Background/Purpose:  More than 27 million adults in the US have knee osteoarthritis (OA), a painful and life-altering disease. Various non-pharmacologic and pharmacologic agents are recommended to alleviate pain, inflammation, and to improve function. When patients have inadequate response to these therapies, viscosupplementation with hyaluronic acid (HA) injections helps restore synovial fluid properties in the knee, leading to less pain and improved clinical outcomes. Total knee replacement (TKR) usually is reserved as the final treatment option. The present study examined the impact of the use of HA injections in delaying TKR in patients with knee OA.

Methods:   A retrospective analysis of the Truven Marketscan databases was performed.  All OA patients continuously enrolled during calendar year January, 2007 through December, 2011, that went on to TKR. Two cohorts were identified with this population: A cohort of patients receiving HA injections and the second cohort with no HA injections. Time-to-event analyses were performed in both cohorts, where the starting time was defined as the patients’ 1^st visit to an OA specialist (primarily orthopedic surgeons), and the ending point was TKR.  The initial visit had to have occurred after January, 2008, providing a 12-month clean period.  HA usage was grouped into episodes of treatment, where successive HA injections within 15 days were grouped into the same episode. An episode encompassed one course of a series of weekly injections or a single injection of HA, based on the standard use of HA. A propensity scoring methodology was implemented to adjust for baseline characteristics of patients in HA and non-HA cohorts.

Results:   The number of patients in the HA and non-HA cohorts were 7,000 and 19,627, respectively.  Of the 7,000 HA patients, we successfully propensity score matched 6,891 with the non-HA cohort (98%) with caliper of 0.001.  The majority of the patients were female (66%). The median times from the initial specialist visit to TKR were 199 and 443 days for non-HA cohort and HA cohort with one episode of HA treatment, respectively.  There was a well defined “dose response curve”, with each treatment episode increasing the median gap by on average 202 days. As an example, for HA cohort with 3 and 4 or more episodes of treatments, the median times to TKR were 784 (585 days delay) and 1,009 (810 days delay) days, respectively.

Conclusion:    This observational, descriptive analysis of an administrative database provides evidence that HA injections delay patients’ progression to TKR (up to approximately 2.2 years documented in the present study population).  Although the analysis attempted to control for disease severity by propensity score matching, there could be remaining differences between the HA and non-HA populations not recorded in the database which could affect the interpretation of the results.