From 1999 to 2008, the utilization rate of total knee replacement procedures in the United States more than doubled for the overall population, and tripled for individuals aged 45 to 64 years.1 Additional research has shown an increase in the rate of primary and revision knee replacement surgeries, as well as hip replacements.1,2 These studies indicate a growing trend for total knee and total hip replacement procedures among the US population aged 45 to 64 years and those aged ≥65 years, with the trend being more pronounced for those who underwent total knee replacement.
Although studies have shown that knee and hip replacements generally result in improvement in pain and functionality for patients who undergo these procedures,3,4 the dramatic increase in number and high cost of such surgeries has garnered much attention nationally. As demonstrated by Losina and colleagues, increases in population size and in obesity rates can only partially explain this increase.1 Furthermore, their study begs the question of whether all joint replacement procedures are clinically necessary, and whether alternative conservative treatments have been explored before joint replacement is considered.
Two earlier studies focused on viscosupplementation as a viable alternative to knee and hip replacements from a managed care perspective. Arnold and colleagues suggested “its use…may generate savings in hospitalizations and other costs,”5 whereas Waddell and Bricker suggested that in candidates for total knee replacement, the need for the procedure can be delayed with hylan G-F 20 (a type of viscosupplementation) when used for the treatment of osteoarthritis (OA) knee pain.6 Until recently, the American Academy of Orthopaedic Surgeons (AAOS) was inconclusive regarding viscosupplementation as a treatment modality for patients with OA of the knee. However, in May 2013, the AAOS recommended against viscosupplementation based on the lack of evidence showing benefit.7
Given the dramatic increase in the rate of knee and hip replacement procedures, and the controversy surrounding the effectiveness of viscosupplementation, there is a need to compare the downstream healthcare utilization and costs between patients who undergo a knee or hip replacement and those who receive intraarticular injections to relieve pain and improve functionality. To examine differences between these 2 groups, lower OA-related healthcare utilization and costs were assumed to be reasonable proxies of better outcomes, consistent with lesser need for ongoing interaction with the healthcare provider.
The aim of this longitudinal study—which focused on Medicare members who underwent joint replacement or received intraarticular injections for pain—was to determine which approach resulted in fewer OA-related physician office visits and OA-related treatments, as well as lower OA-related healthcare costs, after recovery.
The data were obtained from the Humana Research Database (Louisville, KY), which contains enrollment, medical, and pharmacy claims information for Humana’s Medicare Advantage with Prescription Drug plan membership. All data sources were merged using deidentified member identification. The finalized protocol was approved by an independent Institutional Review Board.
This retrospective longitudinal cohort study included Medicare members diagnosed with OA who underwent knee or hip replacement and those who had no joint replacement but received 1 or more intraarticular glucocorticoid or viscosupplementation injections into a major joint. Specifically, members with OA who underwent primary knee or hip replacement surgery between July 1, 2007, and June 30, 2012, were identified for this study. For this cohort, the date of joint replacement was considered the index date. The index date for the comparison cohort that received pain-relief injections was 180 days after the first injection.
Difference-in-difference analyses were conducted to compare the changes in OA-related healthcare costs, postsurgery versus presurgery, for those who received knee or hip replacement relative to those who received pain-relief injections. The difference-in-difference models included variables for depression, anxiety, fibromyalgia, and coexisting conditions causing chronic pain. Chronic pain conditions were included because patients who have coexisting conditions of pain with central sensitization may experience persistent pain after surgery8,9 and, consequently, may incur higher healthcare utilization and cost. A separate analysis of secondary endpoints (postsurgical readmissions, infections related to the joint prosthesis, revision surgeries, and venous thromboembolic events [VTEs]) was conducted for those who underwent knee or hip replacement.
