292 REDUCED RATE OF HIP FRACTURE IN PATIENTS WITH HIP OR KNEE OSTEOARTHRITIS

Abstract

Purpose: There are conflicting reports of the association between hip OA and hip fracture, and little is known of any possible relationship between knee OA and hip fracture. Hence our objective was to study the rate of hip fracture in hip and knee OA patients compared with the general population seeking health care using a comprehensive cohort study design. Methods: Sweden has a publicly funded health care system with inand outpatient health care utilization prospectively registered by the patient’s personal identifier. We studied residents of the county of Skane, Sweden by 31 Dec 2003 (total population 1.15 million) who sought health care at least once the following 4-years (2004-2007), thus being captured in the Skane Health Care Register (SHCR). We identified all subjects aged 20 years or older with an ICD-10 code given by a physician for hip OA (M16), n=11 901, 57.1% women, or knee OA (M17), n=23 866, 58.8% women. To obtain hip fracture rates we calculated the person-time for each OA patient from the day of his/her first OA diagnosis within the period until the day of first hip fracture (S72.0, S72.1 or S72.2) or another censoring event (death, relocation, or end of study period by cross-referencing with the national population register). The person-time for each subject in the general population aged 20 years or older seeking health care (n=761 210, reference population) started to count by his/her first diagnostic code (any ICD-10 code) in the SHCR within the period until first hip fracture or another censoring event (in an identical fashion as for OA patients). We calculated the expected number of fractured OA patients by multiplying the person-time in OA patients with the rate of hip fracture in the corresponding stratum (age and sex) of the reference population. We calculated the expected (standardized) rate of hip fracture by using weights derived from the person-time from the OA patients. We then calculated standardized fracture-rate ratios by dividing the observed rate of hip fracture in OA patients by the expected rate. Thus, a fracture-rate ratio <1 equals a reduced rate of hip fracture among OA patients compared with the reference population. Results: We observed 233 hip fractured hip OA patients (2.0%) while 271 (2.3%) were expected. We registered 398 hip fractured knee OA patients (1.7%) while 472 (2.0%) were expected. The observed rate of hip fracture in hip OA patients was 884 per 100 000 person-years (py) (expected rate 1028 per 100 000 py) and the observed rate in knee OA patients was 763 per 100 000 py (expected rate 904 per 100 000 py). The resulting ageand sex standardized fracture rate-ratio for having hip fracture in hip OA patients was 0.86 (95% confidence interval [95% CI] 0.75, 0.98) and in knee OA patients the ratio was 0.84 (95% CI 0.76, 0.93).