Decreasing Trends in Arthroscopic Treatment of Knee Osteoarthritis After Publication of the 2013 Academy of Orthopaedic Surgeons Clinical Practice Guidelines

Abstract

Introduction:

The purpose of this study was to analyze trends in arthroscopic utilization among patients with an isolated diagnosis of knee osteoarthritis (OA) after publication of the 2013 American Academy of Orthopaedic Surgeons (AAOS) Clinical Practice Guideline (CPG) regarding nonarthroplasty management of knee OA. The 2013 AAOS CPG recommended strongly against the use of arthroscopy for the treatment of knee OA. Our hypothesis was that rates of arthroscopic utilization would decrease after publication of the CPG.

Methods:

Analysis was conducted using the International Business Machines (IBM) MarketScan Commercial and Medicare Supplemental Databases from 2012 to 2019, which contain healthcare data for more than 43.6 million individuals. This study included patients aged 21 years or older who underwent knee arthroscopy for the isolated diagnosis of knee OA without concomitant pathology.

Results:

The overall rate of arthroscopic intervention was 59.15 per 1000 patients in 2012 and decreased by 14.91 points to 44.24 per 1000 patients in 2019. This equates to an average decrease of 4.0% year over year from 2012 to 2019, with an overall decrease of 25.2%. The 40-49, 50-59, and 60-69 age groups demonstrated a decrease of 41.5%, 37.2%, and 35.9%, respectively, from 2012 to 2019. Over the same period, there was increased utilization of physical therapy and corticosteroid injections and decrease in viscosupplementation injections.

Discussion:

This analysis demonstrates a change in practice trends after publication of the 2013 AAOS CPG. Both overall and age cohort–specific arthroscopic utilization rates consistently decreased year over year for all cohorts in accordance with the publication of the 2013 AAOS CPG.

The foundation for effective healthcare quality improvement involves providing physicians, specifically orthopaedic surgeons, with ever-evolving education regarding the most current research findings in the relevant field of interest. There are a myriad of treatment options available for patients facing any number of orthopaedic diseases. The goal of developing clinical practice guidelines (CPGs) is to improve quality of care by creating treatment recommendations based on the best available scientific evidence. The American Academy of Orthopaedic Surgeons (AAOS) defines “Quality” as “the successful delivery of appropriate, evidence-based musculoskeletal care in an effort to achieve sustained patient-centered improvements in health outcomes and quality of life.”¹ For the better part of two decades, the AAOS has collated the best available evidence to assess the comparative effectiveness of various procedures and treatments for musculoskeletal diseases, providing orthopaedic surgeons and their patients with CPGs and guideline-derived products aimed at improving the quality of care.

However, evidence-based quality improvement initiatives, such as CPGs, require time for clinician uptake and implementation. The timely integration of translational research is notably difficult, with some estimates of time lag from publication to incorporation into clinical practice at 17 years.² The reasons behind elongated uptake of CPGs in clinical practice are multifactorial and may include physician awareness, ease of accessibility, delayed compliance by insurance companies, and guideline complexity. To affect practice trends at the point of care, CPGs and translational research require a mix of aggressive dissemination, accessibility, and trustworthiness (ie, transparency and methodological rigor).

Measuring clinician adherence to CPGs can be extremely difficult to accomplish, with endless caveats to exploring the correlation between the evidence and adoption. However, the resources, both financial and human, used by medical specialty societies to develop CPGs justify the question, “Do utilization rates decline for treatments of which CPGs exhibit little to no efficacy?” Another way to view this question would be “Does a physician's choice of treatment follow the best available evidence?”

