Journal of Thrombosis and Haemostasis, 12: 1626–1635

DOI: 10.1111/jth.12675

ORIGINAL ARTICLE

Rivaroxaban vs. low molecular weight heparin for the prevention of venous thromboembolism after hip or knee arthroplasty: a cohort study A . L A Z O - L A N G N E R , * † ‡ J . L . F L E E T , ‡ § E . M C A R T H U R ‡ and A . X . G A R G † ‡ § *Division of Hematology, Department of Medicine, University of Western Ontario; †Department of Epidemiology and Biostatistics, University of Western Ontario; ‡Institute for Clinical Evaluative Sciences – Western (ICES Western); and §Division of Nephrology, Department of Medicine, University of Western Ontario, London, ON, Canada

To cite this article: Lazo-Langner A, Fleet JL, McArthur E, Garg AX. Rivaroxaban vs. low molecular weight heparin for the prevention of venous thromboembolism after hip or knee arthroplasty: a cohort study. J Thromb Haemost 2014; 12: 1626–35.

Summary. Background: Rivaroxaban is increasingly used to prevent venous thromboembolism after hip or knee arthroplasty. Studies evaluating the effectiveness of rivaroxaban compared to low molecular weight heparin after orthopedic surgery in routine practice are scarce. Patients and Methods: We conducted a retrospective cohort study in 121 hospitals in Ontario, Canada, between 2002 and 2012. We included patients aged 66 years or older (median age 73 years) who received an outpatient prescription for subcutaneous low molecular weight heparin (n = 11 471) or oral rivaroxaban (n = 12 850) on hospital discharge after a total knee or hip arthroplasty. The two coprimary outcomes assessed within 30 days of the prescription date were emergency department visit or hospitalization with venous thromboembolism (either deep vein thrombosis or pulmonary embolism; primary efficacy outcome) and a hospitalization with non-traumatic major hemorrhage (primary safety outcome). Results: Rivaroxaban use increased over the study period. Compared to low molecular weight heparin, rivaroxaban was associated with a lower 30-day risk of hospitalization with venous thromboembolism (0.47% vs. 0.81%; relative risk 0.58; 95% confidence interval 0.42–0.81; P = 0.001) with no significant difference in hospitalizations for major bleeding (0.18% vs. 0.20%; relative risk 0.89; 95% confidence interval 0.50–1.59; P = 0.700). Conclusions: In routine practice, anticoagulant prophylaxis with rivaroxaban compared to low molecular weight heparin after hospital Correspondence: Alejandro Lazo-Langner, Hematology Division, London Health Sciences Centre, 800 Commissioners Rd E, Rm E6-216A, London, ON N6A 5W9, Canada. Tel.: +1 519 685 8500 Ext. 58833; fax: +1 519 685 8477. E-mail: [email protected] Received 1 April 2014 Manuscript handled by: M. Cushman Final decision: F. R. Rosendaal, 14 July 2014

discharge from total hip or knee arthroplasty is associated with a lower risk of symptomatic venous thromboembolism with no difference in the risk of bleeding. Keywords: arthroplasty; low molecular weight heparin; prophylaxis; rivaroxaban; venous thromboembolism.

Introduction About 2.5 million total hip or knee replacement surgeries are performed worldwide each year. These surgeries are cost effective and result in better patient mobility, pain, and quality of life [1–3]. However, without the use of prophylactic anticoagulation, venous thromboembolism (i.e. deep vein thrombosis or pulmonary embolism) occurs in up to 40%–60% of patients when assessed by venography, with ~ 4.3% of all thromboembolic events being symptomatic [4–6]. The risk of thrombosis is most apparent in the first 35 days after surgery, although it probably extends longer [7]. For these reasons, many evidencebased clinical practice guidelines recommend the use of preventative anticoagulation after hip or knee replacement surgery for a minimum of 10–14 days [4,8]. In this setting, the use of an anticoagulant vs. placebo reduces the risk of venous thromboembolism by 23%–82% [9]. Since the 1980s, subcutaneous low molecular weight heparins (LMWHs) have been the standard anticoagulants used for this purpose. In recent years, rivaroxaban has been approved for this indication. It is a direct factor Xa inhibitor with rapid onset of action, predictable pharmacokinetics, and oral administration [10]. To date, eight randomized trials enrolling almost 14 000 patients show that rivaroxaban is either non-inferior or superior to enoxaparin for preventing venous thromboembolism after hip or knee arthroplasty with no significant differences in major bleeding rates between both drugs [11–18]. However, a recent © 2014 International Society on Thrombosis and Haemostasis

