Patients and physicians may be reluctant to abandon widely used treatments that have been found to be ineffective. In 2002 and 2008 the New England Journal of Medicine published the results of clinical trials showing that arthroscopic debridement and lavage—surgical treatments to remove damaged tissue and debris—do not benefit patients with osteoarthritis of the knee. To determine whether the trials’ publication was associated with changes in practice patterns, we examined ambulatory surgery data from Florida and found that the number of debridement and lavage procedures per 100,000 adults declined 47 percent between 2001 and 2010. The reduction translates into national savings of $82–$138 million annually. These reductions may be offset by increases in the use of other procedures. The results indicate that clinical trials of widely used therapies can lead to cost-saving changes in practice patterns.

Clinical trials and comparative effectiveness studies of existing treatments have the potential to reduce costs by identifying treatments that are no better than less expensive alternatives. Cost savings, however, will be realized only if patients, providers, and payers change their behavior in response to new evidence. The same characteristics of the health system that promote the adoption of new, untested technologies—specialization and organizational fragmentation, fee-for-service reimbursement, and third-party payment for medical care—may slow the rate at which existing therapies found to be ineffective are actually abandoned.

Arthroscopic debridement and lavage for osteoarthritis of the knee provide a good case study for assessing the impact of clinical trials of standard medical therapies on practice patterns. Two randomized clinical trials have found that these procedures do not relieve pain or improve function relative to no treatment for osteoarthritis of the knee. In this article we assess the trials’ impact on the use of this surgical treatment.

Background

Arthroscopic debridement and lavage are minimally invasive procedures, normally done in the same operation, to treat damage to the knee joint. A patient undergoing arthroscopic surgery receives several small incisions in the knee. The surgeon inserts a lens attached to a camera (the arthroscope) to inspect the structures and surfaces of the knee joint. After validating the initial diagnosis, which is based on the patient’s history and a physical examination as well as x-ray and magnetic resonance imaging findings, the surgeon will wash out debris (lavage) and remove (debride) torn or damaged soft tissue (for example, articular cartilage and menisci).

During the 1990s debridement and lavage were widely used to treat osteoarthritis of the knee. In the 2000s the New England Journal of Medicine published results of two randomized controlled trials of arthroscopic debridement and lavage for patients with osteoarthritis. Exhibit 1 reports a timeline of the publication of the trials and subsequent changes in payers’ coverage policies.

Exhibit 1 Major Trials And Coverage Events For Arthroscopic Debridement And Lavage For Osteoarthritis Of The Knee Date Event July 11, 2002 Trial 1 (Moseley et al.) is published: “The outcomes after arthroscopic lavage or arthroscopic debridement were no better than those after a placebo procedure” a October 10, 2002 CMS initiates a National Coverage Determination process July 3, 2003 CMS releases decision memo: “CMS has determined that the evidence is adequate to conclude that arthroscopic lavage alone is not reasonable and necessary for patients with osteoarthritis of the knee; therefore, we intend to issue a national noncoverage determination” b September 26, 2003 The insurer Aetna revokes coverage of arthroscopic surgery for people with “with knee pain only or with severe osteoarthritis” c September 11, 2004 CMS releases National Coverage Determination: Arthroscopic surgery for osteoarthritis is a “noncovered” service d September 11, 2008 Trial 2 (Kirkley et al.) is published: “Arthroscopic surgery for osteoarthritis of the knee provides no additional benefit to optimized physical and medical therapy” e

The first trial, conducted by J. Bruce Moseley and colleagues and published in July 2002, randomized 180 male patients with moderate to severe osteoarthritis of the knee to arthroscopic lavage, arthroscopic debridement, or sham surgery. 1 A single surgeon performed all procedures. Neither treatment relieved pain or improved function relative to sham surgery at any time during the two year follow-up period.

Based on the trial, the Centers for Medicare and Medicaid Services (CMS) initiated a coverage review process in 2002. CMS then released a decision memo in July 2003 stating its intent to discontinue coverage of arthroscopic debridement and lavage for patients with severe osteoarthritis or knee pain only. CMS would continue to cover debridement and lavage for patients with “less severe and/or early degenerative arthritis.” 2 Many private insurers also withdrew coverage. CMS issued its National Coverage Determination in June 2004. 3

In September 2008 the New England Journal of Medicine published a second trial with more far-reaching treatment implications. Alexandra Kirkley and colleagues randomized 188 patients with osteoarthritis of the knee to physical and medical therapy alone versus physical and medical therapy plus arthroscopic surgery. 4 Patients with large meniscal tears were excluded.

