DISCUSSION
This study demonstrated that multiple courses of HA treatment can delay the need for surgery for up to 3 years, with risk for both TKR and partial knee replacement decreasing in a dose-dependent manner. The potentially confounding effect of differences in baseline characteristics that could influence patients’ propensity to receive TKR in a database study was controlled by performing a multivariate analysis with covariate adjustment. The TKR-delaying effect of HA injection was more prominent in cohorts with a high number of HA treatment courses: 19 out of 20 patients in the cohort of ≥5 HA courses were free of TKR at the end of the 3-year post-index period. Such a high proportion of patients avoiding TKR with repeated courses of HA suggests that some patients may be able to successfully delay TKR well beyond the 3-year time span. This finding is counter-evidence to the frequently made assumption15 that all patients with knee OA will eventually progress to a state of disability, making TKR inevitable. The patients with end-stage radiographic knee OA can also benefit from IA HA injections for an extended period of time;16 the latest evidence indicates that nonoperative management can improve symptoms irrespective of radiographic disease severity, implying that TKR needs not to be the only therapeutic option for patients with end-stage radiographic knee OA.17 This finding suggests that HA treatment should be considered an important clinical treatment option for patients with knee OA.
Although the incidence rates of certain adverse events, such as arthralgia/joint pain, are sizable, these temporary adverse events commonly occur among patients who receive IA injections for knee OA; most of these events may simply include symptoms of the remaining underlying knee OA. These results are consistent with those of previous literature reporting the safety of repeated treatment with IA HA injections in a prospective clinical trial18 and demonstrating that repeated courses of HA treatment pose no greater safety risk than a single course of HA treatment.
Multivariate modeling outcomes of factors influencing risk of receiving TKR are broadly consistent with the generally accepted notions that different levels of disease severity and patients’ willingness to consider TKR at baseline influence the likelihood and timing of receiving TKR.19,20 Age and obesity are common risk factors for progression of OA. Orthopedic surgeons are more likely to recommend surgery than non-surgeons. The pre-index use of corticosteroids and high pre-index healthcare costs could be associated with more severe symptoms at baseline. Patients with vascular disease or severe comorbidities, as evidenced by high CCI scores, make poor candidates for major elective surgeries such as TKR. These results are intuitive and validate the clinical insights of this study. Moreover, inclusion of these covariates in the analysis model allows for indirect adjustment of the most important prognostic factors for TKR at baseline, permitting proper statistical comparison of the results for different cohort groups.
Recently, the efficacy of HA injections for OA patients has become the subject of debate when the American Academy of Orthopaedic Surgeons (AAOS) revised its clinical practice guideline, recommending against the use of HA.21 The AAOS’ findings differ from those of other clinical societies, such as the American College of Rheumatology22 and the European League Against Rheumatism,23 which provide no strong recommendation against the use of HA injections. The announcement of the new guideline by AAOS caused concern among clinicians and payers who had valued IA HA injections as a means to control knee OA pain before patients progress to TKR;24 on the other hand, the demand for nonoperative treatment of knee OA remains high. Utilization rates of TKR have increased dramatically, and surgeries are now performed on younger patients with increasing burden on the healthcare system,25,26 in spite of the fact that as high as a third of TKR surgeries may have been performed in inappropriate patients.27 Part of the confusion surrounding clinical utility of HA stems from the fact that up until recently, relatively little research looked into the practical benefits of HA in actual clinical practice. Analyses of databases such as registries are now gaining attention to overcome that problem. Examination of large administrative databases maintained by commercial payers offers the benefit of probing realistically the safety and efficacy of treatments in actual clinical environments in a very large number of patients with heterogeneous backgrounds. Recently, the Agency for Healthcare Research and Quality’s Technology Assessment Program in the US called for such studies to determine whether HA injections can delay progression to TKR.28 The results of this study and several others11,13,14,16 suggest that use of HA to treat OA of the knee is associated with the delay of TKR, supporting the utility of HA in clinical practice and the healthcare system. Potential clinical benefits of delaying TKR may include the reduced risk of aseptic loosening if younger patients can wait for TKR or more time to allow the modification of risk factors in patients who will ultimately undergo TKR.
LIMITATIONS
Follow-up period was limited to 3 years post-index date because longer follow-up data were not available at the time of the study design. If an incorrect adverse event or OA diagnosis was listed in the medical record, or if the medical record was incomplete, then patients might have been misclassified, resulting in selection bias. The claims dataset includes no uninsured and Medicare patients, as the population in the database consisted primarily of commercially-insured patients in the US. Therefore, the results are most generalizable to other commercially-insured patients in the US. Generalizability to other populations may not be assured if they differ in their accessibility to physician services or prescriptions from the patients in this study. Other treatments such as the nonsteroidal anti-inflammatory drugs used by patients were not included within the pre-specified statistical model because their potential effects were assumed to be short-lived and much less than those of corticosteroid. Including these treatments would overload the statistical model with too many covariates, leading to potential computational instability. The database used provides no information on systemic factors, including plan limits on medication use, that could affect care. Given the large and diverse nature of the healthcare plans in the database. However, these factors should not have materially affected our study results. The claims database also lacks direct indicators of OA disease severity, such as Kellgren-Lawrence scores or patient-reported outcomes, including pain and function questionnaire scores. Our multivariate analysis indirectly makes up for this deficiency by considering other baseline characteristics or clinical indicators that may be correlated with information unavailable in a claims database. Patients who opt to undergo repeated courses of HA treatment may be more inclined to avoid surgery or may naturally experience OA disease progression more slowly, making them potentially different from patients who select to undergo surgery earlier without repeated courses of HA treatment. This condition may introduce a bias that causes difficulty in proving the causality between repeated HA use and delay of TKR.
CONCLUSION
Analysis of the knee OA patient data from a real-world database showed that repeated courses of treatment with HA are safe and are associated with the delay of TKR for up to 3 years. Additional research is needed to evaluate the effects of repeated HA courses on delaying TKR beyond a 3-year period.