|Home | About | Journals | Submit | Contact Us | Français|
Total hip and knee replacement (THR and TKR) are high risk settings for venous thromboembolism (VTE).
(1) Summarize cost-effectiveness of VTE prophylaxis regimens for THR and TKR a
Medline (from January 1997 to October 2009), EMBASE (January 1997 until June 2009), and the Economic Evaluation Database (1997- October 2009)
We identified recent cost-effectiveness studies examining five categories of comparisons: (1) anticoagulants (warfarin, low molecular weight heparin - LMWH, or fondaparinux) vs. aspirin; (2) LMWH vs. warfarin; (3) fondaparinux vs. LMWH; (4) comparisons with new oral anticoagulants; and (5) extended (≥3 weeks) vs. short duration prophylaxis (< 3 weeks). We abstracted information on cost and effectiveness for each prophylaxis regimen in order to calculate an incremental cost-effectiveness ratio. Because of variations in effectiveness units reported and horizon length analyzed, we calculate two cost-effectiveness ratios, one for the number of symptomatic, proximal VTE events avoided at 90 days and the other for quality adjusted life-years (QALYs) at the one year mark or beyond.
We identified 33 studies with 67 comparisons. After standardization, comparisons between LMWH and warfarin were inconclusive whereas fondaparinux dominated LMWH in nearly every comparison. The latter results were derived from radiographic VTE rates. Extended duration prophylaxis after THR was generally cost-effective. Small numbers prohibit conclusions about aspirin, new oral anticoagulants, or extended duration prophylaxis after TKR.
Fondaparinux after both THR and TKR and Extended duration LMWH after THR appear to be cost cost-effective prophylaxis regimens. Small numbers for other comparisons and absence of trials reporting symptomatic endpoints prohibit comprehensive conclusions.
In 2005, there were 580,000 total hip or knee replacements (THR or TKR) performed in the U.S and that number is projected to increase to 4.5 million by 2030. Although THR and TKR are generally safe procedures, they have been identified as high-risk events for venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary embolism (PE). For almost 20 years, physicians have been offering pharmacological prophylaxis to patients undergoing THR and TKR. Uncertainty exists, however, about the optimal pharmacological regimen for prophylaxis.
Guidelines published by the American College of Chest Physicians (ACCP) in 2008 support using potent anticoagulant regimens with agents such as fondaparinux, low molecular weight heparin (LMWH), and warfarin (target INR 2-3) and discourage aspirin therapy alone. Guidelines by the American Academy of Orthopedic Surgeons (AAOS), in contrast, support use of aspirin or a lower potency warfarin regimen (INR < 2) in addition to LMWH and fondaparinux, stating that the latter agents do not offer increased protection against PE but substantially raise the rate of bleeding complications. New oral anticoagulants such as rivaroxaban, a factor Xa inhibitor, and dabigatran, a direct thrombin inhibitor, are expected to gain FDA approval within the next several months and it is anticipated that they too will be supported by the above professional societies.
Several studies have attempted to address these risk-benefit and cost issues using decision analysis methodology regarding specific strategies implementing VTE prophylaxis. Some studies[7-9] substantiate the cost-effectiveness of newer regimens more potent in terms of preventing VTE while others do not. Individual study results vary depending on the setting, economic perspective (e.g., groups for which cost and effects will be aggregated – patients, payers, or others), horizon (time course over which cost and effectiveness information was assessed), and effectiveness outcome analyzed (e.g., VTE events averted, life-years gained, quality adjusted life years (QALYs) gained). Measuring effectiveness in QALYs, particular over a horizon of one year or greater, permits comparison of cost-effectiveness of interventions across diseases but some authors may choose not to measure QALYs because their focus resides in the economics related to the period immediately following surgery. To more meaningfully compare VTE prophylaxis regimens, we systematically reviewed recently published studies that evaluated the cost-effectiveness of the different pharmacologic options in patients undergoing THR and TKR. We abstracted information about cost and effects both for a short and long horizon. In each case, we calculated the incremental cost-effectiveness ratios using our abstractions and then converted our estimates based in many currencies into 2009 US dollars (USD).