Rivaroxaban vs. LMWH after orthopedic surgery 1627

meta-analysis suggests that rivaroxaban might slightly increase the risk of clinically relevant bleeding [19]. Whereas randomized trials inform about the efficacy of a treatment, they do not necessary reflect on its effectiveness, mainly because in routine care, conditions differ from the highly controlled setting of a randomized trial, and it has been advocated that both types of studies (i.e. interventional and observational) are complementary [20,21]. Many practice guidelines have now incorporated rivaroxaban in their recommendations based on the information of randomized trials, without considering its effectiveness. This prompted the present cohort study to assess the outcomes of rivaroxaban compared to LMWH for venous thromboembolism prevention among older adults who underwent hip or knee arthroplasty for osteoarthritis. Methods Study design and setting

We conducted a retrospective cohort study using linked health care databases in Ontario, Canada, with a population of ~ 13 million residents, of whom 14% are 65 years of age or older [22]. In Ontario, residents have universal access to hospital and physician services; in addition, those 65 years of age or older have universal prescription drug coverage through the provincial drug formulary. Rivaroxaban was added to the formulary in June 2009. We conducted this study at the Institute for Clinical Evaluative Sciences (ICES) according to a prespecified protocol that was approved by the research ethics board at Sunnybrook Health Sciences Centre (Toronto, Canada).

revision (ICD-10) (post-2002) codes were in use to define baseline comorbidities in the 5 years before receipt of the relevant coprescription. All events occurred in a period after implementation of the ICD-10 coding system. Patients

We included residents of Ontario, Canada, between June 2002 and March 2012 who received an outpatient prescription for rivaroxaban, dalteparin, enoxaparin, or tinzaparin on the date of or the day after a hospital discharge for total knee or hip arthroplasty for gonarthrosis or coxarthrosis. Patients were included if they were aged 66 years or older at the time of the initial prescription. The date of the study drug prescription served as the index date (start time for follow-up). Patients were excluded if they (i) were in their first year of eligibility for prescription drug coverage (age 65) (to avoid incomplete medication records), (ii) had received a prescription for any of the study drugs or for warfarin in the 180 days before the index date, (iii) had evidence of a condition necessitating chronic anticoagulation such as a prosthetic mechanical heart valve that (to ensure the postoperative anticoagulation was used solely for the purpose of venous thromboembolism prophylaxis) or had evidence of a deep vein thrombosis or pulmonary embolism in the 180 days before the index date (to restrict to new events in follow-up), (iv) had evidence of having received a prescription for more than one study drug on the index date (to compare mutually exclusive groups), (v) required dialysis (to avoid confounding from heparin utilization during dialysis), or (vi) lived in a long-term care facility (as these patients might be significantly less mobile). Cohort creation is shown in Fig. 1.

Data sources

We obtained patient characteristics, drug use, covariate information, and outcome data using records from six databases. The Ontario Registered Persons Database contains demographic and vital statistics information on all Ontario residents who have ever been issued a health card. The Ontario Drug Benefit Plan database contains highly accurate records of all outpatient prescriptions dispensed to patients aged 65 or older, with an error rate of < 1% [23]. The Canadian Institute for Health Information Discharge Abstract Database and the Ontario Health Insurance Plan database contain diagnostic and procedural information on all inpatient and outpatient physician services. The National Ambulatory Care Reporting System contains information on all emergency department visits. Finally, the ICES Physician Database contains prescriber information. These data sets were held securely in a linked, deidentified form and analyzed at the ICES. The databases were complete for all variables used in this study (with the exception of income, which was missing in 0.3% of patients). Both International Classification of Diseases, Ninth Revision (ICD-9) (pre-2002) and the 10th © 2014 International Society on Thrombosis and Haemostasis

Study outcomes

The study had two prespecified coprimary outcomes. The primary efficacy outcome was hospitalization or emergency department visit for venous thromboembolism (deep vein thrombosis or pulmonary embolism ) within the 30 days after the index date. The primary safety outcome was hospitalization or emergency department visit with a non-traumatic major hemorrhage (defined as subarachnoid, intracerebral, upper and lower gastrointestinal tract) within the first 30 days after the index date. The secondary outcomes were the same as the primary outcomes but ascertained at 90 days after the index date, as well as hospitalization with a digestive system endoscopy (a proxy for gastrointestinal bleeding) and all-cause mortality, both evaluated at 30 and 90 days after the index date. Coding definitions are shown in Tables S1 and S2. The codes for the outcomes have been reported to have high positive predictive values in Ontario and elsewhere [24–26]. For hospitalization records, patients with codes for multiple study outcomes were accounted for in the assessment of each outcome.

