The efficiency of the matched-pairs design of The Community Intervention Trial for Smoking Cessation (COMMIT)

Abstract The Community Intervention Trial for Smoking Cessation (COMMIT) was a randomized trial to evaluate the effects of a community-wide smoking cessation intervention on smoking behavior. The statistical design involved 22 pair-matched communities and the randomization of one community in each of the 11 pairs to the intervention, with the other community in the pair acting as a comparison. Communities were matched on the basis of their geographical proximity and similarity of demographic composition. In this paper, we use the data on the rates of quitting smoking among cohorts of heavy and light/moderate smokers in each community to estimate the gains in efficiency achieved by the matched-pairs design compared to an unmatched randomized trial. We find evidence of some gain in efficiency, although the data are not extensive enough to give estimates of efficiency gain that have good precision.