Using published recommendations for identification of cost-effectiveness studies, we searched Medline (from January 1997 to October 2009), EMBASE (January 1997 until June 2009), and the Economic Evaluation Database (1997- October 2009). (Appendix 1) We also searched the bibliographies of included studies.
We included studies that evaluated the cost-effectiveness of pharmacological agents in patients undergoing THR or TKR. Specifically, we focused our search on recent (1997- October 2009) studies published in English that contained complete documentation of methods (as compared with abstracts or brief reports), had discrete information available for TKR or THR (i.e., not combined with other orthopedic surgeries), and contained enough information to calculate an incremental cost-effectiveness ratio for at least one of five important comparisons. The five comparisons were: (1) anticoagulants (fondaparinux, LMWH, warfarin) vs. aspirin; (2) LMWH vs. warfarin; (3) fondaparinux vs. LMWH; (4) comparisons with new oral anticoagulants and (5) extended duration prophylaxis (3 weeks or more with any agent) vs. short duration prophylaxis (< 3 weeks with any agent). We did not analyze information about regimens not routinely recommended as sole therapy by the ACCP or AAOS. These include unfractionated heparin, parenteral thrombin inhibitors, or non-pharmacological means such as intermittent pneumatic compression or graduated stockings. Two authors (A.K. and N.R.) evaluated each study for inclusion. Disagreements were resolved by discussion.
We derived an abstraction instrument based on the recommendations of the Panel on Cost-Effectiveness[13-15]. Two abstractors (N.R. and W.C.) assisted the primary author (A.K.) in recording, in duplicate, the description of the study setting, cohort age, economic perspective, and presence of pharmaceutical industry sponsorship.
To summarize the cost-effectiveness information of our 5 main comparisons, we abstracted data on the incremental cost and effectiveness for both a short and longer horizon when available. The horizon represents the period of time over which costs and effectiveness are aggregated. For certain diseases such as the common cold, a short-horizon analysis may suffice. In other cases, long-term consequences must be accounted for, even for short-term interventions.[13-15] For the short horizon, we abstracted data on the projected costs incurred and VTE events avoided for the period closest to 90 days from surgery. For the purpose of calculating effectiveness, we abstracted data on the combined incidence of deep venous thrombosis (DVT) and pulmonary embolism (PE) that would be detected in routine clinical practice. If a study did not report such an outcome, we also accepted the incidence of radiographically detected events and noted the distinction. If effectiveness was defined only by the life-years or QALYs, we recorded that information.
For the long horizon, we accepted any information that projected the cost and effectiveness for one year or more. We abstracted effectiveness information preferentially for the outcome of QALYs or unadjusted life-years.
For each study with missing information about drug regimen, dosage, duration of therapy, horizon of analysis, major bleeding rate, DVT, PE, and death rate, we contacted corresponding authors first by email and then by letter. If the authors did not respond, we recorded the information as not specified.
We adjusted all cost information to 2009 USD by inflating or deflating to the year 2005 according to readily available consumer price indices for each country [16, 17], converted to USD via World Health Organization purchasing power parity indices , and then inflated to USD using the Bureau of Labor Statistics consumer price calculator available at www.bls.gov . This approach followed the example of Bachmann et al.
To assess study quality, we created an instrument adapted from “Drummond’s List” and one other instrument from Brauer et al. These included items about the use of cost data from a randomized controlled trial or other primary source, use of efficacy data from pooled results of a systematic review, identification of credible sources for all input parameters, appropriate calculation of an incremental cost-effectiveness ratio (ICER), and use of comprehensive one-way sensitivity analyses. The ICER is an expression of how much additionally it costs (in dollars) to achieve an additional unit of benefit (e.g. one more QALY). Policy makers are interested in the ICER value because it facilitates determination about whether newer, more effective interventions represent good value compared to existing, less expensive programs. Interpreting the results of cost-effectiveness analysis can be problematic, making it difficult to decide whether to adopt a diagnostic test or treatment. The threshold for adoption in the United States is thought to be somewhere between $20,000 /QALY gained and $100,000 /QALY gained, with a threshold of $50,000 /QALY gained frequently proposed. In one-way sensitivity analysis, the decision analyst examines what change will occur in the ICER if the value of an input parameter varies across a range of plausible values.
