Validation of the Cummings' risk score; how well does it identify women with high risk of hip fracture: The Tromso Study

We examined a two-step case-finding strategy where the Cummings' risk score (NEJM 1995) was applied in a population-based setting together with bone mineral density (BMD) measurements in order to validate its ability to identify women with high risk of hip fracture. All Tromso women aged between 65 and 74 were invited to the Tromso Osteoporosis Study (TROST) together with a 5% random sample of women aged 75-84 years (n = 1410). All had forearm BMD measurements in 1994/95 and were followed for 5 years with respect to first hip fracture. A risk score was constructed matching the Cummings score as closely as possible. In all 759, 578 and 73 women had 0-2, 3-4 and 5+ risk factors, respectively. Women with 5+ risk factors had a 5-year hip fracture risk of 11% (95% confidence interval (CI) 3.7-18.2%). BMD screening applied to these women identified 74% of them as osteoporotic and 19% as osteopenic with, respectively, 5-year hip fracture risk of 13% and 7.1%. In a population different from the one the score was generated in, this simple risk score identifies a group of women with high risk of hip fractures. With no additional BMD measurements, those high-risk women could benefit from early intervention measures.