The study population comprised Medicare Advantage with Prescription Drug members aged 45 years or older. (Members younger than 65 years qualify for Medicare by way of a disability, diagnosis of end-stage renal disease, or diagnosis of amyotrophic lateral sclerosis.10) Primary knee or hip replacement procedures (partial or total) were identified by the presence of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code or Current Procedural Terminology (CPT) code in any diagnosis position (see Appendix at the end of the article). Intraarticular glucocorticoid or viscosupplementation injections were identified by associated CPT or J-codes (Appendix). At least 1 diagnosis of OA (ICD-9-CM 715.xx) in any position on a claim was required during the preindex period. Although the procedural codes for intraarticular glucocorticoid or viscosupplementation injections were not documented as being specific to the knee or the hip, OA is generally more symptomatic in weight-bearing joints.11
Members were required to have continuous enrollment 180 days preindex and through 360 days postindex. The preindex period reflected the health plan’s policy for viscosupplementation injections (ie, limit of 2 or 3 within 6 months). The postindex period was based on previous literature indicating the ability to capture signs of improvement by 3 to 12 months after surgery.12,13
Members were excluded if they had undergone a primary knee or hip replacement procedure during the preindex period, a second primary knee or hip replacement surgery during follow-up, or a hip fracture (ICD-9-CM 733.14, 733.96, 820.xx) or visit to the emergency department up to 3 days before or after the joint replacement procedure, indicative of nonelective surgery. In addition, members were excluded if they had other conditions, including pregnancy (ICD-9-CM 630.xx-679.xx, V22.xx, or V23.xx), cancer (ICD-9-CM 140.xx-172.xx, 174.xx-208.xx), organ transplantation (V42.xx), Alzheimer’s disease or dementia (ICD-9-CM 331.0 or 290.0, 290.1x, 290.2x, 290.3, 290.4x, 290.8, 290.9, 291.2, 292.82, 294.1x, 294.8, 331.1x, 331.7, 331.82, 331.9), or a coagulation defect (ICD-9-CM 286.xx). Members in skilled nursing facilities for >90 days of the study period were excluded because of incomplete capture of healthcare utilization and costs in claims data.
Baseline demographics and clinical characteristics were compared between cohorts. For all statistical comparisons, t-tests were used to compare continuous variables and chi-square tests for categorical variables. OArelated healthcare utilization was captured for each cohort in 90-day intervals. Difference-in-difference analyses were performed to determine whether the change in healthcare costs for the knee or hip replacement cohort was larger than for the injection cohort, for the 180 days postindex versus the 180 days preindex. The models controlled for age, sex, ethnicity, geographic region, plan type, the Deyo-Charlson comorbidity index (DCI) score, RxRisk-V comorbidity index score, depression, anxiety, fibromyalgia, and chronic pain conditions. The RxRisk-V comorbidity index score is derived from drug claims data and thus can be applied to data from a narrow window of claims as opposed to the broader window typically needed for medical claims–based comorbidity scores.14-19
Because healthcare expenditure data are known to show nonnormal distribution, often with nonconstant variance, the difference-in-difference generalized linear model with log link and gamma distribution was fitted to member healthcare costs. The generalized linear model is an extension of linear regression models but can handle data that have nonnormal distribution and nonconstant variance. The generalized estimation equation method was used to account for repeated measures of the same member.20
Time-to-event analyses were performed to measure secondary outcomes, including hospital readmission rates related to the knee or hip; infection rates related to the joint prosthesis (ICD-9-CM 996.66); incidence of revision surgery (Appendix); and rates of VTE (ICD-9-CM 453.xx). All data analyses were conducted using SAS Enterprise Guide 5.1 (SAS Institute, Cary, NC). Statistical significance was defined as P <.05.
Complete data were available for 25,181 members with OA who underwent knee (N = 17,864; 71%) or hip (N = 7317; 29%) replacement and for 64,484 members with OA who received a steroid or viscosupplementation injection (Figure 1). The mean age was 70.7 years for the knee replacement cohort, 71.7 years for the hip replacement cohort, and 71.1 years for the injection cohort (P <.01 for each joint replacement group vs the injection group; Table 1). The majority of members in each cohort were female, white, and residing in the South. A slightly higher percentage of members in the injection cohort qualified for a low-income subsidy, were dual eligible (ie, qualified for Medicare and Medicaid), or met the criteria for a low-income subsidy and dual eligibility (P <.01 for all comparisons).