1628 A. Lazo-Langner et al

Patients included

Patients excluded

Patients with a prescription for study drug (n = 112 591)

Age less than 66 (n = 4880) Patients remaining in the cohort (n = 107 711) Evidence of prescription for a study drug or warfarin 180 days prior to prescription date (n = 24 328) Patients remaining in the cohort (n = 83 383) Evidence of mechanical heart valve or deep vein thrombosis, pulmonary embolism within 180 days prior to prescription date (n = 5805) Patients remaining in the cohort (n = 77 578) Evidence of prescription for more than one study drug on prescription date (n = 48) Patients remaining in the cohort (n = 77 530) Evidence of dialysis within 1 year prior to prescription date (n = 1902) Patients remaining in the cohort (n = 75 628) Evidence of long term care facility utilization (n = 13 757) Patients remaining in the cohort (n = 61 871) Patients not discharged from hospital on the day of or the day prior to prescription date (n = 30 211) Patients remaining in the cohort (n = 31 660) Patients not admitted for gonarthrosis or coxarthrosis within 1 week of prescription date (n = 7339) Patients included in the final cohort (n = 24 321) Rivaroxaban (n = 12 850) Dalteparin (n = 5941) Enoxaparin(n = 4126) Tinzaparin (n = 1404) Fig. 1. Flow diagram of cohort integration.

Statistical analysis

The analysis of all outcomes compared patients receiving rivaroxaban to the reference group of patients receiving any study LMWH. Baseline characteristics were compared between groups using standardized differences. This metric describes differences between group means relative to the pooled standard deviation and is considered clinically meaningful if > 10% [27]. We calculated 95%

confidence intervals (CIs) for proportions using the Wilson score method [28]. To estimate odds ratios and 95% CIs, we used logistic regression analysis, unadjusted and adjusted for age (per year), sex, hospital setting (academic hospital vs. non-academic), history of major cancer (yes or no), and type of surgery (hip or knee arthroplasty). We interpreted odds ratios as relative risks, which is appropriate given the incidences observed [29]. Survival analysis was performed using the Kaplan–Meier method © 2014 International Society on Thrombosis and Haemostasis

Rivaroxaban vs. LMWH after orthopedic surgery 1629 Table 1 Baseline characteristics of the cohort Low molecular weight heparin (n = 11 471) Age (yrs) (mean [SD]) Age group (n [%]) 66–70 yrs 71–75 yrs 76–80 yrs 81–85 yrs 86–90 yrs > 90 yrs Women (n [%]) Rural place of residence (n [%]) Income quintile† (n [%]) Quintile 1 (lowest) Quintile 2 Quintile 3 Quintile 4 Quintile 5 (highest) Missing Year of cohort entry (n [%])‡ 2002–2004 2005–2008 2009–2012 Number of primary care visits in prior year (median [IQR]) Hospitalization setting (n [%]) Academic hospital Length of hip or knee arthroplasty hospitalization in days (median [IQR]) Date of most recent hospital discharge On index date One day before index date Arthroplasty performed within 30 days before index date Hip arthroplasty within 30 days before index date Knee arthroplasy within 30 days before index date Comorbidities§ (n [%]) Major cancer¶ Bowel cancer Breast cancer Lung cancer Prostate cancer Chronic kidney disease Coronary artery disease (excluding angina) Diabetes mellitus Congestive heart failure Cerebrovascular event** Atrial fibrillation/flutter†† Medications‡‡ (n [%]) ACE inhibitors Angiotensin receptor blockers Anticonvulsants Antineoplastic medications Antipsychotics b-Blockers Calcium channel blockers Loop diuretics Potassium-sparing diuretics Thiazide diuretics Aspirin§§

Rivaroxaban (n = 12 850)

