Method: Through a search with Google and Google Scholar, we identified a total of 30 DR evaluation studies, the majority of which were non-peer-reviewed evaluation reports. Seventeen out of the 30 studies were included in the meta-analysis because they: a) contained a comparison between AR and IR groups; b) included an outcome measure of re-reports, substantiated re-reports, or out of home placements; and c) reported the necessary statistics to compare the effect size, namely the numbers of subjects in AR and IR groups and number of subjects who experienced the outcome events. Two researchers independently retrieved the data from the evaluation studies and discrepancies were examined to validate the data. Bayesian Hierarchical Modeling with SAS was applied to estimate the random effect of each study individually and the pooled effect.
Results: All 17 studies included measures of maltreatment re-reports, but only two found that AR subjects had a slightly lower recurrence rate than IR groups with a 95% confidence interval (CI). The pooled Odds Ratio (OR) suggests that there is no difference in maltreatment re-report between the AR and IR groups (OR = 1, CI=0.9 to 1.1). Four studies included measures of substantiated re-reports and 6 studies examined out of home placements. Although the meta-analysis shows that AR groups had lower odds of experiencing these outcomes than IR groups, the difference were not statistically significant (OR=.86, CI=0.3 to 1.4 for substantiated re-report; OR=.77, CI=1.5 to 1.0 for out of home placement).
Implications: The results of a meta-analysis confirm that children who receive an assessment response are no greater risk of maltreatment re-reports, substantiated re-reports, or out-of-home placements compared to children who receive a traditional CPS investigation. Given the widely documented positive impact on family engagement and worker satisfaction, it would be worthwhile to continue DR experiments and refine its practice models.