These questions guided the aim of this analysis. The authors chose to analyze the utilization trends of arthroscopic lavage and/or débridement for treatment of osteoarthritis (OA) of the knee in patients with no concomitant injuries (ie, meniscal tear or loose body). The rationale for exploring arthroscopic utilization rates was mainly due to consistency within the 2013 and 2021 AAOS CPGs on nonarthroplasty management of OA of the knee, which included over three decades of published research on this topic. Both guidelines found convincing evidence against the use of arthroscopy for this patient population. The 2013 AAOS CPG had a strong recommendation against arthroscopy, and the most recent 2021 AAOS CPG has a moderate strength recommendation stating, “Arthroscopy with lavage and/or débridement in patients with a primary diagnosis of knee OA is not recommended.”^3,4 Guidelines with “strong” or “moderate” strength of recommendations indicate that the evidence supporting the recommendation was of high quality and the likelihood that the recommendation would be overturned by future publications on the chosen topic is low.

It is also important to consider utilization of conservative treatment options for knee OA that may have influenced the trends of arthroscopic intervention. The most common conservative treatment options include physical therapy (PT), intra-articular corticosteroid injections (CSIs), and viscosupplementation injection (VSI), such as hyaluronic acid. The 2013 AAOS CPG gave an inconclusive recommendation for the use of CSIs for knee OA.³ The 2021 AAOS CPG issued a moderate recommendation in favor of intra-articular CSIs in the setting of OA, stating that they “could provide short-term relief for patients with symptomatic OA of the knee.”⁴ Hyaluronic acid on the contrary received a strong recommendation against its use in the treatment of symptomatic knee OA in the 2013 AAOS CPG.³ While these conservative therapies have been studied as treatments for knee OA, no study has been conducted examining the utilization trends of the abovementioned conservative treatment options in the year preceding arthroscopic lavage/débridement of the osteoarthritic knee.

The purposes of this study were to analyze trends in utilization of arthroscopic and conservative treatment options for knee OA before and after the publication of the 2013 AAOS CPG on this topic to assess the effect of the guideline on patient care.

Methods

Databases

The analysis of practice patterns related to the AAOS CPG regarding arthroscopic lavage and/or débridement was conducted using the IBM MarketScan Commercial and Medicare Supplemental Databases from 2012 to 2019, which contains healthcare data for more than 43.6 million individuals in the United States. The IBM MarketScan Commercial and Medicare Supplemental Databases represent the medical experience of insured employees and their dependents, including active employees, early retirees, Consolidated Omnibus Budget Reconciliation Act recipients, and Medicare-eligible retirees with employer-provided Medicare Supplemental plans. The MarketScan Commercial Database provides inpatient and outpatient medical claims linked to drug data, person-level enrollment, and benefit plan design for the working population younger than 65 years and their dependents. The Medicare Supplemental Database includes supplemental and coordination of benefits for retirees covered by their previous employers, including both employer-paid and Medicare-paid components of care (citing IBM documentation).

Study Population Inclusion/Exclusion Criteria

The study population consisted of individuals who met the following specified inclusion and exclusion criteria. The study included patients who underwent one or more knee arthroscopies for the primary International Classification of Diseases (ICD)-9 or ICD-10 diagnosis of knee OA, without concomitant pathology including loose bodies or meniscal tears. Patients were excluded from the study if they had an ICD-9 or ICD-10 code diagnosis consistent with meniscus tears or loose bodies. They were also excluded if they had a reported Current Procedural Terminology (CPT) code pertaining to meniscal débridement or loose body excision. The laterality of diagnosis and procedure was accounted for as well. All data were extracted directly from the IBM MarketScan Database, which collects data on all patient visits and procedures. Inclusion required continuous enrollment for 12 months before and after surgery to allow for analysis of interventions before and after arthroscopy. The study included data from the years 2012 to 2019. For measures evaluating trends before and after surgery requiring continuous enrollment before and/or after the procedure, we report on procedures performed from 2013 to 2018. For this study, however, there was no interpretation of data related to interventions performed in the year after arthroscopy. Individuals were excluded if they were younger than 21 years at the time of surgery.

Explanatory Variables

A multitude of variables were analyzed as a part of this study including demographic characteristics, comorbidities, and geographic area. Demographic characteristics included sex (male and female), as well as age groups 21-29, 30-39, 40-49, 50-59, 60-69, 70-79, and 80+ years. The only comorbidity analyzed was obesity based on body mass index. This study focused on variables related to overall population trends and utilization patterns among different age groups.