Standardized difference*

73.89

5.53

73.63

5.55

3763 3528 2619 1254 279 28 6831 2536

32.80 30.76 22.83 10.93 2.43 0.24 59.55 22.11

4541 3774 2890 1331 295 19 7535 2702

35.34 29.37 22.49 10.36 2.30 0.15 58.64 21.03

0.05 0.03 0.01 0.02 0.01 0.02 0.02 0.03

1969 2400 2319 2391 2358 34

17.17 20.92 20.22 20.84 20.56 0.30

2065 2518 2586 2678 2960 43

16.07 19.60 20.12 20.84 23.04 0.33

0.03 0.03 0.00 0.00 0.06 0.01

762 4682 6027 12

6.64 40.82 52.54 8–18

0 0 12 850 11

0 0 100 7–17

0.38 1.17 1.34

2375 4

20.70 3–5

1944 4

15.13 3–5

0.15 0.17

9690 1781 11 444

84.47 15.53 99.76

11 584 1266 12841

90.15 9.85 99.93

0.17 0.17 0.04

3947

34.41

4692

36.51

0.04

7497

65.35

8149

63.42

0.04

1395 352 466 112 559 518 2501

12.16 3.07 4.06 0.98 4.87 4.52 21.80

1617 410 511 105 700 537 3174

12.58 3.19 3.98 0.82 5.45 4.18 24.70

0.01 0.01 0.00 0.02 0.03 0.02 0.07

1770 743 123 336

15.43 6.48 1.07 2.93

1854 614 110 368

14.43 4.78 0.86 2.86

0.03 0.07 0.02 0.00

3533 2248 401 385 111 2893 3113 645 621 2595 480

30.80 19.60 3.50 3.36 0.97 25.22 27.14 5.62 5.41 22.62 4.18

3857 2564 535 388 145 2713 3291 615 479 2718 296

30.02 19.95 4.16 3.02 1.13 21.11 25.61 4.79 3.73 21.15 2.30

0.02 0.01 0.03 0.02 0.02 0.10 0.03 0.04 0.08 0.04 0.11

© 2014 International Society on Thrombosis and Haemostasis

1630 A. Lazo-Langner et al Table 1 (Continued) Low molecular weight heparin (n = 11 471) NSAIDs (except aspirin) Statins

4265 4638

37.18 40.43

Rivaroxaban (n = 12 850) 4506 5499

35.07 42.79

Standardized difference* 0.04 0.05

ACE, angiotensin-converting enzyme; NSAIDs, non-steroidal anti-inflammatory drugs. *Standardized differences provide a measure of the difference between groups divided by the pooled standard deviation; a value > 10% (0.1) is interpreted as a meaningful difference between the groups. †Income was categorized into fifths of the neighbourhood average income on the index date. The cohort entry date (index date) is the date of the prescription of the study drug. ‡In earlier years, a lower proportion of patients were receiving prophylactic low molecular weight heparin, or no prescription information was available. §Comorbidities were assessed by administrative database codes in the 5 years before index date. ¶Major cancer included esophagus, lung, bowel, liver, pancreas, breast, male/female reproductive organs, leukemias, and lymphomas. **Includes stroke and transient ischemic attack. ††As per the exclusion criteria used to construct the cohort, no patient with a history of atrial fibrillation or flutter was receiving chronic anticoagulants. ‡‡Baseline medication use was assessed during the 120 days before index date anticoagulants. §§Aspirin includes only prescription aspirin.

[30]. To evaluate the robustness of the findings, we conducted a number of additional analyses as described in the Results section. For subgroup analyses, interactions were assessed using a Wald v2 test, when appropriate. Absolute risk was also expressed as the ‘number needed to treat’ (1/ absolute risk difference). This measure indicates how many patients need to receive rivaroxaban to prevent one venous thromboembolic event if all patients received LMWH. The number needed to treat was calculated for ease of interpretation and not to imply causality in this observational study. For all results, a P-value < 0.05 was considered statistically significant. We performed all analyses using Microsoft Excel 2010 (Microsoft Corp., Redmond, WA, USA), OpenEpi version 3.01 (www.openepi.com or www.epiinformatics.com) [31] and SAS 9.3 (SAS Institute Inc., Cary, NC, USA). Results Population characteristics

The cohort consisted of 24 321 patients, of whom 11 471 received LMWH and 12 850 received rivaroxaban. The baseline characteristics of the cohort are shown in Table 1, and the two groups were well balanced. The mean age was 74 years; 59% were women; and 34% underwent hip replacement, whereas the remainder had a knee replacement. Arthroplasty was performed within the 7 days preceding the index date in 91% and 95% of patients in the LMWH and rivaroxaban groups, respectively. Anticoagulants were dispensed for a median of 14 days after hospital discharge (interquartile range 10–21), with no differences among individual anticoagulants (Table S3). Primary outcomes