The mean RxRisk-V comorbidity index score was 4.7 for the knee replacement cohort, 4.4 for the hip replacement cohort, and 4.8 for the injection cohort (P <.01 for each joint replacement cohort vs the injection cohort). The mean DCI score was 0.7 for those with knee replacement, 0.6 for those with hip replacement, and 0.7 for the injection group (P <.01 only for the hip vs injection cohort). The proportions of members with specific comorbidities were remarkably similar, with few exceptions, including diabetes (21.4% hip vs 29.3% injection; P <.01); low back pain (13.0% knee, 24.7% hip, 18.9% injection; P <.01 for hip vs injection and knee vs injection); and osteoporosis (15.1% knee, 23.5% hip, 17.3% injection; P <.01 for hip vs injection and knee vs injection). Healthcare costs in the preindex period were highest for the injection cohort—$717.16 versus $609.99 and $522.61 for those with knee or hip replacement, respectively (P <.01 for each comparison); however, the median cost values were similar ($364, $333, and $340, respectively) among the cohorts (Table 1).
The data for OA-related healthcare utilization are shown in Table 2. Within the knee replacement cohort, healthcare resource utilization was high during the 90 days preindex for emergency department visits (50.7%), outpatient visits (82.3%), office-based physical therapy (88.0%), and use of oral pain medication (60.4%). Similar proportions of the hip replacement cohort had emergency department visits (52.1%), outpatient visits (78.6%), office-based physical therapy (84.9%), and use of oral pain medication (64.5%) during the 90 days preindex. As expected, inpatient healthcare utilization spiked at the time of joint replacement. Afterward, there was a substantial decline over time in most utilization categories, to levels lower than those observed preindex, except for the use of oral pain medication (Table 2).
Within the injection cohort, healthcare resource utilization was high from 91 to 180 days preindex (when the pain-relief injection was administered) for emergency department visits (32.9%), outpatient visits (61.3%), office-based physical therapy (65.3%), pain-relief injections (69.2%), and use of oral pain medication (62.4%). Healthcare resource utilization declined over time during the remaining study periods. Anticoagulant use by those who underwent joint replacement doubled from 91 to 180 days preindex to the 90 days preceding surgery (Table 2). During the first 90 days postsurgery, anticoagulant medication use increased to 55.9% and 56.6% for the knee and hip replacement cohorts, respectively. Anticoagulant medication use was flat (at approximately 10%) among the injection group during the entire study period, and was similar to that of the joint replacement cohorts during the 91 to 180 days preindex and 271 to 360 days postindex.
Figure 2 shows the mean OA-related healthcare costs in 90-day increments. As expected, there was a spike at the time of joint replacement for the surgery cohorts, but these costs dropped to below preindex levels by 360 days postindex. For the injection cohort, the costs were highest from 91 to 180 days preindex, during which time the glucocorticoid or viscosupplementation injection was administered, and then decreased throughout follow-up (Figure 2).
Difference-in-difference analyses were used to compare data for the 181- to 360-day postindex period and the 91- to 180-day preindex period (Table 3). For both replacement cohorts, the OA-related costs were lower than for the injection cohort: coefficient of –0.165 for knee replacement versus injection (P <.001); coefficient of –0.276 for hip replacement versus injection (P <.001) (Table 3). Over time, the OA-related costs for the joint replacement cohorts decreased more than that for the injection cohort (–1.233 and –1.236 for the knee and hip replacement cohorts, respectively; P <.001 vs injection cohort for both comparisons). The interaction term was –0.603 for knee replacement*time and –0.438 for hip replacement*time (P <.001 vs injection cohort for both comparisons). The exponentiated coefficients indicate that costs were 45.3% lower for the knee replacement cohort and 35.5% lower for the hip replacement cohort compared with the injection cohort (Table 3).
For the knee replacement model, parameter estimates for age, race, and geographic region were significant. Each additional year of age was associated with a 1% lower OA-related healthcare cost (P <.001). Costs for white members were 19.8% higher than for nonwhites (P = .005), and costs for members in the Midwest were 21.6% higher than for those in the South (P <.001). Each incremental point in the RxRisk-V comorbidity index was associated with a 6.8% increase in costs (P <.001). Members in the subgroup diagnosed with depression had 18.2% higher costs than those not diagnosed with depression (P = .05).
For the hip replacement model, the costs were 17.1% lower for members in the South compared with those in the Midwest (P = .005). Each incremental point in the RxRisk-V comorbidity index was associated with an 8.3% increase in costs. Members diagnosed with chronic obstructive pulmonary disease (COPD) had 19.3% lower OA-related costs than those not diagnosed with COPD (P = .002).