We also recorded quality items specific to VTE including assessment of joint function following hemarthrosis, propagation of asymptomatic DVT to symptomatic PE, incidence of post-thrombotic syndrome, costs of major and minor bleeding, and future costs related to VTE including blood monitoring and physician visits. Studies ignoring downstream bleeding consequences could make newer, more potent regimens appear more cost-effective whereas studies ignoring downstream costs of treating VTE will bias our interpretation in the other direction. We did not specifically document if individual studies included death costs related to VTE or bleeding. On the whole, death events were rare and the associated costs would be largely paid by the family of the patient and not the institution or health system which was the economic perspective chosen by all but three of the studies analyzed.
We did not pool the results of individual studies given the various modeling assumptions adopted by each author. Instead, we qualitatively compared studies to determine trends in the cost-effectiveness of certain regimens in comparison to others.
We identified 370 titles and abstracts meeting our search criteria. Of the 370, 56 were relevant and were entered for full text review. Of these, 33 studies met all inclusion criteria.[8-10, 25-46] [7, 10, 31, 47-53] (Figure 1)
Most studies were set in the United States (14 of 33)[28-30, 32-38, 41, 42, 44, 46] or Europe (14 of 33).[7-9, 26, 27, 31, 39, 45, 47-50, 52, 53]. Twenty studies[7, 8, 25, 28-38, 41, 42, 44-47] adopted an institutional perspective; only three[10, 49, 50] adopted a societal perspective. Ten studies[7, 9, 25, 29, 32, 37, 47, 48, 51, 53] reported pharmaceutical company sponsorship.
There was substantial variation in the quality of reporting. Only six of the 30 studies reported performing a systematic review and meta-analysis of efficacy data.[10, 27, 29, 36, 37, 51] In addition, only 12 studies documented comprehensive use of one-way sensitivity analysis.[7, 10, 26, 27, 29, 33-36, 45, 48, 49, 53]] Only three of 30 studies[10, 29, 50] measured effectiveness in quality-adjusted life-years to at least the one year horizon. (Table 1 and Appendix Table 2).
We included two studies[25, 26] with three comparisons of an anticoagulant to aspirin. (Table 2) In all three comparisons, results were available for THR exclusively. Sarasin et al. found that the ICER was $1700 /VTE avoided for four weeks of warfarin compared to aspirin and $1300 /VTE avoided for four weeks of LMWH compared to aspirin. There was no apparent pharmaceutical company sponsorship for that study. The final comparison, sponsored by Sanofi-Aventis, the manufacturer of enoxaparin, was set in South Africa, reported an ICER of $7200 /VTE avoided for 10 days of enoxaparin compared with 10 days of aspirin.
We included 15 studies with comparisons of LMWH and warfarin.[10, 26-38, 51] (Table 3) Twelve of 14 compared these agents in patients receiving THR. Of those documenting a short horizon cost-effectiveness result, 3 studies [26, 35, 51] found that the ICER for LMWH was ≤$2,000 / VTE avoided compared with warfarin. In two other studies[29, 30], LMWH cost an additional $2100 /VTE avoided. In a sixth study, LMWH cost $5,200 /VTE avoided. In the next study, LMWH cost $109,000 /VTE avoided. This study by Skedgel et al. examined four additional weeks (in addition to the hospital period) of LMWH compared with four additional weeks of warfarin. It found that the cost, in Canada, would be almost ten-fold higher for LMWH given the significant proportion of patients (39% at baseline) that would require daily nursing supervision of LMWH injection in their homes compared with the same proportion that would require weekly home phlebotomy for monitoring INR while using warfarin. In the remaining four studies of short horizon[27, 31, 32, 37], warfarin dominated LMWH.
In comparisons that analyzed cost-effectiveness in the setting of TKR (or TKR cases combined with THR cases), LMWH dominated or cost less than $2,000 / VTE avoided in four studies.[28, 34-36] In the final study, warfarin dominated LMWH.
