Over 35 RCTs were included.
Only the following two trials (alendronate and vitamin D plus calcium) met all five criteria for demonstrating benefit.
Alendronate plus vitamin D and calcium vs vitamin D plus calcium (1 RCT, 2027 community dwelling patients with low bone mass and previous fracture followed for 3 years): Relative hazard (RH) of hip fracture = 0.49 (95% CI: 0.23, 0.99); radiographic vertebral fracture RH = 0.53 (95% CI: 0.41, 0.68), clinical vertebral fracture RH = 0.45 (95% CI: 0.27, 0.72); wrist fracture RH = 0.52 (95% CI: 0.31, 0.87).
Vitamin D 20 mg (800IU) plus 1.2 g tricalcium phosphate vs placebo (1 RCT, calcium and vitamin D deficient population of 3270 nursing home residents followed for 3 years): Hip fracture RH= 0.73 (95% CI: 0.62, 0.84). Non-vertebral fracture: RH= 0.72 (95% CI: 0.60, 0.84).
Other therapies.
Vitamin D monotherapy (1 study): no impact on hip or peripheral fractures. Calcium monotherapy (3 studies): Three studies reported decreased fracture risk (as opposed to patients with fracture).
Hormone replacement therapy (2 retrospective studies): No statistically significant effect on number of patients with fractures. No prospective RCT evidence of reduced incident of fracture was found. Etidronate (3 studies): inconsistent results. One study reported statistically significant reduction in patients with fractures at 2 years but no statistically significant difference at 3 years. One study reported no statistically significant difference. One study did not have sufficient detail reported to calculate the RR of patient with fracture. Other biphosponates (1 RCT of pamidronate): no statistically significant effect on fracture rate. Calcitrol (5 RCTs): inconsistent results. One study reported statistically significant reduction in patients with fractures. Three studies reported no statistically significant impact of treatment. One study did not have sufficient detail reported to calculate the RR for patients with fracture.
Other vitamin D preparations (2 RCT): no statistically significant effect.
Fluoride (7 studies): inconsistent results. Often only fracture rates were reported. Caclitonin (5 studies): 1 study reported statistically significant reduction in patients with fractures. Two studies did not have sufficient detail reported to calculate the RR for patients with fracture. One study was 'too small to report benefit'. One study has only been produced in abstract and the full report is awaited.
Parathyroid hormone (2 very small studies): studies extremely limited.
Selective oestrogen receptor modulators (1 study with interim analysis): no published RCT with a fracture end point were identified.