Incidence and time-to-event analyses are reported in Table 4 for specific outcomes after joint replacement. The rate of OA-related hospital readmissions was 0.3% for both joint replacement cohorts. Rates of infection related to joint prosthesis were 1.4% for those with knee replacement and 1.5% for those with hip replacement. The rates of revision surgery were 3.4% and 4.5%, respectively, and the corresponding VTE rates were 5.6% and 5.1%. Results of time-to-event analysis were similar for these cohorts, except for time to revision surgery: the mean (median) number of days to revision surgery was 133.6 (119) for the knee replacement group and 105.6 (44) for the hip replacement group (Table 4).
Although previous studies have shown that knee or hip replacement surgery results in an improvement in pain and functionality,3,21 to our knowledge the present study is the first to explicitly examine a comparison cohort of health plan members who received steroid or viscosupplementation injections. Our findings indicate that members without significant comorbid conditions who underwent knee or hip replacement procedure had a greater decrease in OA-related healthcare resource utilization and costs after they recovered from surgery, compared with presurgery, and compared with the members who received intraarticular injections. These results also suggest that, although initially generating lower cost, treatment with steroid and viscosupplementation injection may result in increased utilization and cost over time.
These results are consistent with the AAOS’ recent change in evidence-based guidelines for viscosupplementation for symptomatic OA pain of the knee, from “inconclusive” to “recommend against.”7 A meta-analysis based on a review of 14 studies that assessed outcomes of intraarticular hyaluronic injections demonstrated that the overall effect was not clinically meaningful.7 The group concluded that, because of the clear evidence of lack of efficacy, it could not recommend viscosupplementation.
Our findings also are consistent with a systematic review by Rutjes and colleagues, indicating that viscosupplementation was associated with a small but clinically irrelevant benefit.22 The AAOS guidelines regarding the use of intraarticular corticosteroids were inconclusive (ie, unable to recommend for or against use).7
With respect to the impact of knee or hip replacement on healthcare costs, our findings may contradict those of a study by Bozic and colleagues, who found that the costs for patients with knee replacement increased after the surgery.23 However, their identification periods and populations differed from ours, and their study focused on total healthcare costs rather than OA-related costs. Moreover, their study did not include a comparison cohort to control for changes in external factors that may have affected costs over time. The inclusion of a comparison cohort, as in our study, is an important enhancement to the methodology of Bozic and colleagues.23
Patients diagnosed with depression who undergo knee replacement may have specific treatment needs that are separate from the need for joint replacement, which may lead to an increase in healthcare costs. Symptoms of depression may manifest as joint pain, and hence these patients would continue to experience joint pain after knee replacement surgery. Our findings are consistent with those of Wylde and colleagues, indicating that depression, among other comorbid conditions involving chronic pain, may be a risk factor for poor outcomes after surgery.9
The comorbid conditions of depression, anxiety, fibromyalgia, and chronic pain noted in the current study have been reported to increase the risk for persistent pain,8,9,24 and thus could have resulted in increased costs despite a surgery performed primarily to relieve pain. Patients with comorbid conditions were identified by a recorded diagnosis code on a medical claim; if such conditions are underreported in normal medical practice, then they would be underrepresented in the analysis. Underrepresentation resulting from a lack of medical records for patients in these risk groups may lead to a reduced statistical power to show a significant effect. A prospective study would be more appropriate for thoroughly investigating the potential effect of these comorbid conditions on costs after total joint replacement.
Patients who have been identified as having comorbid conditions and/or other psychosocial risk factors may require concomitant treatment of the comorbid conditions before joint replacement, or further evaluation to assess whether the joint pain likely would be relieved by the surgery. Comorbid conditions are common among older Medicare beneficiaries, as well as 45- to 64-year-old patients, who may have comorbidities that qualify them for Medicare coverage.
Treatment of the comorbid conditions and further evaluation will help ensure that appropriate candidates are selected for knee or hip replacement, will improve the probability of successful outcomes, and will increase the overall cost-effectiveness of these procedures. However, even if candidates are selected appropriately, gaps in appropriate treatment remain, as evidenced by the rates of VTEs in our study, which were 5.6% for knee replacement and 5.1% for hip replacement. These rates are much higher than those reported by Cram and colleagues (0.3%-0.4%) in a recent study,25 and much higher than our preindex period rates (1.4% for knee and 1.3% for hip; Table 1).
