Utility of medical and drug history in fracture risk prediction among men and women

Preventive strategies against osteoporotic fracture depend, in part, on the availability of simple risk prediction tools whereby pharmacological therapy may be targeted to those at greatest risk. The objective of this study was to evaluate the performance of a series of risk factors routinely listed in primary care records for the prediction of future fracture. Information was obtained from the UK General Practice Research Database, which contains the general practitioner medical records of around 6% of the UK population. We performed a case-control study of all 231,778 adult patients with a recorded fracture between 1988 and 1999 and an equal number of controls, individually matched by age, gender, and medical practice to the cases. In addition to a previous history of fracture, 11 items were identified that independently predicted fracture risk (history of anemia, dementia, cerebrovascular disease, and chronic obstructive pulmonary disease; recent use of oral corticosteroids, anticonvulsants, nonsteroidal anti-inflammatory drugs [NSAIDs], antiarrythmics, hypnotics/anxiolytics, antidepressants, and anti-Parkinson drugs). Three or more of these medical risk factors increased the risk of a vertebral fracture 8.1 times (95% confidence interval [CI] 7.0–9.4) and of hip fracture by 4.6 times (95% CI 4.3–5.0), when compared with subjects without these attributes. The optimum screening characteristics of the risk factors for prediction of vertebral fracture revealed a sensitivity 66%, and specificity 89%. These values were 61% and 68%, respectively, for hip fracture. The positive predictive value (PPV) for fracture over a 5 year period improved with the addition of age and previous fracture history to the number of medical risk factors. PPV rose from 8.4% for women aged 65 years without risk factors or fracture history, compared with 26.8% if women had sustained a previous fracture and had three or more risk factors recorded. The data suggest that routinely recorded medical risk factors permit identification of groups of patients with a substantial increase in future risk of fracture. Further investigations, such as bone densitometry, might be conveniently targeted at this group of patients