Change over time in alcohol consumption in control groups in brief intervention studies: systematic review and meta-regression study

Abstract

Reactivity to assessment has attracted recent attention in the brief alcohol intervention literature. This systematic review sought to examine the nature of change in alcohol consumption over time in control groups in brief intervention studies. Primary studies were identified from existing reviews published in English language, peer-reviewed journals between 1995 and 2005. Change in alcohol consumption and selected study-level characteristics for each primary study were extracted. Consumption change data were pooled in random effects models and meta-regression was used to explore predictors of change. Eleven review papers reported the results of 44 individual studies. Twenty-six of these studies provided data suitable for quantitative study. Extreme heterogeneity was identified and the extent of observed reduction in consumption over time was greater in studies undertaken in Anglophone countries, with single gender study participants, and without special targeting by age. Heterogeneity was reduced but was still substantial in a sub-set of 15 general population studies undertaken in English language countries. The actual content of the control group procedure itself was not predictive of reduction in drinking, nor were a range of other candidate variables including setting, the exclusion of dependent drinkers, the collection of a biological sample at follow-up, and duration of study. Further investigations may yield novel insights into the nature of behaviour change with potential to inform brief interventions design.

Introduction

Brief alcohol interventions usually involve feedback on risk status and advice or brief counselling on change (Heather, 1995). There are numerous reviews attesting to effectiveness in this large research literature. Unexpected reductions in drinking in control groups have been repeatedly identified in these reviews (see for example Bien et al., 1993, Kaner et al., 2007). This phenomenon is not straightforward to interpret as there are a number of possible explanations. Fleming et al. (1997), for example, commented in their trial that

“The 20% reduction in alcohol use in the control group was interesting. A majority of the other trials have similar reductions in alcohol use. The reason for this change is unknown but may be related to regression to the mean, historical changes in alcohol use, and the intervention effect of the research procedures. It is our impression that research procedures can have a significant intervention effect.”

Dedicated experimental study of this phenomenon has recently developed. Assessment reactivity has been evaluated and effects of different types and magnitudes have been identified (Carey et al., 2006, Clifford et al., 2007, Kypri et al., 2007). An effect of simply completing a brief screening questionnaire and thereby becoming aware of a study focus on drinking behaviour has also been identified McCambridge and Day (2008). These studies indicate the possible value of screening or other forms of brief assessment as a brief intervention even in the absence of feedback. If control groups are exposed to assessment procedures that are themselves components of brief interventions, effectiveness may have been underestimated (Kypri et al., 2007).

These recent randomised studies of previous observations of possible effects of assessment have potentially important implications for the interpretation of the existing evidence-base on brief interventions. Whilst post hoc commentary on unexpected change in control groups is entirely appropriate, such observations do not constitute a strong form of evidence nor are they straightforward to interpret (Finney, 2008). It has not yet been established that reduced drinking in control groups has been consistently observed in the existing literature, as this possibility has not been the subject of dedicated quantitative study. We identified surprisingly high levels of self-reported smoking cessation in a previous review of this literature McCambridge and Jenkins (2008). For these reasons, we decided to systematically evaluate change over time in control groups in brief intervention studies, and to examine study-level predictors of change in alcohol consumption.