Although the proportion of members treated with anticoagulants in the immediate postsurgery period was markedly higher than during the preindex period, almost half the members who underwent surgery did not receive treatment in the immediate postindex period, when the risk for VTEs is highest. This finding conflicts with current guidelines stating that prophylactic anticoagulant therapy is recommended to mitigate the increased risk for VTE shortly after joint replacement.26-28
The limitations of this study include the possibility that the cohort receiving pain relief injections might have received them into a joint other than the knee or hip, since the medical claim specified “OA” and “joint” but not a particular joint. Although the knee and hip are common sites for OA, other joints, such as the shoulder or ankle, could have been the site of injection.
Another limitation was the possibility that members of the injection cohort may have been sicker and of lower socioeconomic status than those who had joint replacement, as indicated by differences in demographic and clinical characteristics. Although these differences were small, they may have precluded some members of the comparison cohort from being considered for joint surgery. Even though the difference-in-difference models were adjusted for these differences, it is possible that not all characteristics were adequately measured or controlled for in the adjusted analyses.
If OA was not documented on the claim, postsurgery healthcare utilization and costs might have been underestimated as a result of not including the cost of thromboembolic events, nosocomial infections, or other potential consequences of the surgery and inpatient stays that are known to be more common among patients who undergo surgery.25
This study focused solely on members enrolled in the Medicare Advantage with Prescription Drug plan at Humana. Therefore, the findings may not be generalizable to members enrolled in commercial health plans. Future research should include a broader population to improve generalizability. Examining healthcare utilization and costs from 180 days preindex through 360 days postindex is conducive to the difference-in-difference analysis used in this study and is appropriate for capturing any chronic or persistent postsurgical pain.29
Expanding the study period would have captured patients with pain-relief injections who received a joint replacement beyond 360 days postindex, but it would have increased the chance of including healthcare utilization and costs of new, unrelated, major clinical events. Restricting the study to members with continuous enrollment might have led to an underestimation of the number of readmissions, infections, and thromboembolic events resulting in death.
The results of this study suggest that patients with OA who undergo knee or hip replacement surgery have larger decreases in healthcare utilization and costs than those who receive steroid or viscosupplementation injections. However, the high rates of VTE postsurgery emphasize the need to increase prophylactic therapy with anticoagulant medications, as appropriate.
We thank Mary Costantino, PhD, for her review and assistance with editing the original manuscript.
This study was funded by Humana and by Pfizer.
Author Disclosure Statement
Dr Pasquale is an employee of Humana. Dr Louder is an employee of and owns stocks in Humana, and owns stocks in Pfizer and in Anthem. Dr Cheung is an employee of and owns stocks in Pfizer. Dr Reiners is an employee of Humana. Dr Mardekian is an employee of and owns stocks in Pfizer. Dr Sanchez was an employee of Pfizer at the time of study, and is now Senior Director at Regeneron. Dr Goli is an employee of and owns stocks in Pfizer.
1. Losina E, Thornhill TS, Rome BN, et al. The dramatic increase in total knee replacement utilization rates in the United States cannot be fully explained by growth in population size and the obesity epidemic. J Bone Joint Surg Am. 2012;94:201-207.
2. 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;87:1487-1497.
3. Kane RL, Saleh KJ, Wilt TJ, et al. Total knee replacement. Evidence Report/Technology Assessment No. 86. AHRQ Publication No. 04-E006-2. Rockville, MD: Agency for Healthcare Research and Quality; December 2003. http://archive.ahrq.gov/downloads/pub/evidence/pdf/knee/knee.pdf. Accessed May 29, 2014.
4. Räsänen P, Paavolainen P, Sintonen H, et al. Effectiveness of hip or knee replacement surgery in terms of quality-adjusted life years and costs. Acta Orthop. 2007;78:108-115.
5. Arnold W, Fullerton DS, Holder S, May CS. Viscosupplementation: managed care issues for osteoarthritis of the knee. J Manag Care Pharm. 2007;13(4 suppl):S3-S19; quiz S20-S22.
6. Waddell DD, Bricker DC. Total knee replacement delayed with Hylan G-F 20 use in patients with grade IV osteoarthritis. J Manag Care Pharm. 2007;13:113-121.