Eight of 15 studies comparing LMWH to warfarin reported some pharmaceutical company sponsorship, grant support, or involvement of pharmaceutical company consultants. In each case, the pharmaceutical company was the manufacturer of LMWH, either Sanofi-Aventis, Pfizer, or a company which merged with these two. All but two[32, 37] of these eight found favorable cost effectiveness ratios for LMWH. The two studies by government agencies indicated that LMWH was either poor value for its cost or was dominated by warfarin.
We included 10 studies with comparisons of fondaparinux to LMWH.[8, 9, 39-46] (Table 4) Nine of 10 analyzed prophylaxis for THR. Six studies [8, 9, 39, 40, 42, 46] analyzed cost-effectiveness over a short horizon. In all 6, fondaparinux dominated or cost less than $1300 /VTE avoided. In four studies with a long horizon, fondaparinux dominated LMWH. In a fifth, LMWH cost $40 /VTE avoided.
Of the 8 studies reporting cost-effectiveness results for TKR[8, 9, 39, 40, 43-46], all but one found that fondaparinux dominated LMWH over the short and long horizon. In this study , fondaparinux, cost an additional 660 / VTE avoided.
Among the 10 studies comparing fondaparinux to LMWH, a pharmaceutical company sponsored one and supported five more through grants. In each case the sponsor or grantor was Sanofi-Aventis, the manufacturer of enoxaparin (the inferior comparator). Each result demonstrated good value with dominance by the use of fondaparinux.
Only two studies to date have made comparisons with new oral anticoagulants. In the only one which made this comparison in patients undergoing THR, Wolowacz et al. found that dabigatran dominated LMWH over a 60 year horizon (equivalent to a lifetime analysis given the elderly age of the average patient undergoing THR).
In the setting of TKR, McCullagh et al., found that in the short horizon of 180 days, rivarobaxan dominated both LMWH and dabigatran; dabigatran cost only an additional $750 /VTE avoided compared with LMWH. In the long horizon, Wolowacz et al. found that dabigatran dominated LMWH.
The study by Wolowacz et al. was sponsored by Boehringer Ingelheim, the manufacturer of dabigatran, whereas McCullagh reported no sponsorship or support.
We found nine studies[7, 10, 31, 47-52] with comparisons of extended duration vs. short duration prophylaxis in patients undergoing THR. (Table 5) Among short horizon results, three studies [31, 47, 52] with 5 comparisons, found that extended duration therapy after THR either dominated short duration prophylaxis or the ICER was less than $120 /VTE avoided. In Skedgel et al., extended duration warfarin prophylaxis cost an additional $3,200 / VTE avoided but extended duration LMWH cost an additional $27,400 /VTE avoided. In five other studies, the ICER for extended duration therapy was between $7800 and 13,200 /VTE avoided. In McCullagh et al., dabigatran administered for 35 days cost $730,000 /VTE avoided compared with short duration LMWH; the high ICER results mainly from the many fold increased bleeding rates found with dabigatran compared with LMWH (2.0% vs. 0.08%).
Among two THR studies with long horizon results available, Bischof  found that extended duration fondaparinux dominated short duration fondaparinux. Haentjens et al.  found that extended duration enoxaparin cost an additional $9,300 /QALY gained compared with short duration enoxaparin.
For TKR, we found only two studies. At a 35 day horizon, Dranitsaris et al. found that the extended duration dalteparin cost an additional $14,600 /VTE compared with short duration warfarin and $60,000 /VTE compared with short duration dalteparin. At a one year horizon, Haentjens et al. found that extended duration enoxaparin cost an additional $73,000 /QALY gained compared with short duration enoxaparin.
Six of the 10 studies comparing extended duration to short duration therapy included pharmaceutical company sponsorship or grant support. Their was no clear trend among the results with respect to the presence of sponsorship although two of the three studies sponsored exclusively by a government agency found that extended duration therapy with LMWH or dabigatran delivered improved effectiveness at a relatively high cost (between $27,400 and $730,000 /VTE avoided). As mentioned above, the third study by Haentjens et al. found that extended duration LMWH was clearly cost-effective after THR but much less good value after TKR.
