Comparison of nonrandomized trials with slow-release sodium fluoride with a randomized placebo-controlled trial in postmenopausal osteoporosis.

The results of slow-release sodium fluoride (SR-NaF) treatment in two nonrandomized trials involving 65 patients with postmenopausal osteoporosis from the primary site and 121 patients from collaborative sites were compared with those obtained from 54 treated patients and 56 patients taking placebo from a randomized controlled trial. Spinal fracture data were analyzed separately in mild to moderate bone loss of lumbar spine (baseline L2-L4 bone density [BD] > or = 65% young normal) and in severe bone loss (BD < 65%). Since demographic and fracture data were similar among fluoride-treated patients from the three trials at each stratum of bone loss, their data were combined. In mild to moderate bone loss, SR-NaF treatment in the combined group virtually eliminated new spinal fractures with 96.6% of patients remaining fracture-free. The Fluoride group had a markedly lower individual vertebral fracture rate (0.025 vs. 0.188/patient year, p = 0.0001) and group vertebral fracture rate (0.029 vs. 0.175/patient year, relative risk [RR] 0.12, p = 0.0001) than the Placebo group. In severe bone loss, the combined treated group had a significantly lower new spinal fracture rate than the Placebo group, although the differences were not as marked (group vertebral fracture rate of 0.150 vs. 0.276/patient year, RR 0.54, p = 0.03). In the combined fluoride-treated group, the L2-L4 bone mass rose by 4-6%/year for 4 years, and the femoral neck BD increased by 1-2%/year during first 2 years. The radial shaft BD did not change. The Placebo group did not show a change in bone mass at any site. The prevalence (percentage) of patients with related gastrointestinal side effects and nonvertebral fracture rates did not differ significantly between the combined SR-NaF group and the Placebo group (hip fracture rate of 0.0045/patient year in SR-NaF and 0.0053/patient year in Placebo; appendicular fracture other than hip (see text) rate of 0.0193/patient year in SR-NaF and 0.0159/patient year in Placebo). A subgroup analysis showed a low baseline L2-L4 BD, high prevalent spinal fractures, and reduced body weight to be important determinants of the development of spinal fracture during SR-NaF treatment. Concomitant medications (estrogen, vitamin D, thiazide and thyroid hormone) were not independent predictors of the spinal fracture risk. Only 17% of fluoride-treated patients were nonresponders (new spinal fractures or a fall/no change in L2-L4 bone mass). Thus, the effects of SR-NaF treatment on the spinal fracture rate from nonrandomized trials were similar to those of the treated group of the randomized trial but different from those of the Placebo group. The similarity of response of nonrandomized trials with that of the randomized controlled trial and the resultant combined analysis further validate the efficacy and safety of SR-NaF in the treatment of postmenopausal osteoporosis.