There is a great need for simple means of identifying persons at low r
isk of developing osteoporosis, in order to exclude them from screenin
g with bone mineral measurements, since this procedure is too expensiv
e and time-consuming for general use in the unselected population. We
have determined the relationships between body measure (weight, height
, body mass index, lean tissue mass, fat mass, waist-to-hip ratio) and
bone mineral density (BMD) in 175 women of ages 28-74 years in a cros
s-sectional study in a county in central Sweden. Dual-energy X-ray abs
orptiometry was performed at three sites: total body, L2-4 region of l
umbar spine, and neck region of the proximal femur. Using multiple lin
ear regression models, the relationship between the dependent variable
, BMD, and each of the body measures was determined, with adjustment f
or confounding factors. Weight alone, in a multivariate model, explain
ed 28%, 21% and 15% of the variance in BMD of total body, at the lumba
r spine and at the femoral neck according to these models. The WHO def
inition of osteopenia was used to dichotomize BMD, which made it possi
ble, in multivariate logistic regression models, to estimate the risk
of osteopenia with different body measures categorized into tertiles.
Weight of over 71 kg was associated with a very low risk of being oste
openic compared with women weighing less than 64 kg, with odds ratios
(OR) of 0.01 (95% confidence interval (CI) 0.00-0.09), 0.06 (CI 0.02-0
.22) and 0.13 (CI 0.04-0.42) for osteopenia of total body, lumbar spin
e and femoral neck, respectively. Furthermore a sensitivity/specificit
y analysis revealed that, in this population, a woman weighing over 70
kg is not likely to have osteoporosis. Test specifics of a weight und
er 70 kg for osteoporosis (BMD less than 2.5 SD compared with normal y
oung women) of femoral neck among the postmenopausal women showed a se
nsitivity of 0.94, a specificity of 0.36, positive predictive value (P
PV) of 0.21, and negative predictive value (NPV) of 0.97. Thus, exclus
ion of the 33% of women with the highest weight meant only that 3% of
osteoporotic cases were missed. The corresponding figures for lumbar s
pine were sensitivity 0.89, specificity 0.38, PPV 0.33, and NPV 0.91.
All women who were defined as being osteoporotic of total body weighed
under 62 kg. When the intention was to identify those with osteopenia
of total body among the postmenopausal women we attained a sensitivit
y of 0.92 and a NPV of 0.91 for a whereas we found that weight could n
ot be used as an exclusion criterion for osteopenia of femoral neck an
d lumbar spine. Our data thus indicate that weight could be used to ex
clude women from a screening program for postmenopausal osteoporosis.