CMRO
Current Medical Research & Opinion Volume 26, Number 5 May 2010

Original Article

Knee-attributable medical costs and risk of
re-surgery among patients utilizing non-surgical
treatment options for knee arthrofibrosis

Judith J. Stephenson Ralph A. Quimbo Tao Gu HealthCore, Inc., Wilmington, Delaware, USA Address for correspondence: Judith J. Stephenson, SM, HealthCore, Inc., 800 Delaware Avenue Fifth Floor, Wilmington, DE 19801-1366, USA. Tel.: þ1 302 230 2142; Fax: þ1 302 230 2020; jstephenson@healthcore.com Key words: Cost – High and low intensity stretch devices – Knee arthrofibrosis – Medical resource use – Non-surgical treatment options Accepted: 4 February 2010; published online: 12 March 2010 Citation: Curr Med Res Opin 2010; 26:1109–18
Abstract
Objective:
To determine if differences in costs and risks of re-hospitalization and/or re-operation exist between arthrofibrosis patients treated with low intensity stretch (LIS) or high intensity stretch (HIS) mechanical therapies, or physical therapy alone (No Device). Study Design: This observational cohort study utilized administrative claims data to identify arthrofibrosis patients, age 565 years, with continuous enrollment for the 6 months prior to and following the index knee event date. Methods: The index knee event was defined as the knee injury/surgery preceding device use for the LIS and HIS groups and the knee injury/surgery prior to the diagnosis of arthrofibrosis for the No Device group. Knee- attributable medical costs (KAMC), accrued over 6-month pre- and post-index periods, as well as risks of re-operation, re-injury, and re-hospitalization were compared between groups. Multivariate models were used to evaluate group differences in utilization and costs when controlling for age, sex, and comorbidities. Results: A total of 60 359 patients (143 HIS; 607 LIS; 59 609 No Device) met the inclusion criteria. Unadjusted post-index KAMC were significantly less ( p 50.0001) among HIS patients ($8213 10 576) relative to LIS ($16 861 17 857) and No Device ($9345 14 120) patients. A significantly greater percentage of LIS Device patients had total knee replacements than HIS Device or No Device patients, and the LIS group had a significantly higher percentage of patients with musculoskeletal disease. When controlling for these group differences, the multivariate predictive model results were similar to the unadjusted results, with greater post-index KAMC for the LIS patients (24%, p 1⁄4 0.025) and No Device (9%, p 1⁄4 0.323) relative to HIS patients. No Device patients were 71% ( p 50.0001) more likely to have a subsequent knee event than HIS patients, and HIS patients had significantly lower rates of re-hospitalization than LIS and No Device patients (p50.0001). Conclusions: Patients treated with HIS mechanical therapy demonstrated significantly reduced rates of re-hospitalization which corresponded to reduced knee-attributable medical costs. Limitations: Limitations of this study include those inherent in the use of retrospective claims data to identify the cohorts and for analytic purposes. The authors attempted to control for these as much as possible with the multivariate analyses, and inclusion of the model covariates specified above demonstrated a scaled deviance of 1.16 indicating a reasonable goodness-of-fit for the selected model covariates.
