Seven RCTs (n=6,884) were included. Six were individually randomised and one was a cluster-randomised trial.
Methodological quality.
Five studies used centralised methods of randomisation, no studies used a sham control, and all studies allowed for intention-to-treat analysis. Losses to follow-up ranged from 2 to 69%.
Community-dwelling elderly (4 RCTs, n=5,696).
No statistically significant difference in the occurrence of one of more hip fractures was shown between hip protectors and control; the RD was 0% (95% CI: -1, 1, p=0.37) and the RR was 1.07 (95% CI: 0.81, 1.42, p=0.55). There was no evidence of statistical heterogeneity (p=0.61 and p=0.55, respectively).
Institutionalised elderly (2 RCTs, n=248; 1 cluster RCT, n=942).
In the 2 individual RCTs, no statistically significant difference in the occurrence of one or more hip fractures was shown between hip protectors and control; the RD was -7% (95% CI: -24, 11, p=0.45) and the RR was 0.49 (95% CI: 0.06, 3.97, p=0.51). There was evidence of statistical heterogeneity (p=0.03 and p=0.06, respectively).
In the cluster-randomised trial, no statistically significant difference in the occurrence of one or more hip fractures was shown between hip protectors and control; the RD was -3.5% (95% CI: -7.3, 3.0, p=0.072).
When combining the individually- and cluster-randomised trials (n=1,425), no statistically significant difference in the occurrence of one or more hip fractures was shown between hip protectors and control; the RD was -3.7% (95% CI: -7.3, 0.1) and the RR was 0.56 (95% CI: 0.31, 1.01).
The subgroup analysis of nursing home residents showed that hip protectors were associated with a statistically significant decrease in the risk of hip fractures compared with control; the RD was -4.4% (95% CI: -8.09, -0.76) and the RR was 0.50 (95% CI: 0.28, 0.91), based on 1,014 participants in 2 studies.
No studies evaluated health-related quality of life.