|| Checking for direct PDF access through Ovid
We evaluated the influence of dietary, anthropomorphic, and hormonal factors on bone density in a cross-sectional sample of 281 pre- and perimenopausal women age 50–60 years living in Massachusetts. The sample included only women who had intact ovaries and were not currently using estrogen. Information on diet was obtained through a semiquantitative food frequency questionnaire. We measured bone density using single-photon absorptiometry in the non-dominant arm in two sites: the midshaft and the ultradistal radius. We observed no important associations between mid-shaft bone density and dietary variables but found linear relations between ultradistal radius bone density and body mass index [b = 1.10 gm/cm2 per kg/m2, standard error (SE) = 0.56], follicle-stimulating hormone (FSH) (b = −0.36 gm/ cm2 per IU/liter, SE = 0.15), and several nutrients: calcium (b = 0.012 gm/cm2 per mg/day, SE = 0.007), retinol (b = 0.002 gm/cm2 per IU/day, SE = 0.0008), vitamin C (b = 0.025 gm/cm2 per mg/day, SE = 0.013), and vitamin D (b = 0.040 gm/cm2 per IU/day, SE = 0.018). We could not clearly distinguish the independent contribution of these micronutrients, however, because many were commonly ingested together in the form of supplements. Caffeine was inversely associated with bone density (b = −0.035, SE = 0.017) independent of dietary, anthropometric, and hormonal factors. Analyses of individual caffeinated beverages revealed consistent inverse associations for coffee (b = −3.42 gm/cm2 per cups/day, SE = 1.49), tea (b = −2.85 gm/cm2 per cups/ day, SE = 1.56), and caffeinated cola (b = −0.14 gm/cm2 per cans/day, SE = 0.06), but not for decaffeinated coffee or decaffeinated cola. These data indicate that caffeine may be an important determinant of bone density.