@2010 Informa UK Ltd www.cmrojournal.com
Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al. 1109
Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
Introduction
The number of major knee surgeries performed each year in the United States (US) has continued to increase dramatically over the past decade1,2, and is expected to continue to escalate in the future3–5. Varying degrees of motion loss are common following major knee surgical procedures such as total knee arthroplasty (TKA) or ante- rior cruciate ligament (ACL) reconstruction6–9. Mild forms of motion loss have been attributed to postoperative inflammation and pain, and usually resolve with conven- tional treatments of aggressive outpatient physical therapy and/or nonsteroidal anti-inflammatory drugs (NSAIDs). However, between 1% and 17% of TKA and ACL recon- struction patients have been reported to develop a severe form of motion loss termed arthrofibrosis10–14. The rate of arthrofibrosis has been reported to be doubled or even tripled for patients who have undergone reconstruction for multiple ligament injuries9,15–17 or who have sustained injuries or have undergone surgical procedures that are associated with prolonged immobilization such as tibial plateau18 or supracondylar femoral fractures19,20. Using the current annual rates of TKA and ACL reconstruction procedures4,5, as well as the reported rates of postoperative arthrofibrosis13,14, the number of major knee surgery patients affected by arthrofibrosis in the US can be con- servatively estimated to be 85 000 per year. The most common cause of chronic postoperative arthrofibrosis is a biological process that results in abnor- mal, excessive scarring. As a result of an increased number and activity of fibroblasts21,22, scar tissue adhesions form across the suprapatellar pouch, between the extensor mechanism and the peripatellar gutters, and/or between the proximal patella and anterior femur23. These adhe- sions restrict the degree of knee flexion, which in turn limit a patient’s ability to perform such activities of daily living as navigating stairs, rising from a chair, or entering and exiting a car24–27. Patients with arthrofibrosis often present clinically with pain, limited range of motion, and functional limitation as described above. Other common symptoms include anterior knee pain, abnormal gait, quadriceps fatigue, and difficulty walking long distances13,25. Surgical treatment options for patients with arthrofi- brosis are limited, and include manipulation under anes- thesia, arthroscopic and open lysis of adhesions, or revision surgery28,29. Although generally successful, approximately 25% of arthrofibrosis patients treated with a motion- restoring surgical procedure require multiple surgeries29,30, and satisfactory results were only reported by 37% of patients surgically treated for more diffuse arthrofibrosis that affected more than just the anterior compartment of the knee31. Using the annual incidences of arthrofibrosis previously discussed, this equates to approximately 21 000 patients each year who are at risk of requiring additional
surgery to restore motion. An additional 54 000 patients per year are at risk of an unsatisfactory outcome, defined as demonstrating either an ‘abnormal’ or ‘severely abnormal’ result according to the criteria of the International Knee Documentation Committee (IKDC), which includes the arc of motion, joint stability, patient-reported subjective symptoms, and the physician’s clinical findings31. In addi- tion to an increased risk of re-operation, these procedures are not indicated throughout the period of recovery. Manipulations are not recommended after 12 weeks fol- lowing the initial surgery, as there is a risk of rupturing the patellar tendon or fracturing the patella due to the increased tensile strength of the adhesions8,25,28. Invasive revision procedures are not recommended in the first 6 months after the initial procedure29 and often occur well after 1 year29,32. With the limited availability and inherent risks asso- ciated with motion-restoring surgical options, conserva- tive non-operative treatment options are stressed in the early postoperative period. Instead of improving motion by breaking apart or releasing scar tissue adhesions during surgical interventions, non-operative treatments attempt to regain motion by stretching the tissue. For patients who do not respond to physical therapy alone, adjunctive mechanical therapy has been reported to be effective at restoring motion and reducing the risk of re-operation24,25,28,33. The efficiency of mechanical ther- apy can be improved by increasing the number of treat- ment sessions per day or week, the duration of time spent stretching during each session, or the intensity of the torque being delivered to the tissue during each session34. Mechanical therapy devices may be categorized as either low intensity stretch (LIS) or high intensity stretch (HIS) devices, based on the applied torque delivered to the joint. Devices that provide lower torque than that of the physical therapist are considered LIS devices, whereas devices that mimic the torque applied by physical thera- pists are considered HIS devices35. HIS devices apply a load to the knee using either hydraulics or pneumatics, and HIS treatment protocols require the patient to stretch near or at the pain threshold. LIS devices are either spring-loaded or apply a load using gearing, and the lower torque applied by LIS devices may produce less strain for the patient. However, LIS devices are generally associated with longer treatment sessions (6–8 hours vs. 1 hour) as well as a greater number of weeks of use than has been reported with HIS devices33,36. Because it is unclear if one treatment philosophy is more beneficial than the other for knee arthrofibrosis patients, the purpose of this study was to determine whether differences in cost and the risks of re-hospitalization and re-operation exist between knee arthrofibrosis patients treated with LIS mechanical therapy devices, HIS mechanical therapy devices, or phys- ical therapy alone (No Device).
