Hip fractures constitute a major health problem. For effective prevention,
high-risk groups need to be identified. The objective here was to develop h
ip fracture risk scores while assessing the added value of bone mineral den
sity relative to more conventional risk indicators. We prospectively follow
ed during 4 years a cohort of 5208 persons (2193 men) aged 55 years and ove
r from the Rotterdam Study, a population-based cohort study conducted in th
e Netherlands. Risk scores for hip fracture were constructed using several
conventional risk indicators requiring interview and anthropometry only, an
d bone mineral density. During follow-up, 50 persons (14 men) suffered hip
fracture. Hip fracture risk was independently determined by age, gender, he
ight, the use of a walking aid, cigarette smoking, and either bone mineral
density or weight, We developed two risk scores, with and without bone mine
ral density. The observed ii-year risk ranged from 3/3389 (0.1%) to 17/169
(10.1%) for the lowest and highest category of the score including bone min
eral density, respectively. For the score without bone mineral density, the
se risks were 8/3117 (0.3%) and 16/144 (11.1%), respectively. The area unde
r the receiver operating characteristic curve indicating discriminatory pow
er was 0.88 for the risk score including, and 0.83 far the score excluding,
bone mineral density (p for difference = 0.04). In conclusion, risk scores
with and without bone mineral density measurement can be used for hip frac
ture risk assessment in elderly persons. While the score with bone mineral
density has a modestly better performance, the score requiring interview an
d anthropometry only may be especially useful in primary care settings. (C)
1999 by Elsevier Science Inc. All rights reserved.