Although multiple VTE prophylaxis regimens are supported by the American College of Chest Physicians (ACCP) and the American Academy of Orthopedic Surgeons (AAOS), our systematic review suggests that not all of them may be cost-effective relative to other regimens. There was no consensus about the cost-effectiveness of LMWH compared with warfarin. By contrast, fondaparinux dominated LMWH in nearly every comparison we found. Extended duration prophylaxis with LMWH after THR appeared to be cost-effective with multiple studies indicating extended duration prophylaxis dominates short duration LMWH or cost no more than an additional $10,000 /VTE avoided. Small numbers, predominance of studies analyzing only a short horizon, lack of established cost-effectiveness thresholds for VTE based effectiveness units, and reliance by study authors on venographic endpoints prohibit robust conclusions about the comparisons analyzed.
Comparisons of our work with previous reviews of the economic literature are limited by differences in type of surgery included and publication dates of the included articles. Sullivan et al. summarized the prophylaxis literature between 1984 and 2000 and found that most studies presented consistent findings including that LMWH is cost-effective compared with warfarin. Our results do not support this conclusion. Sullivan et al. based their conclusions on many studies that we excluded because they were published prior to 1997 or which included outcomes from patients undergoing hip fracture surgery. We believe temporal trends[55, 56] in the care of total hip and knee replacement necessitated excluding earlier studies. We also felt that hip fracture surgery identified a distinctive patient population with respect to cost, risk, and benefit issues. Similar to our findings, Sullivan et al. also found that extended duration LMWH was generally cost-effective compared with short duration therapy.
Ivanovic et al. summarized the literature about fondaparinux. These authors concluded that fondaparinux was more cost-effective than LMWH (enoxaparin) 40 mg daily initiated preoperatively but less cost-effective than LMWH 30 mg twice daily initiated postoperatively. Our review did not specifically compare the cost-effectiveness of regimens with LMWH initiated at different times but we found that fondaparinux dominated LMWH in all but one when considering the longer horizon. LMWH dosages in the included studies were evenly distributed between 40 mg daily and 30 mg twice daily. Ivanovic et al. also report not being able to calculate ICERs for two studies whereas we were able to calculate them based on data presented in tables included by the study authors.
Wolowacz et al. also published a review discussing the evolution of model building over a twenty year time span (1987-2006). In terms of quality, the findings of that review were generally consistent with the abstractions we performed, particularly with respect to the paucity of studies measuring QALYs over a sufficiently long period. Unlike their review, we abstracted cost and effect information and independently calculated incremental cost effectiveness ratios for each comparison discussed. We converted costs to 2009 USD and measured effects in common units (total VTE events avoided for short horizon studies and QALYs for long horizon studies). This facilitated comparisons between the multiple regimens supported by major professional societies.
The most salient finding of our review is that fondaparinux dominates LMWH. These results should, however, be interpreted cautiously. There have been only four randomized controlled trials comparing fondaparinux with enoxaparin[59-62] and only one involved patients with TKR surgery. A summary estimate of risk calculated by Turpie et al suggested that fondaparinux offers a 55% reduction in the odds of venographic VTE but no difference in the incidence of symptomatic VTE at postoperative day 11 when screening venography was performed. The studies of cost-effectiveness evaluating fondaparinux generally extrapolated these short horizon venographic rates to estimate the number of symptomatic VTE events. Recent evidence suggests that the ratio of asymptomatic venographic DVT rate to symptomatic DVT rate is between 3 and 7 for THR and between 15 and 24 for TKR. These ratios, however, came from trials using enoxaparin only. Although they do not address this point specifically for fondaparinux, the 2008 ACCP guidelines state that initial efficacy studies using venographic endpoints should be followed with trials that use symptomatic (and objectively confirmed) VTE as endpoints.
