We appreciate Dr. Gordan's comments, particularly because of his outstanding contributions to this area during the last 30 years. Although our study was not designed to evaluate the effects of progestins on lipids and bone density or to evaluate the contribution of calcium intake to preservation of bone mass, we agree in principle with Dr. Gordan's views.

Dr. McGivney is correct that several factors may impede photon penetration and spuriously increase the value of spine bone mineral density as assessed by dual photon absorptiometry. However, we feel that compression fractures of lumbar vertebrae did not skew our results for the following reasons:

1) Only one woman suffered a fracture of L2, L3, or L4 during the trial. She was in the estrogen group, but her rate of change in bone mineral density was in the lowest quartile of rates for women randomized to estrogen.

2) If compression fracture had spuriously altered our bone mineral density results, we would have expected a greater increase to be seen in the placebo group because this group developed more compression fractures.

3) We observed a statistically significant negative association between vertebral fractures and the rate of change in bone mineral density for the estrogen group, but not for the placebo group, during logistic regression analyses.

4) It seems most unlikely that estrogen would cause a greater progression of degenerative joint disease, vascular calcifications, or vertebral compressions in the treatment group during the short duration of this study. We agree with Dr. McGivney that randomization should have eliminated the influence of these factors.

We did not collect information on exercise or resistance weight training, but virtually all women in this trial had sedentary life styles. We included women with hysterectomies because a large fraction of women with osteoporosis have had hysterectomies (41% in our trial). For clinical treatment, estrogen alone should probably be used in such patients. In our clinical trial, the use of both drugs was necessary to balance contingent variables.

In reply to Dr. Petrovic we were careful to exclude patients from the study who had recently used estrogen, because this factor would most likely reduce the intended response, not increase it. Previous, but more distant, estrogen use did not influence the magnitude or direction of responses.