Figure 2 shows the Kaplan–Meier curves for the primary outcomes. In the whole cohort, within 30 days there were 154 venous thromboembolic events and 46 major bleeding events for a 30-day event rate of 0.63% and 0.19%,

respectively. In the rivaroxaban vs. LMWH groups, there were 61 (0.47%) vs. 93 (0.81%) venous thromboembolic events, respectively. Rivaroxaban was associated with a 42% risk reduction for venous thromboembolism (95% CI 19%–58%), with a number needed to treat of 298 (95% CI 185–749). During the same period there were 23 (0.18%) vs. 23 (0.20%) major bleeding events in the rivaroxaban and LMWH groups, respectively, with an unadjusted relative risk (RR) of 0.89 (95% CI 0.50–1.59; P = 0.700). Adjusting for potential confounders did not change the results (Table 2). Secondary outcomes

Secondary outcomes results are shown in Table 2. Compared to LMWH rivaroxaban was associated with a 41% reduction in the risk of developing venous thromboembolism within 90 days, with no difference between the two groups in major bleeding events. No differences between the groups were observed in hospitalizations with digestive system endoscopy (a proxy for gastrointestinal bleeding) within 30 and 90 days of the index date, nor was there a difference between the groups when this outcome was combined with bleeding events assessed with hospital diagnostic codes (Table 2). There was no significant difference between the groups in all-cause mortality within 30 and 90 days. Additional analyses

We conducted additional analyses to assess the robustness of the findings. The results were not different when we used a Cox proportional hazards model rather than a logistic regression model (Table S4). Since rivaroxaban became publicly funded in 2009, we restricted the analysis to patients accrued between 2009 and 2012. The results were consistent with those of the primary analysis (Table S5). Additionally, we did not observe any difference in secular trends for venous thromboembolic events among patients receiving LMWH over the accrual period, © 2014 International Society on Thrombosis and Haemostasis

Rivaroxaban vs. LMWH after orthopedic surgery 1631

Thrombosis - free survival

A

Venous thromboembolism 1 0.9

1

0.8

0.995

Bleeding - free survival

0.99

0.7

0.985

0.6

0.98

0.5

0.975

0.4

0.97 0.965

0.3

0.96

0.2

0.955

0.1

0.95

0

B

LMWH Rivaroxaban

0

15

15

0

30

45

60

30 45 60 Days since prescription Major hemorrhage

1

0.8

0.995

75

90

90

Discussion LMWH Rivaroxaban

1 0.9

75

0.99

0.7

0.985

0.6

0.98

0.5

0.975

0.4

0.97 0.965

0.3 0.96

0.2

0.955

0.1

0.95

0

0

0

15

15

30

45

60

30 45 60 Days since prescription

difference in event rates for the various outcomes between the three types of LMWH. The point estimate of 30-day incidence of venous thromboembolism was lower with rivaroxaban, 0.47% (95% CI 0.37–0.60%), than any of the three LMWHs. Finally, we conducted additional analyses to exclude the possibility of a differential effect in patients diagnosed with a venous thromboembolic event without any hospital or emergency department encounter. At 30 days, there were 42 events (0.37%; 95% CI 0.27–0.49) in the LMWH group compared to 35 (0.27%, 95% CI 0.20–0.38) in the rivaroxaban group (P = 0.19). After adding all events, the unadjusted RR was 0.63 (95% CI 0.49–0.82; P = 0.001), whereas the adjusted RR was 0.58 (95% 0.42–0.80; P = 0.001), in favor of rivaroxaban.

75

75

90

90

Fig. 2. Kaplan–Meier curves for venous thromboembolism (A) and major hemorrhage (B) (see text for definitions).

suggesting that there were no other recent changes in care that influenced outcomes (v2 P-value 0.94; Table S6). When the analysis was restricted to patients seen by physicians who prescribed both LMWH and rivaroxaban during the study period (which reduced cohort size by 39%), rivaroxaban (vs. LMWH) was associated with a non-significant trend in a lower 30-day risk of venous thromboembolism (RR 0.68, 95% CI 0.38–1.20; P = 0.182). When the analysis was stratified by type of joint replacement, a trend toward greater benefit of rivaroxaban vs. LMWH was observed in knee replacement rather than hip replacement, but subgroup results were not statistically significant (P-value for interaction 0.24; Table S7). To consider potential differences in outcomes with different LMWHs, we separately analyzed the proportion of events observed with dalteparin, enoxaparin, and tinzaparin (Table S8). There was no significant © 2014 International Society on Thrombosis and Haemostasis