Descriptive Analysis

For descriptive analysis, we calculated univariate statistics for variables in the analytic file, including mean, median, percentile, standard deviation, skewness (distortion or asymmetry relative to a normal distribution), and kurtosis (heavy-tailed or light-tailed relative to a normal distribution) for continuous variables. For categorical variables, frequency distributions of values were generated, and we counted the number of diagnoses and procedures, and other pertinent measures, per individual. Where appropriate, we examined bivariate statistics such as correlations and contingency tables. The IBM MarketScan Research Databases analytics tool was used to conduct statistical analysis.

Data Analysis

To address Research Question 1—“What are the trend rates regarding the utilization of arthroscopy for treatment of knee OA by orthopaedic surgeons?”—we created tables of the values of each measure in each year, allowing for the comparison of the rates in all classes (eg, male vs. female rate). The overall rates were determined as actual rates or adjusted for demographic characteristics such as age group, sex, and census division using multivariate statistical models to isolate the time trends. Adjusted rates are useful because unadjusted rates may be affected by changes in the underlying composition of the patient sample. Changes in arthroscopic trend rates were calculated using both the number of patients per 1000 and percent change in arthroscopic utilization rates from year to year.

To address Research Question 2—“What other interventions were performed in these patients in the year preceding arthroscopic surgery?”—we (1) found the date of the arthroscopic surgery for patients meeting diagnostic criteria and other inclusion/exclusion criteria as defined in the study population section and (2) ensured that patients were continuously enrolled in the 12 months before surgery. This avoids trends being censored because of disenrollment. We created coded criteria for a list of interventions and determined the percentage of continuously enrolled patients receiving these interventions (eg, PT, VSIs, and CSIs).

Results

The trend in knee arthroscopy procedures in individuals with a diagnosis of OA aged 21 years and older shows an overall decline from 2012 to 2019 (Figure 1). The overall rate of arthroscopy in 2012 was 59.15 per 1000 patients and decreased by 14.91 points to 44.24 per 1000 patients in 2019. This decreasing trend in arthroscopic treatment of OA equates to an overall decrease of 25.2%.

Figure 1.

Plot demonstrating the rate per 1000 patients indicating an overall decreasing trend in knee arthroscopy for the diagnosis of knee osteoarthritis for all age groups from 2012 to 2019.

Arthroscopy trend rates were evaluated for individual age groups by decade from 21 to 80+ years of age. 85.6% of patients who underwent arthroscopic intervention for OA were between 40 and 69 years of age (Figure 2). The largest cohort was the 50-59 age group, accounting for 41.2% of arthroscopic procedures. The next two cohorts in decreasing order were the 60-69 and 40-49 age groups, which accounted for 23.3% and 21.1% of arthroscopic procedures, respectively. Patients aged 39 years or younger accounted for only 9.3% of arthroscopic procedures, and patients aged 70 years or older accounted for only 5.1%.

Figure 2.

Chart demonstrating the percentage of arthroscopic interventions for knee osteoarthritis by age between 2012 and 2019.

Overall trends in knee arthroscopy procedures declined for each individual age group, with the largest percentage decreases in those aged 70 to 79 years and 80 years and older from 2012 to 2019. The arthroscopy rate decreased each year on average by 4.87% in the 21-29 age group, 5.75% in the 30-39 age group, 5.93% in the 40-49 age group, 5.31% in the 50-59 age group, 5.14% in the 60-69 age group, 7.06% in the 70-79 age group, and 7.17% in the 80+ age group. Figure 3 demonstrates the decreasing trend in arthroscopy rates per 1000 patients for the three most prevalent groups including those aged 40 to 69 years.

Figure 3.

Plot demonstrating the rate per 1000 patients indicating a decreasing trend in arthroscopy for 40-49, 50-59, and 60-69 age groups from 2012 to 2019.