1110 Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al.
www.cmrojournal.com @2010 Informa UK Ltd

CMRO
Current Medical Research & Opinion Volume 26, Number 5 May 2010
Methods Data source This observational cohort study utilized administrative claims data from the HealthCore Integrated Research Database (HIRD) for services incurred between January 1, 2004 and March 31, 2008, to identify knee arthrofibrosis patients under the age of 65 years with at least 12 months of continuous medical health plan eligibility. The HIRD is a broad, fully-integrated database consisting of the eligi- bility, medical and pharmacy claims of approximately 30 million patients from 14 geographically-dispersed US health plans in the northeast, midwest, south, and west regions of the US All study materials were handled in compliance with the Health Insurance Portability and Accountability Act (HIPAA) of 1996 and a limited data- set was utilized for the analyses. Patient identification and treatment cohort assignment All patients selected for the study were initially identified from a review of available medical claims. Patients were required to have at least one medical claim with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code(s) and/or Current Procedural Terminology (CPT) code(s) for knee surgery or injury (Appendix A lists the specific knee sur- gery and knee injury codes that were used) during the patient identification period from January 1, 2003 through March 31, 2007 (Figure 1). This initial event is called the index knee event. A hierarchical algorithm, based on the type of therapy utilized subsequent to the index knee event, was used to assign arthrofibrosis patients to one of the three treatment cohorts (HIS, LIS, and No Device). Arthrofibrosis is brought about by an excessive synovial inflammatory response followed by increased activation and proliferation of fibroblastic cells and a significant increase in extracel- lular matrix proteins21. The excessive inflammatory reac- tions associated with arthrofibrosis may be caused by a number of stimuli, including trauma, surgery, or infec- tion37. For this reason, the index knee event was defined as either the knee injury or surgery that immediately pre- ceded the diagnosis of arthrofibrosis. HIS device users were identified first, followed by LIS device users; the No Device cohort consisted of patients who were neither HIS not LIS users. For the LIS and HIS device cohorts, the index date was defined as the date of the knee event immediately preceding initial device use. For the purpose of this study, the index knee event could be either a surgery or injury for the reasons described above. For the No Device cohort, the index date was defined as the date of the knee event prior to the diagnosis of arthrofibrosis. The HIS device protocol required a diagnosis of arthrofibrosis for its use. After identifying the HIS patients, the remaining LIS and No Device patients were required to have a diagnosis relating to osteoarthrosis, ankylosis, contracture/fracture, or stiffness in the lower leg in order to make the groups more compar- able in terms of severity (See Appendix A for the diagnosis codes). Patients were required to have at least 6 months con- tinuous medical coverage pre- and post-index date. Full pharmacy coverage was not required for study inclusion given the nature of the treatments of interest, as well as the clinical (re-surgery) and economic (medical costs) out- comes of interest. Outcomes assessment The major outcomes of this study were the 6-month post-index knee-attributable medical costs (KAMC) and the incidence and risk of knee surgery/injury subsequent to the index knee event. KAMC consisted of health plan and patient paid costs of all inpatient, emergency room (ER) and outpatient encounters. In addition to office visits and diagnostic procedures, outpatient encounters also included all post-surgery rehabilitative physical therapy encounters that were billable to the health plan. Medical claims with codes relating to knee device use were not included as part of KAMC in order to assess the non-device medical cost burden that was observed among the three treatment cohorts. Additionally, since full pharmacy coverage was not mandatory for all study patients (only medical health plan eligibility was an inclusion requirement), pharmacy costs were not included as part of the assessment of KAMC. Therefore, it was assumed that pharmacotherapy patterns for the treatment of the injuries under investiga- tion were similar between treatment groups, with medical encounters relating to surgery and physical therapy consti- tuting the majority of all derived economic outcomes. In addition to patient characteristics, the prevalence of con- current diseases related to knee injury and surgery (i.e., total knee surgery (ICD-9-CM: 81.54, CPT: 27447), osteoarthritis (ICD-9-CM: 715.X6), and musculoskeletal disease (Appendix A)) was assessed over the 6-month period prior to the index date. Similarly, the overall co-morbidity burden was assessed using the Deyo– Charlson co-morbidity index (DCI)38 based on an assess- ment of medical claims over the same 6-month period. Statistical analysis Statistical comparisons between cohorts were performed in two stages. The first stage consisted of bivariate or unad- justed comparisons in which continuous characteristics were compared using appropriate analysis of variance
@2010 Informa UK Ltd www.cmrojournal.com
Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al. 1111

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