We conducted this cohort study to evaluate the outcomes of older patients receiving rivaroxaban for venous thromboembolism prevention after hip or knee arthroplasty. Our data show that in routine care, there has been a rapid uptake of rivaroxaban for the prevention of venous thromboembolism after such surgeries. After rivaroxaban was included in the provincial formulary in 2009, there were twice as many prescriptions for rivaroxaban compared to LMWH. However, many centers in Ontario (and likely elsewhere) are still using LMWH for venous thromboembolism prophylaxis. We believe that our data in conjunction with data from the randomized trials support a decision to shift practice toward the preferential use of rivaroxaban (vs. LMWH) in this setting. Our study results are similar to those reported in randomized controlled trials and a recent large cohort study. A recent systematic review and meta-analysis including eight randomized trials and 13 764 patients reported that rivaroxaban was associated with a 52% risk reduction for venous thromboembolism compared to LMWH [19], a similar effect size to that found in our study. However, unlike our results, the same authors reported that rivaroxaban was associated with a 25% increase in clinically relevant bleeding. The difference between the two estimates may be explained by the difference in the definition of bleeding events, since our study captured only events requiring hospitalization and did not account for other potential bleeding events that might still be relevant. Furthermore, the recently published XAMOS prospective cohort study that included 17 701 patients receiving either rivaroxaban or standard-of-care also reported thrombosis and major bleeding incidences that were similar to our own with the caveat that in such study ~ 20% of the patients in the standard-of-care arm received other agents instead of LMWH [32]. Our study has several strengths. By using comprehensive provincial healthcare databases we provide generalizable results of all patients undergoing hip or knee

1632 A. Lazo-Langner et al Table 2 Primary and secondary clinical outcomes in patients undergoing total hip or knee replacement receiving rivaroxaban or low molecular weight heparin

n/N

% (95% CI)

Unadjusted RR

95% CI

P-value

Adjusted RR*

95% CI

P-value

Primary outcomes Emergency department visit or hospitalization with venous thromboembolism within 30 days LMWH 93/11 471 0.81 (0.66–0.99) 1 Ref. 1 Ref. Rivaroxaban 61/12 850 0.47 (0.37–0.61) 0.58 0.42–0.81 0.001 0.58 0.42–0.80 0.001 Hospitalization with non-traumatic major hemorrhage within 30 days LMWH 23/11 471 0.20 (0.13–0.31) 1 Ref. 1 Ref. Rivaroxaban 23/12 850 0.18 (0.12–0.27) 0.89 0.50–1.59 0.700 0.88 0.49–1.58 0.675 Secondary outcomes Emergency department visit or hospitalization with venous thromboembolism within 90 days LMWH 134/11 136 1.20 (1.01–1.43) 1 Ref. 1 Ref. Rivaroxaban 84/11 823 0.71 (0.57–0.88) 0.59 0.45–0.77 < 0.001 0.59 0.43–0.81 0.001 Hospitalization with non-traumatic major hemorrhage within 90 days LMWH 30/11 136 0.27 (0.19–0.38) 1 Ref. 1 Ref. Rivaroxaban 28/11 823 0.24 (0.16–0.34) 0.88 0.53–1.47 0.623 0.63 0.34–1.16 0.138 Hospitalization with digestive system endoscopy (a proxy for gastrointestinal bleeding) within 30 days LMWH 26/11 471 0.23 (0.16–0.33) 1 Ref. 1 Ref. Rivaroxaban 31/12 850 0.24 (0.17–0.34) 1.06 0.63–1.79 0.814 1.05 0.62–1.76 0.869 Hospitalization with digestive system endoscopy (a proxy for gastrointestinal bleeding) within 90 days LMWH 42/11 136 0.38 (0.28–0.51) 1 Ref. 1 Ref. Rivaroxaban 47/11 823 0.40 (0.30–0.53) 1.05 0.69–1.60 0.804 0.76 0.47–1.23 0.270 Hospitalization with non-traumatic major hemorrhage or digestive system endoscopy (a proxy for gastrointestinal bleeding) within 30 days LMWH 34/11 471 0.30 (0.21–0.41) 1 Ref. 1 Ref. Rivaroxaban 34/12 850 0.26 (0.19–0.37) 0.89 0.55–1.44 0.639 0.87 0.54–1.40 0.567 Hospitalization with non-traumatic major hemorrhage or digestive system endoscopy (a proxy for gastrointestinal bleeding) within 90 days LMWH 51/11 136 0.46 (0.35–0.60) 1 Ref. 1 Ref. Rivaroxaban 49/11 823 0.41 (0.31–0.55) 0.91 0.61–1.34 0.617 0.90 0.61–1.33 0.591 All cause mortality within 30 days† LMWH 6/11 471 0.05 (0.02–0.11) – – – – – – Rivaroxaban ≤ 5/12 850† ≤ 0.04 (0.01–0.10) – – – – – – All-cause mortality within 90 days LMWH 25/11 136 0.22 (0.15–0.34) 1 Ref. 1 Ref. Rivaroxaban 16/11 823 0.14 (0.08–0.22) 0.60 0.32–1.13 0.114 0.52 0.27–1.02 0.058 CI, confidence interval; LMWH, low molecular weight heparin; n, number of events; N, number of patients; Ref, reference; RR, relative risk. *Adjusted for age (per year), sex, hospital setting (rural vs. urban), history of major cancer (yes or no), and type of surgery (THR or TKR). †Cell sizes of five or less are not reported according to institutional policy.

