OBJECTIVES : To compare the medical resource utilization and costs among Medicare patients with intraoperative periprosthetic hip fractures (PPHFx) versus those without PPHFx during total hip arthroplasty (THA)

METHODS : This retrospective observational cohort study used health care claims from the United States Centers for Medicare and Medicaid Standard Analytic File (100%) sample. Patients aged 65+ with a claim for primary THA between 2010 and 2016 and a diagnosis of hip osteoarthritis (OA) were identified. Intraoperative PPHFx was defined using a combination of codes for THA and fixation or arthroplasty of hip-related procedure during THA. The comparison cohort included patients without PPHFx anytime during the one-year period. A multi-level matching technique, using direct and propensity score matching was used to balance the patient, surgeon and hospital characteristics. The proportion of patients admitted at least once to skilled nursing facility (SNF), inpatient rehabilitation facility (IRF), readmission, outpatient hospital during the 0-90 or 0-365 day period after THA was compared as well as the total all-cause payments during those periods.

RESULTS : A total of 4,460 patients (2,230 per cohort) were included after dual matching, mean age= 75.8 years and female= 74.4%. Utilization of any 90-day post-acute services was statistically significantly higher among patients in the PPHFx cohort versus those in the comparison cohort: SNF (41.7% vs 30.8%, P<0.0001), IRF (17.7% vs 10.1%, P<0.0001) and readmissions (17.6% vs 11.5%, P<0.0001), except outpatient hospital visit (73.1% vs 73.5%, P=0.73). The mean 90-day total all-cause payments were statistically significantly higher for patients in PPHFx cohort versus comparison ($30,114 vs $21,229, P<0.0001). All trends were similar in the 365-day follow up.

CONCLUSIONS : Patients with intraoperative fractures had statistically significantly higher resource utilization and associated costs when compared to patients without PPHFx during THA. The statistically significant differences observed during the 90-day follow up were maintained over the one-year period.