High-concentration nonavian high-molecular weight hyaluronan injections and time-to-total knee replacement surgery

This retrospective study utilized the IBM MarketScan Commercial claims database. This database contains administrative claims and eligibility records for approximately 40 million enrollees in distinct sets of files for commercially insured individuals (i.e., working-age adults and their dependents). A standard extract from the database consists of three files: an enrollment file, a medical file and a drug file. The files are linkable based on an encrypted patient identification number. A prior retrospective commercial claims analysis study by Altman and colleagues explored the association between HA injections, using all types of HAs (2006–2011) and time-to-TKR surgery [5]. The present study updates the results of the prior study using recent data from 1 January 2008 to 31 December 2017 and subsets the HA cohort to patients only receiving Orthovisc (referred to as HMW-HA).

All commercially insured patients who underwent TKR surgery in inpatient settings throughout 1 January 2008 and 31 December 2017 and had a diagnosis for knee OA by an orthopedic surgeon prior to TKR were determined through claims with an International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9-CM) procedure code of 81.54 and associated ICD-10-CM procedure codes. The first entry of knee OA diagnosis in the data was identified as the index date. This criterion was used based on the clinical consideration that the diagnosis by someone other than an orthopedic surgeon would not clearly indicate knee problems related to TKR surgery. Patients’ time-to-TKR was defined as the time interval between the baseline date and that of TKR surgery. Patients were subjected to two additional screenings (Figure 1) and were required to have no other knee OA diagnosis visits to an orthopedic surgeon prior to initial knee OA diagnosis, defined as the ‘minimum clean period’ (MCP). The length of the MCP of the study population (90th percentile) was about 14 months (420 days) and spanned from the time between patients' first and second visits to the orthopedic surgeon. Patients were additionally required to demonstrate a minimum continuous enrollment (MCE) period of 7 years. This was derived from the sum of the MCP of 420 days and the time from initial diagnosis of knee OA to the date of TKR surgery or 2188 days (95th percentile), which totaled to 2608 days (or about 7.1 years). A MCE period was required to make sure all claims were available to record time to TKR accurately from initial OA diagnosis.

The study population was grouped into two cohorts, one comprised of patients who had nonavian HMW-HA injections and the other of patients who did not have nonavian HMW-HA injections (or any other HA injection). Nonavian HMW-HA is administered as a series of 3–4 weekly injections within a 6-month period. The age group, gender and US geographic region of resident patients were determined for each study cohort.

Patients were included in the study if they were between 18 and 64 years of age at time of diagnosis for knee OA (as the database did not cover Medicare-eligible patients). Patients were excluded from the study if they met any of the following exclusion criteria: had other knee OA diagnosis visits to an orthopedic surgeon prior to the initial diagnosis date of knee OA, designated as MCP. This was to remove prevalent cases of knee OA; had prior HA injections (including nonavian HMW-HA) in the MCP and other HA injections from index date to TKR. This was to remove any prevalent users of HA and to restrict the cohort to only new users; had continuous enrollment below the MCE period. This was to remove patients that may have missing TKR claims due to loss of enrollment; had a TKR surgery in any setting any time prior to the first TKR surgery in inpatient settings identified between 1 January 2008 and 31 December 2017. This was to remove bilateral TKR cases and; had a diagnosis for revision TKR any time prior to the first TKR surgery in inpatient settings between 1 January 2008 and 31 December 2017. Revision TKR would indicate a prior TKR and hence this was to remove bilateral TKR cases.

Time-to-TKR surgery was defined by the total number of days between the date of initial knee OA diagnosis and the date of the TKR surgery. For both the nonavian HME-HA and non-HA cohorts, the median and mean times-to-TKR surgery were evaluated, as was time-to-TKR surgery associated with repeat nonavian HMW-HA courses. One course of nonavian HMW-HA was defined as three injections within 6 months or, alternatively, the end of a 6-month period from the first injection.

p-values where calculated to evaluate baseline differences in patient characteristics. Kaplan–Meier curves were developed. The t-test and log-rank test were used independently to estimate the differences in mean and median times to TKR surgery for the two cohorts, respectively. Furthermore, the impact that the number of courses of nonavian HMW-HA injections had on mean and median times to TKR surgery were evaluated using a one-way analysis of variance (ANOVA) and log-rank test, respectively. Statistical significance was set at 0.05 (two sided). Finally, Cox regression models were used to evaluate the effect of nonavian HMW-HA injections (yes/no) and number of courses of nonavian HMW-HA injections on time to TKR surgery adjusting for baseline demographic (age, gender [reference – male], region [reference – west] and insurance type [reference – preferred provider organization]) and clinical characteristics (functional comorbidity score derived from 18 different comorbid conditions). Cox models were run separately for cohorts with and without continuous enrollment restrictions.

Table 1 describes the process through which patients were selected for the study population. A total of 412,430 adult patients who had a TKR surgery between 1 January 2008 and 31 December 2017 were identified overall, 30,028 of which met the study inclusion criteria and were therefore included in the overall study population. There were 1978 patients in the nonavian HMW-HA cohort and 28,050 patients in the non-HA cohort.

Patient characteristics of the overall study population and patient cohort are presented in Table 2. Of the overall study population, 58.2% were women and 66.5% were aged 55–64 years old. The majority of patients were from the south (47.2%) and north central (29.8%) of the USA. Both cohorts were similar in baseline characteristics.