(ANOVA) methodology and categorical variables were compared using chi-square tests. Right-censored continu- ous outcomes were initially presented via Kaplan–Meier curves stratified for each treatment group. The second stage incorporated the use of multivariate regression models to determine statistically significant differences in outcomes while adjusting for potential confounding variables. Two main outcomes were compared among treatment cohorts using multivariate analysis: KAMC and the incidence of post-operative surgical intervention. The KAMC of the three treatment cohorts, accrued over a uniform 6-month pre- and post-index period, were compared utilizing a generalized linear model (GLM) with a gamma distribution and logarithmic link function. This technique was utilized because of the skewed nature of healthcare cost data39. In addition, the multivariate cost model was used as a predictive model to estimate the semi-annual KAMC burden to the health plan for each treatment cohort. In this model, the primary covariate of interest was the treatment cohort while additional model covariates included: age, gender, pre-period DCI score, pre-period KAMC, occurrence of total knee replacement on index date, pre-period osteoarthritis and pre-period musculoskeletal disease. The incidence of a post-operative surgical intervention in the time period from after the index surgery to the end of study follow-up was assessed using a Cox proportional hazards model. In this model, the primary covariate of interest was the treatment cohort while additional model covariates included: age, gender, pre-period DCI score, occurrence of a total knee replacement on index date, pre-period osteoarthritis and pre-period musculoskeletal disease. All statistical analyses were conducted using SAS ver- sion 9.1 (Version 9.1 for Windows, SAS Institute Inc., Cary, NC, USA). An a priori a-level of p 0.05 was con- sidered significant. Results A total of 60 359 patients (143 HIS Device; 607 LIS Device; 59 609 No Device) met the study inclusion criteria (Table 1). The mean patient age ranged from 48.1 years for LIS Device patients to 50.4 years for No Device patients. Less than half of the patients in two of the three cohorts (LIS and No Device) were female. A significantly greater (p50.0001) percentage of LIS Device patients (46.3%)
1112 Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al.
www.cmrojournal.com @2010 Informa UK Ltd

CMRO
Current Medical Research & Opinion Volume 26, Number 5 May 2010
had total knee replacement surgery at index than HIS Device (11.9%) or No Device patients (19.0%). In terms of baseline co-morbidities, LIS Device patients had signif- icantly more (p50.0001) musculoskeletal disease (87.5%) than patients in the HIS Device (79.0%) or No Device (79.8%) cohorts. Although LIS Device patients had more baseline osteoarthritis (71.8%) than patients in the HIS device (59.4%) and No Device (56.6%) cohorts, the dif- ferences were not statistically significant (p 1⁄4 0.0960). Figure 2 shows that the three treatment groups differed significantly (p50.0001) in unadjusted KAMC for both the pre- and post-index periods. In the pre-period, HIS Device patients had the highest mean costs ($11 950 17 767) followed by LIS Device patients ($9605 27 515) and No Device patients ($2118 8512). However, for the 6-month post-index period, mean KAMC were lowest for HIS Device patients ($8213 10 576) followed by No Device patients ($9345 14 120) and LIS Device ($16 861 17 857). The primary drivers of these differences were the incidence and cost of knee-attributable inpatient hospitalizations. More specifically, pre-index knee-attributable inpatient costs
@2010 Informa UK Ltd www.cmrojournal.com
Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al. 1113

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
accounted for the majority of pre-index KAMC among HIS (72%) and LIS (56%) patients, while it represented only a small portion among No Device patients (30%). Similarly, post-index knee-attributable inpatient costs represented a smaller portion of KAMC among HIS patients (44%) compared to LIS (71%) and No Device patients (60%). A similar pattern is seen in Figure 3 when considering the incidence of inpatient hospitalizations during the 6-month period before and after the index knee event. During the 6-month period prior to the index event, 48% of the HIS Device patients had at least one knee-related hospitalization compared to 27% of the LIS Device patients and 9% of the No Device patients (p50.0001). During the 6-month period following the index knee event, the incidence of at least one re-hospitalization was 22% for the HIS Device patients, 29% for the No Device patients and 58% of the LIS Device patients (p50.0001). The results of the GLM model (Table 2) were similar to the unadjusted results. Relative to HIS Device patients, LIS Device patients had a 24% increase (p 1⁄4 0.0248) and No Device patients had a 9% increase (p 1⁄4 0.3232) in post-index KAMC after controlling for covariates. Significantly more (p50.0001) No Device patients (47.4%) had a subsequent knee event within the 6-month period following the index surgery than HIS Device patients (24.5%) and LIS Device patients (22.2%) and the mean time from the index event to the subsequent knee surgery or injury was also shortest for the No Device patients (110.3 81.0 days) relative to the HIS Device patients (146.1 67.7 days) and the LIS Device patients (152.9 61.6 days). A similar pattern was seen when the entire follow-up period was considered (Figure 4). The Cox proportional hazard model (Table 3) evaluat- ing the risk of re-operation showed that, after controlling for covariates, No Device patients were significantly more likely (hazard ratio: 1.71, 95% CI 1.32–2.22, p50.0001) to
1114 Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al.
