2021 ISAKOS Biennial Congress Paper
Characterizing the Incidence of Avascular Necrosis after Corticosteroid Injections of the Hip
Nathan Varady, MD, MBA, New York, NY UNITED STATES
Paul F Abraham, BS, Boston, Massachusetts UNITED STATES
Ahab Chopra, BA, Cambridge, MA UNITED STATES
Michael Peter Kucharik, BS, Boston, Massachusetts UNITED STATES
Wendy Madeline Meek, BBA, Boston, Massachusetts UNITED STATES
Christopher T Eberlin, BS, Boston, MA UNITED STATES
David Freccero, MD, Quincy, MA UNITED STATES
Eric L. Smith, MD, Boston, MA UNITED STATES
Scott D Martin, MD, Boston, MA UNITED STATES
Department of Orthopaedic Surgery, Massachusetts General Hospital / Harvard Medical School, Boston, MA, UNITED STATES
FDA Status Not Applicable
Summary
This study presents the first baseline reference data on the nationwide incidence of AVN after intraarticular CSI of the hip and reveals numerous factors independently associated with increased risk, including type of hip pathology and female sex.
Abstract
Introduction
Corticosteroid injections (CSIs) are a cornerstone in the treatment of numerous hip joint disorders that have failed other forms of conservative management. Notably, recent reports have shown extremely high rates of avascular necrosis (AVN) after CSIs of the hip. However, these studies contained select patient populations, and the absolute incidence of AVN after CSI of the hip has not been defined. Furthermore, whether there are differences in risk of AVN after CSI between various underlying disease states or other patient factors is unknown. Therefore, the purposes of this study were to characterize the nationwide incidence of AVN after hip CSI and to identify risk factors associated with increased risk of AVN development.
Methods
This retrospective cohort study identified all patients undergoing intraarticular CSIs of the hip in the MarketScan database from 2007-2017. Patient age, sex, geographic region, medical diagnoses, and surgeries were collected. The date and indication (osteoarthritis, unspecified pain, other derangements [labral injuries, loose-bodies, etc.]) for the patients’ first CSI were recorded. Cumulative incidence of new AVN was analyzed using Kaplan-Meier analyses censoring on database dropout or any hip surgery for a non-AVN indication. Cox-proportional hazard models were used for adjusted analyses including age, sex, region, year of first CSI, indication for first CSI, and number of subsequent CSIs. Patients with AVN at the time of first injection were excluded.
Results
We identified 143,905 patients undergoing CSI of the hip (65.1% female; median [interquartile range] age 53.0 years [45.0-59.0 years]). The most common indications for CSI were non-arthritis pain (64.8%) and osteoarthritis (32.3%). Across all patients, the one and two-year incidence of AVN was 1.8% (95% confidence interval [CI] 1.72%-1.88%) and 2.45% (95%CI 2.34%-2.55%), respectively. In adjusted analysis, age (hazard ratio [HR] 1.008 per additional year, 95% CI 1.004-1.013, p<0.001), female sex (HR 1.88, 95% CI 1.74-2.03, p<0.001), geographic region (p<0.001), number of additional CSIs (HR 1.08 per injection, 95% CI 1.06-1.1, p<0.001), and CSI indication (p<0.001) were associated with AVN risk. With respect to indication specifically, AVN rates were significantly higher for patients undergoing CSI for osteoarthritis than for joint derangements (HR 2.32, 95% CI 1.66-3.24, p<0.001)or non-arthritis pain (HR 1.71, 95% CI 1.58-1.86, p<0.001).
Discussion And Conclusion
AVN is a relatively rare event after CSI of the hip, as the nationwide one- and two-year incidence of AVN after hip CSI is approximately 1.8% and 2.45%, respectively. Importantly, AVN incidence varies markedly between CSI indication, with significantly higher rates for those undergoing injection for osteoarthritis, even when controlling for patient factors. These results may suggest that AVN is often a function of the underlying disease and/or patient population. Future studies examining the risk of AVN after hip CSI should account for injection indication, and randomized controlled trials would be useful in determining whether CSIs themselves play a causal role in these findings.