A total of 18 different comorbid conditions were used to develop a functional comorbidity score, along with mean functional comorbidity score [14]. Patient comorbid conditions of the overall study population and patient cohorts by use of nonavian HMW-HA are presented in Table 3. Of the overall study population, the mean value for the Functional Comorbidity Index was 2.8 overall. The nonavian HMW-HA cohort index was 2.7 and the non-HA cohort was 2.8. The highest individual comorbidity was congestive heart failure or heart disease with 53.8% overall, 50.9% for nonavian HMW-HA cohort Index and 54.0% for the non-HA cohort.

The median time-to-TKR surgery was 893 days for the nonavian HMW-HA cohort and 399 days for the non-HA, with a difference of 494 days (or 16.2 months; p < 0.001) between them (Figure 2). The mean time-to-TKR surgery was 1201 days (or 39.4 months) for the nonavian HMW-HA cohort, which differed from the timespan of 837 days (or 27.4 months) for the non-HA cohort by 364 days (12 months; p < 0.001).

A total of 32.3% of the patients in the nonavian HMW-HA cohort had two or more courses during the evaluation period, with 20.8% having two courses, 5.7% having three courses and 5.8% having four or more courses of nonavian HMW-HA (Table 4).

The median time-to-TKR increased with each additional course of nonavian HMW-HA, ranging from 773 days among patients with one course of nonavian HMW-HA to 1612 days among patients with 4+ nonavian HMW-HA courses (Table 2). The time-to-TKR stratified by number of courses of nonavian HMW-HA injections is presented in Figure 3.

These results show that treatment with an increasing number of courses of nonavian HMW-HA injections was associated with a stepwise increase in time-to-TKR surgery (p < 0.001).

For patients with at least 7 years continuous enrollment, two multivariable Cox regression models adjusting for baseline demographic and clinical factors were conducted; one with any use of nonavian HMW-HA and the other evaluating the effect of number of courses for nonavian HMW-HA. Any use of nonavian HMW-HA model showed that the risk for TKR was 24.3% lower with receipt of nonavian HMW-HA as compared with no HA at any given time (Table 5). The model for number of courses for nonavian HMW-HA showed that the risk of TKR reduced as the courses of nonavian HMW-HA increased as compared with no HA: 18.2% lower risk with one course, 30.3% lower risk with two courses, 38.5% lower risk with three courses and 43.8% lower risk with four or more courses of nonavian HMW-HA (Table 6).

For patients with no continuous enrollment restriction, similar results were obtained from the multivariable cox regression models. Any use of nonavian HMW-HA model showed that the risk for TKR was 39.1% lower with receipt of nonavian HMW-HA as compared with no HA at any given time. Additionally, the risk of TKR reduced as the courses of nonavian HMW-HA increased as compared with no HA: 34.3% lower risk with one course, 44.8% lower risk with two courses, 56.2% lower risk with three courses and 59.6% lower risk with four or more courses of nonavian HMW-HA.

Strengths of the present study were the large cohort size (n = 30,028), long time horizon (7 years minimum continuous enrollment), the use of real-world patient care data, the multiple institutions and physician practice patterns represented andthe varying geographic locations of patients in the study sample. Another strength included, this study was limited to a single nonavian HMW-HA, which would reduce HA differentiation bias. Additionally, this study used multivariable regression approach to adjust for the baseline differences between the patients receiving nonavian HMW-HA and no HA. The regressions showed similar results with and without 7 years continuous enrollment criteria showing the robustness of the results. the This study was subject to a number of important limitations. First, information potentially impacting study results, including measures of disease severity baseline pain scores or patient willingness in both nonavian HMW-HA injection status cohorts were not available in the claims database. Second, the date of diagnosis by an orthopedic surgeon may not be an accurate reflection of the onset of disease. Furthermore, baseline diagnoses may not necessarily represent the first diagnosis of knee OA. Third, the study population was limited to patients under the age of 65 years, reflecting the patient population included in the database. Therefore, these results may not necessarily be applicable to an elderly population. However, this study is from private payor perspective and thereby Medicare was not analyzed as part of the measured cohort. Fourth, therapy with nonavian HMW-HA injections may have occurred outside of the population included in the claims database or prior to the MCP, excluding such patients from this study. Fifth, medium and small firms were not represented in this study as MarketScan data is primarily supplied by large employers, despite the size of this data source. Sixth, certain plan-specific policies may restrict the usage of nonavian HMW-HA or require patients to try specific therapies prior to undergoing TKR surgery. Due to the inability to identify specific health insurance plans in the claims database, it was not possible to account for such policies. Seventh, the database does not provide information on delays in getting an appointment with an orthopedic surgeon and getting into the operating room. However, this time bias would have affected both the nonavian HMW-HA cohort and non-HA cohort equally. Eighth, no causal inferences can be drawn as this is an observational study and lacks randomization that may cause systematic differences between the groups compared. Finally, as with most retrospective claims analyses, it may be possible that some knee OA diagnoses and nonavian HMW-HA treatments were miscoded. Most of these limitations are inherent to claims database analysis and do not preclude the development of clinically relevant conclusions about the real-world impact of HAs such as nonavian HMW-HA on time-to-TKR in patients with knee OA.