www.cmrojournal.com @2010 Informa UK Ltd

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
undergo re-operation than HIS patients. Although LIS patients were more likely (hazard ratio: 1.12, 95% CI 0.83–1.50, p 1⁄4 0.4608) to undergo re-operation than HIS patients, the result was not statistically significant. Discussion Arthrofibrosis is a chronic condition resulting in loss of motion and function that can be devastating for patients following knee surgery28. As the disease progresses, a single patient may require extensive orthopaedic follow-up, mul- tiple surgeries, with corresponding prescriptions for phys- ical therapy, NSAIDs, and pain medication. Effective treatment of this condition may not only halt the disease progression and restore motion, but may demonstrate resulting cost savings to the healthcare system. This study represents the largest study of knee arthrofibrosis performed to date and, to the authors’ knowledge, is the first to compare patients treated with or without the addition of mechanical therapy, as well as the first to com- pare the two philosophies of mechanical therapy (LIS versus HIS). As suggested by previous authors, the use of adjunctive mechanical therapy has been indicated for patients who have failed traditional rehabilitation25,28,33. The results of the current study demonstrate that this, indeed, has been how these devices have been used clinically. Patients in the LIS and HIS groups had significantly greater pre-index knee-attributable utilization and medical costs than patients who were treated with physical therapy alone. Specifically, the percentage of patients with at least one knee-attributable inpatient hospitalization during the pre-index period differed among the three groups, with 48.3% of the HIS Device group having at least one pre-index inpatient hospitalization, compared to 27.2% of the LIS Device group, and 8.5% of the No Device group. Not surprisingly, pre-index KAMC followed the same trend, with mean costs for the HIS, LIS, and No Device groups of $11 950 17 767, $9605 27 515, and $2118 8512, respectively. The high knee-attributable inpatient utilization and KAMC of HIS Device patients during the 6-month period prior to their index surgery suggest that these patients were perhaps more complicated patients with more extensive histories of knee surgeries. Because of their more extensive history of either failed primary surgery or a documented history of previously
@2010 Informa UK Ltd www.cmrojournal.com
Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al. 1115

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
failed traditional rehabilitation, mechanical therapy was then indicated for these patients. The unadjusted post-index KAMC data also demon- strated clear differences among the three groups. Mean KAMC for the 6-month post-index period were signifi- cantly lower for the HIS Device group ($8213 10 576) than for the No Device ($9345 14 120) and LIS Device groups ($16 861 17 857). The KAMC of the LIS and No Device groups were higher for the 6-month post-index period than they were for the 6-month pre-index period whereas for the HIS device group the opposite was true. HIS device group patients had higher 6-month pre-index KAMC than for the 6-month post-index period. The source of post-index cost differences among the groups appears to be driven by the incidence of knee- attributable re-hospitalizations. A very high number of the patients in the LIS Device group was re-hospitalized (58.2%), compared to 21.7% of the HIS Device group and 29.3% of the No Device group. Although the percentage of patients with re-hospitalizations for knee-related rea- sons was significantly higher for the LIS Device group, this was not associated with an increase in the percent- age of knee-related re-operations (22.2%). More than one-third of the LIS patients had a knee-attributable inpatient hospitalization that was not associated with a re-operation. Conversely, there was no marked disparity between the percentages of re-hospitalization and re-operation for either the HIS or No Device groups. While the limitations of the retrospective claims data prevent us from confirming the exact source, the authors suggest that the dramatically different rates of re-hospitalization between the two device groups reflect two distinct