Trials which randomize practices II: sample size

BACKGROUND: When practices are randomized in a trial and observations aremade on the patients to assess the relative effectiveness of the differentinterventions, sample size calculations need to estimate the number ofpractices required, not just the total number of patients. OBJECTIVE: Ouraims were to introduce the methodology for appropriate sample sizecalculation and discuss the implications for power. METHOD: A workedexample from general practice is used. DISCUSSION: Designs which randomizepractices are less powerful than designs which randomize patients tointervention groups, particularly where a large number of patients isrecruited from each practice. Studies which randomize few practices shouldbe avoided if possible, as the loss of power is considerable and simplerandomization may not ensure comparability of intervention groups.