Limited clinical utility of a self-evaluating risk assessment scale for postmenopausal osteoporosis: lack of predictive value of lifestyle-related factors.

The aim of this study was to assess the efficiency of a self-administered questionnaire to identify subjects with postmenopausal osteoporosis in the setting of first line medical care. A sample of 300 postmenopausal women completed the questionnaire based on 18 items. Bone mineral density at the lumbar spine (BMD-L), total hip (BMD-H), and femoral neck (BMD-N) was used as objective criterion for evaluation. The mean risk score was 8.2 +/- 3.21. BMD was correlated with total risk score: r = -0.32 for BMD-L, -0.36 for BMD-N, and -0.43 for BMD-H. Cutoff points for the risk score (equal likelihood points) according to a T-score threshold of -2.5 were 8.6 for BMD-L and BMD-N and 9.3 for BMD-H; specificity and sensitivity was 62% and 62%, respectively, for BMD-L, 65% and 62% for BMD-N, and 75% and 63% for BMD-H. Stepwise multiple regression analysis of the questionnaire items in relation to BMD showed higher correlation coefficients for models including individual items rather than the overall risk score. Items concerning low weight, older age, and wrist fracture after 50 years of age were always selected as significant determinants of BMD (R = 0.43-0.55). Hormonal replacement therapy was also an important determinant. Lifestyle-related items did not contribute significantly. In conclusion, the diagnostic performance of the 18-item self-administered questionnaire was poorer than a shortened questionnaire omitting lifestyle factors. The clinical utility of a questionnaire should ultimately be evaluated in the specific optic of a chosen global strategy for prevention of osteoporotic fractures.