treatment protocols. One protocol appears to be more dependent on the therapist to apply the neces- sary overpressure to increase joint motion, whereas the other protocol appears to be more dependent on the home mechanical therapy device to apply the necessary overpressure. For the LIS Device group, the disparity between the percentages of re-operation and re-hospitalization may stem from an increased utilization of inpatient rehabilitation or skilled nursing facilities. The purpose of admitting arthrofibrosis patients into these facilities would be to provide them with intensive, moni- tored physical therapy to aggressively treat their condition in addition to the use of the LIS device. In short, in the inpatient/LIS treatment model, the therapist provides the overpressure necessary to increase motion, and the LIS device is used to maintain those motion gains between therapy sessions. Unlike the LIS group, the use of a HIS device may have allowed patients to be treated in an outpatient setting. Where the LIS group may have required daily physical therapy sessions to increase motion, with the LIS mechan- ical therapy device being utilized to maintain motion between physical therapy sessions, the use of the HIS device may have allowed arthrofibrosis patients to be trea- ted with fewer physical therapy sessions. Patients treated in the outpatient setting have fewer physical therapy sessions per week than when in an inpatient setting, often ranging from three to five sessions per week25. Because of the abil- ity to mimic the load applied by the physical therapist35, the HIS device may have been used to increase range of motion between physical therapy sessions instead of acting to maintain motion in the inpatient/LIS treatment model. The ability for patients to be treated in the outpatient setting may be due to the combination of the higher applied loads and the potentially improved compliance associated with shorter daily HIS treatment (60 minutes/ day) compared to the LIS device (6–8 hours/day). Patients who were treated with physical therapy alone (No Device group) were less costly in the pre-index period, suggesting that the degree of arthrofibrosis may not have been as advanced as for the other two groups. These less complicated patients were able to be treated solely with outpatient physical therapy. Patients treated with an HIS device as an adjunct to physical therapy also did not appear to need to be admitted to a rehabilitation or skilled nursing facility, as their motion loss could be treated at home and/ or at an outpatient therapy clinic. After adjusting the post-index KAMC data by control- ling for covariates such as device group, age, sex, total knee replacement surgery, and comorbidity index, the multi- variate predictive model results were consistent with the unadjusted results. The predicted post-index adjusted KAMC of the LIS Device group were 24% (p 1⁄4 0.0248) higher than those of the HIS Device group. Although the No Device group also demonstrated slightly greater pre- dicted post-index adjusted KAMC relative to the HIS Device group (9%, p 1⁄4 0.3232), the increase was not sta- tistically significant; however, No Device patients were 71% (hazard ratio: 1.71, 95% CI 1.32–2.22, p50.0001) more likely to undergo re-operation than the HIS Device patients. This study was not without limitations. There are inher- ent limitations to the observational nature of the retro- spective claims analyses, most notably the potential selection bias that may occur due to lack of randomization between the observed treatment cohorts. The authors attempted to control for this bias through multivariate analyses, which controlled for characteristics such as age, sex, comorbidity index, and surgery that may have differed between the treatment cohorts and confounded the study findings. Specifically, the multivariate analysis demon- strated that the single factor with the greatest effect on post-index KAMC was the presence of total knee arthro- plasty (TKA) as the index event. The three groups differed greatly in the incidence of TKA, with 46.3% of the LIS, 19.0% of the No Device and 11.9% of the HIS groups having this procedure as their index event. The a priori inclusion of TKA, as well as additional patient
1116 Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al.
