Twelve RCTs (n=2,876) were included in the review.
Six studies were reported to be high quality and six to be moderate quality.
Overall results: CM interventions (10 studies) did not appear to be effective in reducing the LOS (AWES 0.094, 95% CI: -0.032, 0.220, p>0.7); significant heterogeneity was detected (p<0.025).
Sensitivity analyses: LOS was not affected by study quality, the number of CM components or the geographical setting. However, the interventions appeared to be effective for patients with heart failure, but not for those with a stroke or those classified as frail elderly; the studies were homogeneous (p>0.05).
Overall results: no statistically significant effect of CM on readmission rates (10 studies) was found (OR 0.87, 95% CI: 0.69, 1.04); the studies were homogeneous (p>0.1). Using a binominal effect size display this was interpreted as a 6% decrease in readmission rates.
Sensitivity analyses: readmission rates were not affected by study quality, the number of CM components or patient diagnosis. Studies conducted in the USA were found to be effective, but not studies conducted elsewhere; the studies were homogeneous (p>0.1).
The authors concluded that there was no significant risk of publication bias in terms of LOS or readmission rates, based on the results of the funnel plot and fail-safe N.