The 40-49 age group demonstrated a 41.5% decrease in arthroscopy rates between 2012 and 2019. During the same time frame, rates in the 50-59 age group decreased by 37.2% and those in the 60-69 age group decreased by 35.9%. Each age group by decade from 21 to 80+ years experienced the largest decrease in arthroscopy rates during the 2013-14 time frame. The 40-49 age group experienced a decrease of 9.6% in this time frame, the 50-59 age group saw a decrease by 8.6%, and the 60-69 age group saw a decrease by 9.4%. The decrease in arthroscopy rates correlates with the publication of the 2013 AAOS CPGs on knee OA. These yearly percentage decreases are presented in Table 1.

Table 1.

Yearly and Overall Percentage Decrease in Arthroscopy for 40-49, 50-59, and 60-69 Age Groups From 2012 to 2019

Year	Age 40-49	Age 50-59	Age 60-69
2012-13	−0.65%	−3.96%	−4.25%
2013-14	−9.58%	−8.60%	−9.41%
2014-15	−7.59%	−6.36%	−5.52%
2015-16	−10.02%	−7.75%	−7.39%
2016-17	−3.73%	−2.65%	−3.51%
2017-18	−4.14%	−2.44%	−0.41%
2018-19	−5.78%	−5.42%	−5.45%
2012-19	−41.49%	−37.18%	−35.94%

Overall percentages for both PT and corticosteroid knee injections in the year before arthroscopy demonstrated increases in rates in the overall population (21 years and older) and within each individual age group. When adjusted for sex, age group, and census division, these trends remained consistent. Physical therapy utilization in the year before arthroscopy for all age groups was 28.6% in 2013 and 34.5% in 2019, demonstrating a 5.9% increase in PT utilization between 2013 and 2019. The largest increase in PT utilization between 2013 and 2019 was reported in the 60-69 age group, with an increase of 17.6%—from 27.1% to 44.7%. Corticosteroid injection utilization was 49.6% in 2013 and 55.9% in 2019, demonstrating a 6.3% increase over this time. The largest increase in CSIs occurred in the 40-49 age group with an increase of 8.3%—from 46.5% to 54.8%.

The percentage of VSIs decreased in the overall population from 8.4% in 2013 to 5.8% in 2019. This decrease remained consistent for each individual age group from 21 to 69 years, demonstrating the same trend in adjusted rates. In the 70-79 and 80+ age groups, VSIs show a slight increase from 2013 to 2019 in overall and adjusted rates.

Figure 4 displays the annual percentage of patients aged 21 years and older who underwent knee arthroscopy for a diagnosis of OA, who received PT, VSI, or CSI interventions in the year before arthroscopy between 2013 and 2019.

Figure 4.

Chart demonstrating the percentage of patients who received physical therapy, corticosteroid injection, and viscosupplementation in the year before arthroscopy from 2013 to 2019. CSI = corticosteroid injection, PT = physical therapy, VSI = viscosupplementation injection

Discussion

The purpose of this study was to evaluate trends in utilization of arthroscopic lavage and/or débridement for treatment of knee OA after consistent recommendations against its use in both the 2013 and 2021 AAOS CPG on the management of knee OA. This analysis conducted in conjunction with IBM MarketScan demonstrated a steep decline in knee arthroscopy procedures across all age categories (from 21 years and older) with a primary diagnosis of knee OA between 2012 and 2019 after the publication of the 2013 CPG. The 2013 CPG issued a strong recommendation against the use of arthroscopic lavage and/or débridement in the setting of primary OA of the knee.³ This guideline did not include patients who had concomitant mechanical derangements such as meniscus tears or loose bodies.

The decreasing trend in arthroscopic utilization suggests that the strong recommendation against arthroscopic lavage and/or débridement in the 2013 AAOS CPG potentially had a notable effect on physician behavior. Each age group by decade from 21 to 80+ years experienced the largest decrease in arthroscopy rates the year after the publication of the 2013 CPG on OA of the knee. Arthroscopy in the setting of knee OA before the publication of this CPG in 2013 had been a relatively controversial topic, with relatively high utilization rates despite a lack of high-quality clinical evidence demonstrating efficacy. These findings demonstrate potential orthopaedic surgeon adherence to this CPG and a shift toward other treatments supported by the CPGs, such as PT.

