Seventeen RCTs were included in the review (n=4,780 patients). Fifteen RCTs reported that they included multidimensional geriatric assessment as part of the intervention. Ten RCTs reported on assignment as part of the intervention. Ten RCTs reported adequate concealment of allocation. Seven RCTs blinded outcome assessors. All RCTs reported intention-to-treat results.
Short term effects (at discharge):
Overall, there was a statistically significant short-term beneficial effect on all outcomes: combined odds ratio 1.75 (95% CI 1.31 to 2.35) for functional improvement; relative risk 0.64 (95% CI 0.51 to 0.81) for nursing home admission; and relative risk 0.72 (95% CI 0.55 to 0.95) for mortality. Heterogeneity between studies was not significant.
Meta-regression analyses showed two significant differences in effect: a greater functional improvement for orthopaedic rehabilitation programmes compared with general geriatric rehabilitation programmes (OR 2.33, 95% CI 1.62 to 3.34 versus OR 1.34, 95% CI 1.12 to 1.60, p=0.04); and a larger reduction in admissions to nursing homes in trials with patients aged 80 years or younger compared with those aged over 80 years (RR 0.42, 95% CI 0.27 to 0.64 versus RR 0.75, 95% CI 0.58 to 0.96, p=0.045).
Longer term effects (at three to 12 month follow-up):
Overall, there was a statistically significant longer term beneficial effect on all outcomes: combined odds ratio 1.36 (95% CI 1.07 to 1.71) for functional improvement; relative risk 0.84 (95% CI 0.72 to 0.99) for nursing home admission; and relative risk 0.87 (95% CI 0.77 to 0.97) for mortality. There was significant heterogeneity for the outcome functional improvement.
Meta-regression analyses showed one significant difference in effect: a greater functional improvement for orthopaedic rehabilitation programmes compared with general geriatric rehabilitation programmes (OR 1.02, 95% CI 0.86 to 1.21 versus OR 1.79, 95% CI 1.24 to 2.60, p=0.01).
NNT to avoid one admission to a nursing home at hospital discharge was between nine and 28. NNT to avoid one admission to a nursing home at follow-up was between 21 and 63. NNT to prevent one death at one year follow-up was 38 (20% one-year mortality rate among controls was assumed). Results that related to rate of re-admission to hospital were reported.
Funnel plots and Begg and Egger bias tests indicated little evidence of publication bias.