The researchers concluded that “surgery…provides no additional benefit to optimized physical and medical therapy.” 4(p1097) The study included female patients, multiple surgeons, and patients with mild osteoarthritis. However, some surgeons have argued that the trial, like that of Moseley and colleagues before it, 1 lacked external generalizability because patients who were good candidates for surgery were not considered for randomization. 5

Although both trials were widely publicized, it is unclear whether they led to a decline in the use of arthroscopic debridement and lavage. Patients, most of whom pay only a small share of the cost of the procedure out of their own pockets, often expect knee arthroscopy to be a part of their treatment and, in our experience, lack enthusiasm for the more involved and slower course of conservative therapy.

Additionally, physicians face financial incentives to perform arthroscopy. Although many insurers withdrew coverage of lavage and debridement for some indications, enforcement may have been hampered by the imprecision of diagnostic coding. For example, codes do not indicate the severity of osteoarthritis and meniscal tears.

To examine trends in debridement and lavage, a recently published study used the case logs routinely collected for surgeons taking the American Board of Orthopaedic Surgery’s examination for board certification. 6 The number of arthroscopic knee procedures in patients with osteoarthritis declined from 2.36 per surgeon in 2001 to 1.40 in 2009. However, the sample was limited to surgeons who were in their first few years of practice. In this study we examined trends in debridement and lavage using a data set that includes procedures performed by all types of surgeons.

Study Data And Methods

Data Source

We measured trends in the rate of arthroscopic debridement and lavage in outpatient surgery centers from 1998 to 2010 using Florida’s State Ambulatory Surgery Database. We selected Florida for the analysis because it is a large, demographically diverse state. The data capture 100 percent of outpatient surgeries and include information such as patient demographics and diagnosis and procedure codes that are typically found in insurance claims databases. The data include procedures performed in freestanding ambulatory surgery centers as well as hospital-based outpatient surgery units.

We identified arthroscopic debridement and lavage procedures using Current Procedural Terminology (CPT) codes 29877 (debridement) and 29871 (lavage) and International Classification of Diseases , Ninth Revision (ICD-9), procedure code 80.86.

We also identified other types of arthroscopic knee procedures and arthroscopic shoulder procedures using the relevant CPT and ICD-9 codes, described in online Appendix Exhibit 1. 7 Examining trends in the use of other arthroscopic joint procedures is useful for gauging underlying trends in the demand for and use of knee arthroscopy.

The vast majority of arthroscopic knee procedures are meniscectomies to remove torn meniscal tissue. Tears in the menisci, which are cartilaginous disks in the knee joint, may cause pain and limit mobility. Patients who undergo arthroscopy for osteoarthritis may undergo a meniscectomy if the surgeon observes torn meniscal tissue during the procedure. These procedures would be billed using codes for debridement and meniscectomy (CPT codes 29880 and 29881) and were not counted in the “debridement and lavage” category for the purposes of our analysis.

Analysis

We report age- and sex-standardized trends in quarterly procedure rates per 100,000 adults. We standardized rates to the 2000 Florida population. Osteoarthritis is not recorded consistently on claims, so we do not report diagnosis-specific trends in procedure rates. The database captures a 100 percent sample of outpatient procedures, and thus quarter-to-quarter differences in procedure rates are not influenced by sampling variability.

The data do not include observations for patients with osteoarthritis of the knee who did not have surgery. Thus, we cannot examine the impact of trial results on the likelihood that patients with osteoarthritis undergo debridement and lavage.

Following Cary Gross and colleagues, 8 we calculated the significance of changes in procedure rates using a regression spline approach. We calculated quarterly procedure rates from 1998 to 2010 ( N = 52 ) and then estimated an ordinary least squares regression of the rates as a function of a variable that indicates the period after publication of Moseley and colleagues’ trial 1 and a time-trend variable that is allowed to vary before and after those results were released. Details are provided in the Appendix. 7

Study Results

Trends In Procedure Rates

The data include observations for 67,878 debridement and lavage procedures and 653,774 arthroscopic knee surgeries, excluding debridement and lavage. The average age of patients undergoing debridement and lavage between 1998 and 2010 was forty-four, and 45 percent were male. Ten percent were insured by Medicare, 65 percent were covered by private payers, and 25 percent were covered by other payers or uninsured (see Appendix Exhibit 2 for further detail). 7

