Article type
Year
Abstract
Introduction:
Objectives: There are few therapies that have been extensively investigated using both observational studies (OS) and randomized controlled trials (RCT) with similar eligibility criteria. An exception is the comparisons of RCT and OS in the evaluation of hormone replacement therapy (HRT) on bone mineral density in post-menopausal women
Methods: By using Medline, searching references and consulting endocrinologists, we identified 13 OS and 10 RCT on this topic from 1984 to 1998. Eighteen of these studies provided information on the bone density in the lumbar spine, and 11 in the distal radius. Fourteen lumbar spine studies were adjusted for patient differences and contained sufficient information for computing confidence intervals.
Results: The mean effect on the bone density and the lower half of the 95% confidence interval for the lumber spine studies are shown below. The median changes in bone density were 6.7% for the OS and 8.2% for the RCT. Wide confidence intervals due to small sample sizes were less often a problem for the OS than the RCT. The only OS that was an outlier compared bone measurements years after HRT therapy was discontinued. The OS and RCT also gave comparable results for bone density in the distal radius.
Discussion:
Objectives: There are few therapies that have been extensively investigated using both observational studies (OS) and randomized controlled trials (RCT) with similar eligibility criteria. An exception is the comparisons of RCT and OS in the evaluation of hormone replacement therapy (HRT) on bone mineral density in post-menopausal women
Methods: By using Medline, searching references and consulting endocrinologists, we identified 13 OS and 10 RCT on this topic from 1984 to 1998. Eighteen of these studies provided information on the bone density in the lumbar spine, and 11 in the distal radius. Fourteen lumbar spine studies were adjusted for patient differences and contained sufficient information for computing confidence intervals.
Results: The mean effect on the bone density and the lower half of the 95% confidence interval for the lumber spine studies are shown below. The median changes in bone density were 6.7% for the OS and 8.2% for the RCT. Wide confidence intervals due to small sample sizes were less often a problem for the OS than the RCT. The only OS that was an outlier compared bone measurements years after HRT therapy was discontinued. The OS and RCT also gave comparable results for bone density in the distal radius.
Discussion: