Twelve studies (n=5,655 participants) met the inclusion criteria, but one was excluded from meta-analysis due to a lack of data. Six of the remaining eleven were RCTs, two were cohort studies, and three were controlled trials. Sample sizes ranged from 19 (RCT) to 3,117 (cohort study). The length of follow-up, where reported, ranged from one month to one year. Publication bias was a potential problem for readmission rates and adherence to out-patient treatment.
Readmission rates (six studies, n=1,517): Discharge planning had significantly lower readmission rates (RR 0.66, 95% CI 0.51 to 0.84; NNT=15). There was no evidence of statistical heterogeneity. The fail-safe N test indicated that 15 null studies would be required to make this result no longer significant.
Adherence to out-patient treatment or continuity of care (six studies, n=3,842): Discharge planning had significantly higher adherence to out-patient treatment or continuity of care (RR 1.25, 95% CI 1.07 to 1.47; NNT=7). There was evidence of high statistical heterogeneity between studies (I2 76%). The fail-safe N test indicated that for adherence to out-patient treatment 45 null studies would be required to make this result no longer significant.
Mental health (four studies, n=374): Discharge planning produced significant improvements in mental health outcomes (Hedges' g -0.25, 95% CI -0.45 to -0.05). There was no evidence of statistical heterogeneity between studies.
Quality of life (four studies, n=544): There was no benefit from discharge planning in quality of life as measured by Lehman’s Quality of life Interview, and the World Health Organization's Quality of Life - BREF. There was no evidence of statistical heterogeneity between studies.
Subgroup analysis indicated that there were significant differences between study designs (cohort studies versus controlled trials), and follow-up periods (less than one month versus one to three months). Sensitivity analysis indicated that the exclusion of any single study did not significantly change the overall results.