Exhibit 2 displays quarterly procedure rates. Rates of debridement and lavage procedures declined 47 percent between 2001 and 2010. (Appendix Exhibit 3 displays annual procedure volumes and rates). 7 Rates dropped sharply following the publication of Moseley and colleagues’ results 1 and then again after CMS issued a decision memo stating its intent to discontinue coverage. 3 Trends were similar across major payer types (see Appendix Exhibit 4). 7 Exhibit 2 Trends In Arthroscopic Procedures Of The Knee Per 100,000 People Ages 18 And Older, By Quarter, 1998–2010 y axis. Other arthroscopic procedures are denoted by the red graphing line and relate to the right-hand y axis. a See Note b CMS is Centers for Medicare and Medicaid Services. See Note c NCD is National Coverage Determination. See Note d See Note SOURCE Authors’ analysis of data from the Florida State Ambulatory Surgery Database. NOTES Rates are standardized by age and sex to the 2000 Florida population. Debridement/lavage is denoted by the blue graphing line and relates to the left-handaxis. Other arthroscopic procedures are denoted by the red graphing line and relate to the right-handaxis.See Note 1 in text).CMS is Centers for Medicare and Medicaid Services. See Note 2 in text.NCD is National Coverage Determination. See Note 3 in text.See Note 4 in text.

The number of arthroscopic knee procedures per 100,000 people, excluding debridement and lavage but including meniscectomy, decreased less than 1 percent between 2001 and 2010. During that same period, the number of shoulder arthroscopies per 100,000 increased 85 percent (data not shown).

The number of debridement and lavage procedures per 100,000 decreased 15 percent between 2008 and 2010 (the trial conducted by Kirkley and colleagues 4 was published in September 2008). The number of arthroscopic knee procedures per 100,000, excluding debridement and lavage, decreased 6 percent over the same period.

The number of shoulder arthroscopies per 100,000 increased 3 percent between 2008 and 2010, but the rate of growth was slower than it had been earlier in the decade. For example, the number of shoulder arthroscopies per 100,000 increased 11 percent between 2006 and 2008. The United States entered a recession in December 2007, which may also have dampened demand for elective surgery.

The regression model indicates that there was a significant decrease in lavage and debridement procedure rates following the publication of Moseley and colleagues’ results ( p < 0.001 ). 1 There was also a significant increase in the use of other types of arthroscopic procedures of the knee ( p < 0.001 ). (Appendix Exhibit 5 provides full results.) 7

Exhibit 3 displays trends in the number of arthroscopic knee procedures (all types, including debridement and lavage procedures and meniscectomies) per 100,000 by age group. In each age group, publication of Moseley and colleagues’ trial results 1 was associated with an immediate decline in arthroscopic knee procedure rates. Surprisingly, rates continued their downward trend among patients ages 18–44. After an initial decline, rates increased among patients age forty-five and older, although it is difficult to discern trends, given the pronounced seasonality of rates. Rates declined after 2008 in both age groups. Exhibit 3 Trends In Arthroscopic Procedures Of The Knee Per 100,000 People, By Age Group And Quarter, 1998–2010 a See Note b CMS is Centers for Medicare and Medicaid Services. See Note c NCD is National Coverage Determination. See Note d See Note SOURCE Authors’ analysis of data from the Florida State Ambulatory Surgery Database. NOTE Rates are standardized by age and sex to the 2000 Florida population.See Note 1 in text).CMS is Centers for Medicare and Medicaid Services. See Note 2 in text.NCD is National Coverage Determination. See Note 3 in text.See Note 4 in text.

Impact On Health Care Costs

We calculated the impact of changes in the use of arthroscopic debridement and lavage on costs by applying the 2001 and 2010 rates of debridement and lavage procedures per 100,000 to the 2010 US population, which imply a reduction of 55,000 procedures annually. Our calculations are described in Exhibit 4 .