www.cmrojournal.com @2010 Informa UK Ltd

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
characteristics as model covariates, produced a more con- servative estimate of the 6-month post-index KAMC for each treatment cohort compared to the estimates based on the unadjusted means. Inclusion of the model covariates specified above demonstrated a scaled deviance of 1.16 indicating a reasonable goodness-of-fit for the selected model covariates. When considering the cost differences demonstrated between the LIS and HIS groups, it is appar- ent that the disparity in TKA was the reason that the 49% reduction in KAMC demonstrated by the HIS Device group in the unadjusted analysis was reduced to 24% in the adjusted predictive model. Conclusions In the 6 months following the index event, the unadjusted knee-attributable medical costs of patients in the three treatment groups differed significantly. Patients who used HIS mechanical therapy devices had the lowest unadjusted knee-attributable medical costs compared to the unad- justed costs of the other two groups. Moreover, the use of HIS mechanical therapy was associated with significantly lower rates of post-index re-hospitalization and post-index adjusted knee-attributable medical costs when compared to patients treated with LIS mechanical therapy. Finally, patients who used HIS mechanical therapy devices also had a significantly reduced risk of knee re-injury or re-operation compared to patients treated with physical therapy alone. We conclude that the reduced post- operative knee-attributable medical costs demonstrated by the group of arthrofibrosis patients treated with HIS mechanical therapy could be attributed to the observed reduced incidence of re-hospitalization. Transparency Declaration of funding This study was funded by End Range of Motion Improvement, Inc. (ERMI), Atlanta, Georgia. Declaration of financial/other relationships J.J.S., R.A.Q., and T.G. have disclosed that they are employees of HealthCore, Inc, who received funding for this study. Peer reviewers have received honoraria from CMRO for their review work. Peer Reviewer 1 and Peer reviewer 2 have disclosed that they have no relevant financial relationships. References 1. Kurtz S, Mowat F, Ong K, et al. Prevalence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg Am 2005;87-A:1487-97 2. Lyman S, Koulouvaris P, Sherman S, et al. Epidemiology of anterior cruciate ligament reconstruction, trends, readmissions, and subsequent knee surgery. J Bone Joint Surg Am 2009;91-A:2321-8 3. Kurtz S, Ong K, Lau E, et al. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am 2007;89-A:780-5 4. Kurtz S, Lau E, Ong K, et al. Future young patient demand for primary and revision joint replacement. National projections from 2010 to 2030. Clin Orthop Relat Res 2009;467:2606-12 5. Brophy R, Wright R, Matava M. Cost analysis of converting from single-bundle to double-bundle anterior cruciate ligament reconstruction. Am J Sports Med 2009;37:683-7 6. Ritter M, Harty L, Davis K, et al. Predicting range of motion after total knee arthroplasty. Clustering, log-linear regression, and regression tree analysis. J Bone Joint Surg (Am) 2003;85-A:1278-85 7. Shelbourne K, Patel D. Treatment of limited motion after anterior cruciate ligament reconstruction. Knee Surg Sports Traumatol Arthrosc 1999;7:85-92 8. Noyes F, Berrios-Torres S, Barber-Westin S, et al. Prevention of permanent arthrofibrosis after anterior cruciate ligament reconstruction alone or com- bined with associated procedures: a prospective study in 443 knees. Knee Surg Sports Traumatol Arthrosc 2000;8:196-206 9. Noyes F, Mangine R, Barber S. The early treatment of motion complications after reconstruction of the anterior cruciate ligament. Clin Orthop 1992; 277:217-28 10. Boldt J, Munzinger U, Zanetti M, et al. Arthrofibrosis associated with total knee arthroplasty: Gray-scale and power Doppler sonographic findings. Am J Roentgenol 2004;182:337-40 11. Hasan S, Saleem A, Bach Jr B, et al. Results of arthroscopic treatment of symptomatic loss of extension following anterior cruciate ligament recon- struction. Am J Knee Surg 2000;13:201-9 12. Kaper B, Smith P, Bourne R, et al. Medium-term results of a mobile bearing total knee replacement. Clin Orthop 1999;367:201-9 13. Harner C, Irrgang J, Paul J, et al. Loss of motion after anterior cruciate ligament reconstruction. Am