UNDERSTANDING THE VARIABILITY IN THE EFFECTIVENESS OF COMMUNITY HEARTHEALTH-PROGRAMS - A METAANALYSIS

Over the past 25 years, community interventions to reduce cardiovascul ar disease (CVD) have been conducted around the world with very mixed results. This study uses meta-analysis to assess whether the variation in the observed effectiveness of community heart health programs (CHH P) is related to characteristics of the intervention program, the popu lation under study, or the evaluation methods. A CHHP is defined as an y primary prevention program that attempted to reduce the population b urden of CVD by shifting the distribution of risk factors in a general population. To be included in the meta-analysis, a study must have ut ilized a reference group in the evaluation, employed a repeated indepe ndent cross-sectional measurement design, and reported sufficient outc ome information for at least one of four major risk factors: smoking, total cholesterol, blood pressure, and body weight. Results of these s tudies are summarized with the effect size measure ((<(Y)over bar(i1)> - <(Y)over bar(i2)>) - (<(Y)over bar(r1)> - <(Y)over bar(r2)>))s(r1) where Y = outcome measure, S = standard deviation of the outcome measu re, 1 = baseline, 2 = follow-up, i = intervention. and i = reference c ommunity. This measure, which reports the net change in the interventi on group in terms of the variability in the reference population befor e the start of the intervention, permits comparison across different o utcome measures and facilitates the aggregation of effects across stud ies. Generalized least squares regression, which permits the incorpora tion of multiple, dependent effect sizes from a single study, was used to assess the impact of characteristics of the intervention (preventi on strategy, type of mass communication, community organization, and e nvironmental change), the population (setting, gender, year of follow- up measurement), and the evaluation design and implementation (the num ber of communities, matching of communities, the follow-up time, the r esponse rate, and covariate adjustment in the analysis) on the effect sizes. The results of this analysis suggest that the characteristics o f the evaluation method account for much of the heterogeneity in the o utcome of CHHPs, though some intervention characteristics also play a role. (C) 1997 Elsevier Science Ltd.