Exhibit 4 Estimates Of Savings From Reduction In The Use Of Debridement/Lavage Between 2001 And 2010 Description 2001 2010 Rate per 100,000 50.1 26.6 Number of procedures a 117,254 62,157 If the cost per procedure is $1,500: Spending (millions) $175.9 $93.2 Savings (millions) — b $82.6 If the cost per procedure is $2,500: Spending (millions) $293.1 $155.4 Savings (millions) — b $137.7

We used $1,500 and $2,500 as low and high estimates of the cost per debridement procedure. These figures are reasonable based on Medicare reimbursement rates 9 and charges to private insurers, 10 but the actual amount may be higher or lower. Moseley and colleagues 1 stated that the cost of arthroscopy was $5,000 per procedure, but we were unable to find a reference or support for this figure in publicly available payment rate databases. 9,10

We estimate that nationwide spending on debridement and lavage declined $82.6–$137.7 million annually over the ten-year period. Kirkley and colleagues’ trial 4 compared surgery and therapy to therapy alone, but it is possible that some patients may view surgery and physical therapy as substitutes. If patients received therapy at a per patient cost of $788 11 instead of debridement or lavage, savings would be $39.2–$94.3 million annually.

Discussion

Use of arthroscopic debridement and lavage declined immediately following publication of the results of a trial conducted by Moseley and colleagues in 2002, 1 before insurers changed their coverage policies, and then again in the second half of 2003, following the release of the CMS decision memo. These results suggest that changes in coverage policies are partly but not entirely responsible for declines in procedure volume. Rates also declined following publication of the results of Kirkley and colleagues’ trial 4 in 2008.

Many patients in Florida continue to receive debridement and lavage: More than 3,800 debridement and lavage procedures were performed in 2010. Direct measurement of appropriateness would require a review of all patients’ medical records, which would be prohibitively expensive.

Prominent payers, including Medicare and Aetna, responded to Moseley and colleagues’ trial 1 by issuing policies specifying termination of coverage for arthroscopic debridement and lavage for patients with severe osteoarthritis in 2003 and 2004. 3,12 Insurance claims, however, do not record the severity of osteoarthritis or meniscal tears, and so these policies can be difficult to enforce. In some cases, insurers may unknowingly pay for noncovered procedures. In other cases, insurers may deny claims for medically appropriate arthroscopic surgery, such as surgery for patients with symptomatic meniscal tears.

We focused on debridement and lavage because the impact of the trials, if any, should be most readily apparent for this group of procedures. However, it is also important to consider the impact of the trials on all types of arthroscopic procedures of the knee. Publication of the trials’ results may have reduced the number of meniscectomies among patients with osteoarthritis where repair of a torn meniscus was not the primary goal of the procedure. Alternatively, physicians may have begun to perform more meniscectomies in patients with osteoarthritis to circumvent insurers’ coverage restrictions.

Based on the trends depicted in Exhibit 2 , it does not appear that surgeons increased the use of meniscectomy to compensate for the decline in debridement and lavage procedures. However, the number of meniscectomies is large in relation to the number of stand-alone debridement and lavage procedures, so it is difficult to determine whether declines in the use of debridement and lavage were offset by increases in meniscectomies.

Procedure rates for all arthroscopic procedures of the knee among patients age forty-five and older continued to increase after publication of Moseley and colleagues’ trial results, 1 while rates among patients ages 18–44 decreased. This finding was surprising. The prevalence of osteoarthritis increases with age, so we expected procedure rates to decline in the older age group. 13 It is possible that among patients with osteoarthritis, younger patients were more likely to undergo arthroscopic surgery before Moseley and colleagues published their results. Older patients may be more accepting of the activity limitations that accompany osteoarthritis and less likely to pursue aggressive therapy.

We estimate that the reduction in the per capita rate of arthroscopic debridement and lavage procedures between 2001 and 2010 translates into savings of $82–$138 million annually. Hypothetically, actual savings would be higher under three scenarios: if the per procedure cost exceeded $2,500; if, in the absence of the two relevant trials, 1,4 procedure volume had increased through 2010, given that the prevalence of symptomatic knee osteoarthritis was increasing over this period; 14 or if the trials depressed the use of other types of arthroscopic procedures in patients with osteoarthritis of the knee.

Savings would be lower if declines in the use of arthroscopic debridement and lavage were offset by increases in the use of physical therapy, other arthroscopic procedures, or total knee replacement surgeries.

Implications For Studies Reporting ‘Negative’ Results

The trials of arthroscopic debridement and lavage are examples of studies reporting “negative” results—that is, where patients receiving the target therapy did not fare better than those in the control condition. Although the studies were not comparative effectiveness studies in the strict sense of the term, understanding their impact on practice patterns is helpful for gauging the potential impact of comparative effectiveness research on spending.