replacement across 121 hospitals. The large sample of > 24 000 patients provided good precision to describe associations of the anticoagulants studied with important but uncommon events. The baseline characteristics of rivaroxaban and LMWH groups were highly similar, and the results proved robust in multiple additional analyses. Finally, the data were complete, drug dispensing and outcomes were accurately recorded, and patient loss to follow-up was minimal due to the short time frame and a low emigration rate (< 1% per year) [33]. The present study has limitations. It was done retrospectively, and a lack of prospective data collection with central adjudication of events raises the possibility of outcome misclassification, although there is no reason to believe that this might have differed between groups. This is further supported by our multiple additional analyses. However, diagnostic codes for VTE have been demonstrated to be reasonably accurate for use in outcomes research in Ontario and elsewhere (sensitivity 97%, specificity 74%) [24,34]. Because we could not exclude the

possibility that some patients were managed entirely as outpatients, we conducted additional analyses in patients not having any hospital encounters. About 0.3% of the population had an event without a hospital or emergency department encounter and, again, the RR was similar to the primary analysis. Also, we do not have information on in-hospital use of thromboprophylaxis, and although it is unlikely that a patient started in-hospital on postoperative rivaroxaban would be switched to LMWH after discharge, we cannot completely rule out an effect of inhospital use of anticoagulants on the occurrence of the study outcomes. Additionally, we cannot completely exclude the possibility of survival bias, and our results apply only to patients who were discharged from hospital without any of the outcomes on discharge. Regarding bleeding events, we only assessed events that had a diagnostic code such as intracranial hemorrhage and gastrointestinal bleeding, but other potentially relevant bleeding events (e.g. compartment syndrome, wound hematoma) were not considered, and there is a possibility that in © 2014 International Society on Thrombosis and Haemostasis

Rivaroxaban vs. LMWH after orthopedic surgery 1633

daily practice rivaroxaban might result in a higher frequency of such events, as suggested by randomized trials. However, our coding definitions did capture most relevant events, in particular the most feared complication of intracranial bleeding. As universal drug coverage in Ontario only includes patients aged 65 or older, we cannot draw any conclusions for younger patients, although it would not be unreasonable to expect similar RRs. We excluded patients with a recent history of dialysis or other indications for chronic anticoagulation, and our findings do not apply to these groups. Furthermore, we had information only on medication dispensing and not on adherence. It is possible that the observed beneficial effects of rivaroxaban compared to LMWH in routine care were due to a higher non-adherence among LMWH users because of the inconvenience of administering daily injections. However, this could not be assessed in our data sources. Also, we could not assess the influence of over-the-counter use of aspirin or non-steroidal antiinflammatory drugs on the risk of bleeding or venous thromboembolism given that recent studies suggest a benefit of aspirin for primary and secondary venous thromboembolism prevention [35–37]. However, the possibility of differential use of these agents among the two study groups is unlikely. Finally, although our primary end points were defined at 30 days after the index date, we also included 90-day outcomes with consistent results. In conclusion, the present study shows that in daily clinical practice, rivaroxaban is at least as effective as LMWH for preventing clinically important venous thromboembolism among older patients undergoing hip or knee arthroplasty who are discharged from the hospital without complications. It was not associated with a higher risk of bleeding as defined in this study. The results support the generalizability of the findings from randomized clinical trials and the adoption of rivaroxaban in routine care.

