Five randomised controlled trials (RCTs; 1,602 women) were included.
Nonvertebral fractures: the RR for nonvertebral fracture for those on alendronate versus placebo was 0.71 (95% CI: 0.502, 0.997, P=0.048). The estimated cumulative incidence of nonvertebral fracture after 3 years was 9.0% in the alendronate group and 12.6% in the placebo group.
For women younger than 65, the nonvertebral fracture rate per 100 patient years was 3.66 in the placebo group and 2.99 in the alendronate group. For women over 65, the nonvertebral fracture rate per 100 patient years was 5.34 in the placebo group and 3.57 in the alendronate group.
Point estimates for RR across studies ranged from 0.61 to 0.76 when any one trial was omitted.
The Cox model including treatment, study and their interaction showed no evidence of heterogeneity (P=0.77).
After 3 years, the risk reduction for fractures of the hip was estimated to be 54% (95% CI: 85% reduction, 36% increase, P=0.15) and the risk reduction for fractures of the wrist was estimated to be 61% (95%CI: 22% reduction, 81% reduction, P=0.006).
Bone mineral density: over 3 years, 10 mg alendronate increased bone mineral density compared to placebo. The bone mineral density increased by the following amounts at the following sites: spine 8.8%, trochanter 7.8%, femoral neck 5.9%. Total body calcium increase by 2.5%.
Treatment with 20 mg daily produced comparable effects, whereas 5 mg daily was only about 70% as effective as 10 mg. 2.5 mg was approximately half as effective as the 5 mg dose.