Can we rely on published reports of control group support? Review of smoking cessation trials

Abstract

Background: Control groups in behaviour change trials often receive some form of behavioural support, which may vary between studies and influence trial effect sizes. Reporting of the content of this support in published papers is often minimal. We examined the quality of control group descriptions in smoking cessation trials and the value of a novel instrument for assessing the content of behavioural support provided to the control groups. Method: As part of a systematic review, authors (first, second, last) of trials (k=148) were requested to complete a specially-developed, 60-item control group support checklist and send any additional materials on control group support. Two independent coders examined published articles, supplementary materials and the checklist for the presence of behaviour change techniques (BCTs). Descriptive and reliability analyses were conducted. Results: The number of control group BCTs delivered was 1775 (MBCTs=14.20, SDBCTs =14.03) and varied within and between types of controls (e.g., usual care (MBCTs=16.83, SDBCTs=15.27) versus brief advice (MBCTs=5.33, SDBCTs=3.45)). Authors of 60 trials completed the control group checklist (α=0.97); 59.3% (786) of their BCTs were identified through this checklist, 19.7% in other materials sent, and 21.1% from publically-available materials. Across 148 trials, 34% (606) of coded BCTs could be extracted from published materials. Discussion: Reporting of control group support in smoking cessations trials is often poor. The proposed methodology for collecting additional information from study authors seems both reliable and valuable, revealing almost 80% of the control group BCTs. Systematic reviewers should consider adopting a similar methodology in their projects.