There is less conclusive evidence about the duration of prophylaxis although extended prophylaxis with LMWH appears cost-effective compared with short duration therapy in the case of THR surgery. Authors of cost-effectiveness studies included in this review generally summarized efficacy of extended duration prophylaxis with LMWH using one or more of the seven randomized controlled trials[65-71] which reported on the efficacy of extended duration prophylaxis. At least two of these trials[65, 66] did not require venography at the time of discharge from the hospital, permitting assessment of symptomatic VTE rates from four to seven weeks after operation. We cannot draw firm conclusions on the question of extended duration versus short duration of therapy with other agents which have not been studied extensively. Our review also suggests that there is insufficient cost-effectiveness evidence to support extended prophylaxis for TKR. The most recent update of the ACCP guidelines “recommends” extended prophylaxis for THR and “suggests” extended prophylaxis for TKR.
Limitations to our work include differences in economic perspective and setting. As our results overwhelmingly suggest that fondaparinux dominates LMWH, we believe our conclusions are sound for this comparison keeping in mind the absence of trial data measuring symptomatic endpoints. The economic perspective did not appear to explain the variations in results found but we did not have sufficient numbers of studies within each major comparison to make firm statements about the influence of individual differences in analytic methods. Although we converted from foreign currencies to USD using purchasing power parity, cost structures between countries may not be comparable as highlighted by Drumond and Tang.
We also acknowledge the potential bias exerted by pharmaceutical company sponsorship of multiple studies. This bias could have played a role in the comparisons between LMWH and warfarin and extended duration with short duration therapy. They do not appear to have played a role in the comparisons including fondaparinux. Multiple studies sponsored by the manufacturer of LMWH found fondaparinux to be dominant to LMWH. In general, however, we did not have sufficient numbers within each comparison type to determine if variation in study results was related to pharmaceutical company sponsorship
Another major limitation is that there is no established threshold for declaring a prophylaxis regimen cost-effective when disease based units are used to express effectiveness. The QALY permits comparing the value of interventions across diseases given that the utilities which are used to calculate them are standardized to estimates between 0 and 1 where 1 represents perfect health and 0 represents death.
Another limitation includes absence of cost-effectiveness analyses about certain comparisons such as fondaparinux versus warfarin, fondaparinux versus aspirin, and low intensity warfarin (INR < 2) vs. any of the other regimens. We also acknowledge the possibility of English language and publication bias as with any systematic review.
The demand for cost-effectiveness research is growing at a fervent pace. In early 2009, the U.S. government dedicated $1.1 billion to comparative effectiveness research including cost-effectiveness research. The U.S. Centers for Disease Control adopted the results of cost-effectiveness research when it prepared guidelines about screening for HIV infection. Similarly, the United States Preventive Services Task Forces incorporated model results when it updated its most recent colorectal cancer screening recommendations. As the demand for cost-effectiveness work grows, the need to be able to summarize and standardize the information will grow as well. Our work was a comprehensive, systematic review of the cost-effectiveness literature regarding VTE prophylaxis for patients undergoing total joint replacement. In addition, we improved upon previous reviews by standardizing cost-effectiveness information to a common currency and effectiveness unit.
In summary, we found that fondaparinux dominated LMWH in virtually all studies we analyzed but firm conclusions cannot be made until trial data are available which measure symptomatic VTE rates. Extended duration LMWH prophylaxis also appears cost-effective compared with short duration prophylaxis in the case of THR. There is limited evidence to determine the cost-effectiveness of other regimens including extended duration fondaparinux, extended duration LMWH after TKR, prophylaxis with new oral anticoagulants, low-intensity warfarin therapy, or aspirin. These knowledge gaps represent important areas for future research.
This work was performed at Boston Medical Center affiliated with Boston University School of Medicine
Conflicts of Interest: Dr. Kapoor has no conflicts of interest to report
Dr. Chuang has no conflicts of interest to report
Dr. Radhakrishnan has no conflicts of interest to report
Dr. Smith has no conflicts of interest to report
Dr. Berlowitz has no conflicts of interest to report
Dr. Segal has no conflicts of interest to report
Dr. Katz has no conflicts of interest to report
Dr. Losina has no conflicts of interest to report