7. American Academy of Orthopaedic Surgeons. Treatment of Osteoarthritis of the Knee: Evidence-Based Guideline. 2nd ed. Rosemont, IL: American Academy of Orthopaedic Surgeons; 2013. www.aaos.org/Research/guidelines/TreatmentofOsteoarthritisoftheKneeGuideline.pdf. Accessed May 29, 2014.
8. Masselin-Dubois A, Attal N, Fletcher D, et al. Are psychological predictors of chronic postsurgical pain dependent on the surgical model? A comparison of total knee arthroplasty and breast surgery for cancer. J Pain. 2013;14:854-864.
9. Wylde V, Hewlett S, Learmonth ID, Dieppe P. Persistent pain after joint replacement: prevalence, sensory qualities, and postoperative determinants. Pain. 2011;152:566-572.
10. Henry J. Kaiser Family Foundation. Medicare: Medicare and nonelderly people with disabilities. Fact sheet. September 2010. https://kaiserfamilyfoundation.files.wordpress.com/2013/01/8100.pdf. Accessed May 1, 2015.
11. Centers for Disease Control and Prevention. Osteoarthritis. Updated May 16, 2014. www.cdc.gov/arthritis/basics/osteoarthritis.htm. Accessed December 10, 2014.
12. Davis AM, Perruccio AV, Ibrahim S, et al. The trajectory of recovery and the interrelationships of symptoms, activity and participation in the first year following total hip and knee replacement. Osteoarthritis Cartilage. 2011;19:1413-1421.
13. Browne JP, Bastaki H, Dawson J. What is the optimal time point to assess patient-reported recovery after hip and knee replacement? A systematic review and analysis of routinely reported outcome data from the English patientreported outcome measures programme. Health Qual Life Outcomes. 2013;11:128.
14. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613-619.
15. Farley JF, Harley CR, Devine JW. A comparison of comorbidity measurements to predict healthcare expenditures. Am J Manag Care. 2006;12:110-117.
16. Fishman PA, Goodman MJ, Hornbrook MC, et al. Risk adjustment using automated ambulatory pharmacy data: the RxRisk model. Med Care. 2003;41:84-99.
17. Sales AE, Liu C-F, Sloan KL, et al. Predicting costs of care using a pharmacy-based measure risk adjustment in a veteran population. Med Care. 2003;41:753-760.
18. Sloan KL, Sales AE, Liu C-F, et al. Construction and characteristics of the RxRisk-V: a VA-adapted pharmacy-based case-mix instrument. Med Care. 2003;41:761-774.
19. Von Korff M, Wagner EH, Saunders K. A chronic disease score from automated pharmacy data. J Clin Epidemiol. 1992;45:197-203.
20. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22.
21. Losina E, Walensky RP, Kessler CL, et al. Cost-effectiveness of total knee arthroplasty in the United States: patient risk and hospital volume. Arch Intern Med. 2009;169:1113-1121; discussion 1121-1122.
22. Rutjes AWS, Jüni P, da Costa BR, et al. Viscosupplementation for osteoarthritis of the knee: a systematic review and meta-analysis. Ann Intern Med. 2012;157:180-191.
23. Bozic KJ, Stacey B, Berger A, et al. Resource utilization and costs before and after total joint arthroplasty. BMC Health Serv Res. 2012;12:73.
24. Berger A, Dukes E, Martin S, et al. Characteristics and healthcare costs of patients with fibromyalgia syndrome. Int J Clin Pract. 2007;61:1498-1508.
25. Cram P, Lu X, Kates SL, et al. Total knee arthroplasty volume, utilization, and outcomes among Medicare beneficiaries, 1991-2010. JAMA. 2012;308:1227-1236.
26. Falck-Ytter Y, Francis CW, Johanson NA, et al; for the American College of Chest Physicians. Prevention of VTE in orthopedic surgery patients: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(suppl):e278S-e325S.
27. Geerts WH, Bergqvist D, Pineo GF, et al; for the American College of Chest Physicians. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest. 2008;133(suppl):381S-453S.
28. Sheth NP, Lieberman JR, Della Valle CJ. DVT prophylaxis in total joint reconstruction. Orthop Clin North Am. 2010;41:273-280.
29. Kehlet H, Rathmell JP. Persistent postsurgical pain: the path forward through better design of clinical studies. Anesthesiology. 2010;112:514-515.