).

Data sources: We searched MEDLINE, EMBASE, PsycINFO, CINAHL, AMED, and the Cochrane Library for articles published between 1966 and September 2004. We also hand-searched several major medical journals, searched reference lists of relevant articles, and contacted authors of published preference designs. The 2 themes in the first filter of the search strategy were preferences and possible determinants of preferences.

Study selection: Comprehensive cohorts and 2-stage trials that measured or recorded patient or physician preference, included allocation of participants to random and preference cohorts, and followed up all participants. We excluded trials with no recording of preference; of decision aids; with measurements of preferences for economic analyses; in which patients who refused randomization were followed up without reference to preferences; and of nonclinical populations.

Data extraction: Up to 4 reviewers independently evaluated the articles, and disagreements were resolved at project steering group meetings. We extracted data on study design, measurement of preference, recruitment, attrition, and summary data on the primary outcome(s) at baseline and each follow-up point.

Data synthesis: Of 10,023 citations identified, 170 articles met screening criteria and 32 (27 comprehensive cohorts and 5 two-stage trials) were determined to be eligible and were used in the final review. Although treatment preferences led to a substantial proportion of people refusing randomization, there was less evidence of bias in the characteristics of individuals agreeing to be randomized. Differences in outcome across the trials between randomized and preference groups were generally small, particularly in large trials and after accounting for baseline measures of outcome. Therefore, there was little evidence that preferences substantially interfere with the internal validity of randomized trials.

Conclusions: Preferences influence whether people participate in randomized trials, but there is little evidence that they significantly affect validity.