Figure 2.

Estimated effects of calcium and vitamin D supplement on the risk of joint pain according to selected baseline characteristics. In these analysis, odds ratios and 95% confience limits for effect of calcium and vitamin D supplementation on joint pain are estimated from a logistic regression model adjusted for linear age and race/ethnicity. P- values testing for interaction separately for each subgroup are from models including terms for the main effects for calcium and vitamin D supplementation and the subgroup, plus their interaction. For testing, age and BMI (log transformed) were modeled as linear terms; physical activity was coded (0-4) representing the 5 categories of physical activity.

Current hormone therapy reflects use at baseline if randomized to the Dietary Modification trial only or randomized to active versus placebo if randomized to one of the Hormone Therapy trials. Calcium supplementation and vitamin D supplementation reflects non- protocol use at study entry. E alone = estrogen alone, E+P = estrogen plus progestin.