Moseley et al⁵ conducted the most widely cited study regarding arthroscopic treatment of symptomatic knee OA in the New England Journal of Medicine in 2002. This high-quality randomized controlled trial was included in the rationale for the 2013 CPG. This study included 180 patients with a diagnosis of OA of the knee who were subsequently randomly assigned to arthroscopic débridement, arthroscopic lavage, or a placebo group. This study demonstrated no difference in subjective pain or patient-reported functional outcome measures between the placebo group and either of the arthroscopic groups. Objectively measured walking and stair climbing were found to be statistically poorer in the débridement group than in the placebo group when measured at 2 weeks and 1 year postoperatively.

Kirkley et al⁶ also conducted a randomized controlled trial in New England Journal of Medicine that was included in the rationale for the 2013 AAOS CPG. Patients were randomly assigned to either surgical lavage and arthroscopic débridement together with physical and medical therapy or treatment with physical and medical treatment alone. No significant differences were noted in the two primary outcome measures of the study, which were the Western Ontario and McMaster Universities OA Index score and the Short-Form 36 (SF-36) Physical Component Summary Score. The conclusion of this study was that arthroscopic surgery for knee OA provides no additional benefit over optimized physical and medical therapy.

As utilization of knee arthroscopy decreased during the study time frame, trends in other conservative treatment options such as PT, CSIs, and VSIs were also evaluated in the year before arthroscopy. It is important to consider the possibility that arthroscopic treatment trends may have decreased in part because of the recognition of alternative conservative treatment methods including PT, CSIs, and VSIs. The 2013 AAOS CPG recommended strongly in favor of supervised and/or unsupervised physical activity in patients with knee OA. Corticosteroid injections received an inconclusive recommendation in the 2013 CPG, which has since been changed to a moderate recommendation in favor of usage in the 2021 CPG. This analysis demonstrated increased utilization of both PT and CSIs in the year before arthroscopy for all age groups between 2013 and 2019. The increasing trends in utilization of these conservative treatments demonstrate potential physician adherence to the published CPG and may suggest that physicians are using treatments other than arthroscopy for OA.

A strong recommendation was given against the use of VSIs, specifically hyaluronic acid injections (HAIs) in the 2013 AAOS CPG, which was changed to a moderate recommendation against its use in the 2021 CPG.^3,4 A decrease in utilization of HAIs was observed between 2013 and 2019 when looking at the overall population. However, for 70-79 and 80+ age groups, VSI utilization in the year before arthroscopy did show a slight increase from 2013 to 2019. The overall decreasing trend in utilization of viscosupplementation once again demonstrates potential physician adherence to the 2013 AAOS CPG.

Since the publication of the 2013 AAOS CPG on nonarthroplasty management of knee OA, we have demonstrated consistent trends in line with the evidence-based recommendations. Utilization of both arthroscopy and VSIs has decreased, and utilization of PT and CSIs has increased. It is historically difficult to evaluate clinical adherence to society recommendations because of a lack of available widespread patient demographic and clinical data. Meiyappan et al⁷ conducted a retrospective review of 1096 patients seen at a single center, with the goal of evaluating their practice adherence to the 2013 AAOS CPG regarding knee OA. Adherence to the AAOS guidelines was 65%, 60%, and 40% in new/never treated, new/previously treated, and return patients, respectively. Intra-articular injections including corticosteroids and viscosupplementation were the most common conservative intervention at 32%, followed by PT at 29%. This article demonstrated inconsistent adherence to the CPG at a single center.