Savings will materialize only if patients and physicians are willing to abandon ineffective treatments. Physicians practicing in a fragmented, fee-for-service system, however, have little incentive to abandon the treatments that help define their specialty and provide a source of income. Likewise, patients may be reluctant to forgo aggressive and expensive treatments if they pay only a small share of the cost of the treatments out of pocket.

Employers and insurers stand to gain when trials lead physicians to abandon costly treatments, but insurers’ ability to influence treatment patterns is limited. 15 In the case of arthroscopic knee procedures, insurance claims do not contain sufficient detail to determine whether patients are appropriate candidates for surgery. Costly review of patients’ medical records and imaging studies is needed to ascertain appropriateness.

Previous studies of the impact of negative trial results have found some instances in which trials led to rapid changes in practice patterns 16 and others where practice patterns changed slowly 17 or not at all. 18 Physicians and patients may be more likely to abandon treatments that are associated with side effects or patient discomfort than those that are simply ineffective but cause no harm. Complications from arthroscopic knee surgery are rare, but, like all surgeries, the procedure can lead to infections and blood clots and entails a recovery period.

Some orthopedic surgeons questioned whether Moseley and colleagues’ trial, 1 which enrolled only male veterans age seventy-five and older, and Kirkley and colleagues’ trial, 4 which took more than six years to enroll subjects, were externally generalizable. Other trials reporting negative results have been subject to similar criticisms, which potentially blunts their impact. 15 Inclusion of broader patient populations, as is required for grantees by the newly formed Patient-Centered Outcomes Research Institute, should facilitate the translation of results into practice.

Our analysis suggests that trials and comparative effectiveness studies targeting widely used therapies of uncertain value have the potential to reduce costs. Through a combination of publication of randomized controlled trials, information dissemination to providers and insurance carriers, and changes in coverage policies, the use of arthroscopic debridement and lavage declined 50 percent following publication of negative trial results. Many older patients continue to receive arthroscopic surgery, so there may be additional opportunities to reduce use.

ABOUT THE AUTHORS: DAVID HOWARD, ROBERT BROPHY & STEPHEN HOWELL In this month’s Health Affairs , David Howard and coauthors report on their study of the impact of clinical trials showing that a certain surgical treatment for knee osteoarthritis—arthroscopic debridement and lavage—did not benefit patients. The trial results, published in 2002 and 2008, prompted payers like Medicare to discontinue coverage of the procedure for patients with severe osteoarthritis. Howard and coauthors, examining ambulatory surgery data from Florida, found a 47 percent decline in the use of the procedure between 2001 and 2010 that nationally may have translated into $134 million in annual savings. Howard is an associate professor in the Department of Health Policy and Management at Emory University. A health economist by training, Howard conducts research using economics and statistics to better understand physicians’ decision making and its implications for public policy. He is currently studying the impact of comparative effectiveness research on medical practice patterns and the value and pricing of new anticancer therapies. Howard has acted as an adviser or consultant to the Medicare Payment Advisory Commission, which advises Congress on Medicare issues; the American Cancer Society; the Division of Transplantation, Department of Health and Human Services; and the Institute of Medicine. Howard received his doctorate in health policy from Harvard University. Robert Brophy is an assistant professor of sports medicine at the Washington University School of Medicine in St. Louis, where his clinical focus is on shoulder and knee injuries. He has a particular interest in the treatment of shoulder instability and rotator cuff tears and meniscus, articular cartilage, and multiligament injury in the knee. Brophy played on two national championship teams in the United Soccer League (1992 and 1996) before enrolling in medical school at Washington University in St. Louis. He entered residency in orthopedic surgery at the Hospital for Special Surgery in New York City. Brophy is a team physician for the St. Louis Rams football team. He has also been a team physician with the St. Louis Blues hockey team and the head team physician for the St. Louis Athletica of Women’s Professional Soccer. Stephen Howell is a professor of mechanical engineering at the University of California, Davis. He is also a clinician, researcher, and innovator in total knee replacement, anterior cruciate ligament reconstruction, and meniscal injury. At his clinical practice in Sacramento, California, he focuses on the treatment of degenerative processes and sports-related injuries to the knee. Howell is a paid consultant for his development and clinical application of “kinematic alignment” in total knee replacement. He also is a paid consultant and receives royalties for his development of anterior cruciate ligament reconstruction instruments and products. He has been awarded eleven US patents, two European patents, and eight patents pending. He is on the editorial board of the American Journal of Sports Medicine . Howell received a medical degree from Northwestern University.

NOTES