While the consistent pattern across these studies suggests a robust finding, some limitations should be noted. Most studies were based on small samples, some involved bivariate analyses only, and only two assessed interactions of race by treatment needed to test differential treatment effects across race. Moreover, the follow-up period was short, typically 12 months or less, and one study observed an attenuation of positive treatment effects among blacks after the 6-month follow-up, suggesting the need for longer-term studies (McKay et al., 2003). Finally, the more rigorous of these studies involved randomized clinical trials, the results of which may not generalize well to disadvantaged populations (2000). As Humphreys and Weisner (2000) point out, use of common exclusion criteria for clinical trials can yield study samples with lower problem severity than that of excluded persons. Moreover, these exclusion criteria disproportionately exempt minorities and persons of low socioeconomic position.