J Sports Med 1992;20:499-506 14. Maloney W. The stiff total knee arthroplasty. Evaluation and management. J Arthroplasty 2002;17:71-3 15. Shapiro M, Freedman E. Allograft reconstruction of the anterior and posterior cruciate ligaments after traumatic knee dislocation. Am J Sports Med 1995; 23:580-7 16. Tzurbakis M, Diamantopoulos A, Xenakis T, et al. Surgical treatment of mul- tiple knee ligament injuries in 44 patients: 2-8 years follow-up results. Knee Surg Sports Traumatol Arthrosc 2006;14:739-49 17. Petersen W, Laprell H. Combined injuries of the medial collateral ligament and the anterior cruciate ligament. Early ACL reconstruction versus late ACL reconstruction. Arch Orthop Trauma Surg 1999;119:258-62 18. Tscherne H, Lobenhoffer P. Tibial plateau fractures. Management and expected results. Clin Orthop 1993;292:87-100 19. Huang H-T, Huang P-J, Su J-Y, et al. Indirect reduction and bridge plating of supracondylar fractures of the femur. Injury, Int J Care Injured 2003; 34:135-40 20. Chapman M, Finkemeier C. Treatment of supracondylar nonunions of the femur with plate fixation and bone graft. J Bone Joint Surg Am 1999; 81-A:1217-28 21. Unterhauser F, Bosch U, Zeichen J, et al. Alpha-smooth muscle actin con- taining contractile fibroblastic cells in human knee arthrofibrosis tissue. Arch Orthop Trauma Surg 2004;124:585-91 22. Murakami S, Muneta T, Furuya K, et al. Immunohistologic analysis of syno- vium in infrapatellar fat pad after anterior cruciate ligament injury. Am J Sports Med 1995;23:763-8 23. Sprague NI, O’Connor R, Fox J. Arthroscopic treatment of postoperative knee fibroarthrosis. Clin Orthop 1982;166:165-72 24. Mont M, Seyler T, Marulanda G, et al. Surgical treatment and customized rehabilitation for stiff knee arthroplasties. Clin Orthop 2006;446:193-200 25. Seyler T, Marker D, Bhave A, et al. Functional problems and arthrofibrosis following total knee arthroplasty. J Bone Joint Surg Am 2007;89-A:59-69 26. Rowe P, Myles C, Walker C, et al. Knee joint kinematics in gait and other functional activities using flexible electrogoniometry: how much knee motion is sufficient for normal daily life? Gait Posture 2000;12:143-55
@2010 Informa UK Ltd www.cmrojournal.com
Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al. 1117

Current Medical Research & Opinion Volume 26, Number 5 May 2010
CMRO
27. Ritter M, Lutgring J, Davis K, et al. The effect of postoperative range of motion on functional activities after posterior cruciate-retaining total knee arthro- plasty. J Bone Joint Surg Am 2008;90-A:777-84 28. Magit D, Wolff A, Sutton K, et al. Arthrofibrosis of the knee. J Am Acad Orthop Surg 2007;15:682-94 29. Christensen C, Crawford J, Olin M, et al. Revision of the stiff total knee arthroplasty. J Arthroplasty 2002;17:409-15 30. Haidukewych G, Jacofsky D, Pagnano M, et al. Functional results after revi- sion of well-fixed components for stiffness after primary total knee arthro- plasty. J Arthroplasty 2005;20:133-8 31. Aglietti P, Buzzi R, De Felice R, et al. Results of surgical treatment of arthrofi- brosis after ACL reconstruction. Knee Surg Sports Traumatol Arthrosc 1995; 3:83-8 32. O’Shea J, Shelbourne K. Anterior cruciate ligament reconstruction with a reharvested bone-patellar tendon-bone graft. Am J Sports Med 2002; 30:208-13 33. Branch T, Karsch R, Mills T, et al. Mechanical therapy for loss of knee flexion. Am J Orthop 2003;32:195-200 34. McClure P, Blackburn L, Dusold C. The use of splints in the treatment of joint stiffness: biologic rationale and an algorithm for making clinical decisions. Phys Ther 1994;74:1101-7 35. Uhl T, Jacobs C. Torque measures of common therapies for the treatment of flexion contractures. J Arthroplasty 2010;In Print 36. Steffen T, Mollinger L. Low-load, prolonged stretch in the treatment of knee flexion contractures in nursing home residents. Phys Ther 1995;75:886-97 37. Bosch U, Zeichen J, Skutek M, et al. Arthrofibrosis is the result of a T cell mediated immune response. Knee Surg Sports Traumatol Arthrosc 2001; 9:282-9 38. Deyo R, Cherkin D, Ciol M. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613-19 39. Manning WG, Basu A, Mullahy J. Generalized modeling approaches to risk adjustment of skewed outcomes data. J Health Econ 2005;24:465-88
1110 Costs of non-surgical treatment options for knee arthrofibrosis Stephenson et al.
www.cmrojournal.com @2018 Informa UK Ltd