Carlson et al⁸ sent a survey to the members of the American Association of Hip and Knee Surgeons with five clinical vignettes, each associated with a different Kellgren-Lawrence classification. The purpose of the survey was to evaluate whether treatment responses chosen by the orthopaedic surgeons were in alignment with the 2013 AAOS CPG on knee OA. They received 345 responses with adherence rates of 80%, 82%, 21%, 50%, and 98% for Kellgren-Lawrence OA at stages 0 through 4, respectively. Intra-articular hyaluronic acid injection was the most commonly selected conservative intervention for stages 2 and 3, which was not recommended by the AAOS. Their conclusion based on the survey was that there is limited adherence to the CPGs in the orthopaedic community regarding knee OA in the moderate-to-severe stages.

Bedard et al⁹ conducted a retrospective analysis of the Humana database between 2007 and 2015 evaluating the effect that the CPG had on utilization rates of HAIs after the first edition of the CPG in 2008 and the second edition in 2013. Over one million patients with a diagnosis of knee OA were included, with 38% receiving a CSI and 12.9% receiving a HAI. Their analysis determined that rates of HAIs in patients younger than 50 years decreased by 0.25 per 100 patients per quarter year after the publication of the 2013 CPG. The rates of hyaluronic acid injections in patients older than 50 years decreased by 0.12 patients per 100 patients per quarter year. These decreases in utilization contrasted with the increasing utilization that was seen year over year before the publication of the 2013 AAOS CPG. This study demonstrated the effectiveness of the CPG related to physician adherence with reduction of HAIs administered. One interesting finding is that the trend in the use of HAIs by orthopaedic surgeons and pain specialists decreased with time after the publication of 2013 AAOS CPG but did not change among primary care physicians or nonsurgical musculoskeletal providers. This may further demonstrate the direct effect of the CPG, given that primary care specialists would not routinely be reviewing the AAOS CPGs.

This analysis was conducted using the IBM MarketScan Commercial and Medicare Supplemental Databases, which allow for the evaluation of an extremely large and diverse patient and provider population. This type of analysis achieves a much more in-depth evaluation of national trends compared with either a single-center review or a survey of individual providers. Surveys can be markedly biased based on voluntary participation and individual provider's stance on certain treatments. A single center only demonstrates a small proportion of geographic and physician data, and this study is once again often biased by the relatively small sample size of treating physicians. Our study using IBM MarketScan allowed for a more critical evaluation of the trends regarding arthroscopic intervention than either a single-center study or a survey would allow.

The decreasing trends in arthroscopic utilization are clear from our data; however, there are limitations with this study when attempting to determine the cause for the trend. It is possible that the trends in arthroscopic utilization were decreasing before publication of the CPG. We attempted to ameliorate this limitation by collecting data in the year before publication and demonstrating the largest decrease in arthroscopic utilization in the year after publication and showing maintenance of this trend through 2019. This suggests a direct correlation. A second limitation is that we did not delineate treatment options based on severity of OA. As with most large database studies, patient inclusion was based on diagnostic codes that do not specify severity. Another inherent limitation to large database studies is dependence on accurate coding by a physician, which is never guaranteed. It is possible that patients who underwent arthroscopy as part of a much large procedure such as a high tibial osteotomy may have been included in our study if they did not have any diagnoses or treatment of meniscal pathology or loose bodies. Finally, it is possible that surgeons experienced a shift in behavior, assigning patients a diagnosis of an internal derangement such as a meniscal tear or loose body instead of OA to avoid a denial from the payors. Such a change in behavior would not be captured in our study because these patients were excluded.

In conclusion, this analysis suggests a change in physician practices potentially influenced by the 2013 AAOS CPG, which strongly recommended against arthroscopic treatment for an isolated diagnosis of knee OA. Both overall and age cohort–specific arthroscopic utilization rates consistently decreased year over year for all cohorts after the publication of the 2013 AAOS CPG. The secondary analysis demonstrated an overall decrease in VSI utilization, with a corresponding increase in utilization of other conservative treatment options including CSIs and PT.

This study demonstrates that the AAOS' efforts to produce robust evidence-based CPGs have had an effect on physician behavior and as such contribute to the AAOS strategic plan of improving quality of care for our patients.