Addendum A. Lazo-Langner, J. L. Fleet, E. McArthur, and A. X. Garg designed the study and analyzed the data. A. LazoLangner wrote the first draft of the manuscript, and J. L. Fleet, E. McArthur, and A. X. Garg edited the manuscript. A. Lazo-Langner and E. McArthur had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All persons who contributed significantly to this work have been acknowledged. Acknowledgements This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and LongTerm Care (MOHLTC). The opinions, results and © 2014 International Society on Thrombosis and Haemostasis

conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred. This study was funded in part by the Canadian Institutes of Health Research. A. Lazo-Langner is supported in part by the ICES Western Scholars program and The Academic Medical Organization of Southwestern Ontario (AMOSO) Opportunities Fund. Disclosure of Conflict of Interest A. Lazo-Langner has received honoraria from Pfizer, Bayer, Leo Pharma, and Boehringer Ingelheim and has participated in studies sponsored by Pfizer, Leo Pharma, Bayer, Daiichi-Sankyo, Novartis, and Celgene. A. X. Garg received an investigator-initiated grant from Astellas and Roche to support a Canadian Institutes of Health Research study in living kidney donors, and his institution received unrestricted research funding from Pfizer. None of these entities were involved in any aspect of this study. A. X. Garg reports grants from Pfizer, outside the submitted work. A. Lazo-Langner reports personal fees from Pfizer, Leo Pharma, and Boehringer Ingelheim; honoraria from Pfizer, Leo Pharma, and Bayer; participation in research studies for Pfizer, Leo Pharma, Boehringer Ingelheim, Bayer, Daiichi Sankyo, Novartis, and Celgene; and an unrestricted grant from Alexion, outside the submitted work. All other authors declare no conflicts of interest.

Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. Coding definitions used in the study. Table S2. Coding definitions for comorbid conditions. Table S3. Duration of anticoagulant prescription periods in days. Table S4. Primary and secondary clinical outcomes in patients undergoing total hip or knee replacement receiving rivaroxaban or low molecular weight heparin estimated using a Cox proportional hazard regression model. Table S5. Sensitivity analyses for outcomes in patients included in the cohort between 2009 and 2012. Table S6. Secular trends for venous thromboembolism in patients receiving low molecular weight heparin. Table S7. Subgroup analyses of 30- and 90- day outcomes according to type of joint replaced. Table S8. Proportion of events according to individual low molecular weight heparin. References 1 de Pina MDF, Ribeiro AI, Santos C. Epidemiology and variability of orthopaedic procedures worldwide. In: Bentley G, ed.

1634 A. Lazo-Langner et al

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

European Instructional Lectures Vol. 11, 11 edn. 12th EFORT Congress, Copenhagen, Denmark. Berlin, Germany: Springer, 2011: 9–19. Kurtz S, Ong K, Lau E, Mowat F, Halpern M. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg 2007; 89: 780–5. Weiser TG, Regenbogen SE, Thompson KD, Haynes AB, Lipsitz SR, Berry WR, Gawande AA. An estimation of the global volume of surgery: a modelling strategy based on available data. Lancet 2008; 372: 139–44. Falck-Ytter Y, Francis CW, Johanson NA, Curley C, Dahl OE, Schulman S, Ortel TL, Pauker SG, Colwell CW. Prevention of VTE in orthopedic surgery patients. Chest 2012; 141: e278S– 325S. Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR, Colwell CW. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008; 133: 381S–453S. Lazo-Langner A. Comparing Strategies for Thromboprophylaxis in Major Orthopedic Surgery Using an Estimation of Net RiskBenefit Through Probabilistic Simulation. A Clinical Cost-Effectiveness Study. Ottawa, Canada: University of Ottawa, 2007. Planes A, Vochelle N, Darmon JY, Fagola M, Bellaud M, Huet Y. Risk of deep venous thrombosis after hospital discharge in patients having undergone total hip replacement: double-blind randomised comparison of enoxaparin versus placebo. Lancet 1996; 48: 224–8. American Academy of Orthopaedic Surgeons. Clinical Guideline on Prevention of Pulmonary Embolism in Patients Undergoing Total Hip or Knee Arthroplasty. American Academy of Orthopaedic Surgeons, 2007. http://www.aaos.org/Research/guidelines/ PE_guideline.pdf. Accessed 26 September 2008. Arcelus JI, Kudrna JC, Caprini JA. Venous thromboembolism following major orthopedic surgery: what is the risk after discharge? Orthopedics 2006; 29: 506–16. Lazo-Langner A, Lang E, Douketis J. Clinical review: clinical management of new oral anticoagulants: a structured review with emphasis on the reversal of bleeding complications. Crit Care 2013; 17: 230. Kakkar AK, Brenner B, Dahl OE, Eriksson BI, Mouret P, Muntz J, Soglian AG, Pap AF, Misselwitz F, Haas S. Extended duration rivaroxaban versus short-term enoxaparin for the prevention of venous thromboembolism after total hip arthroplasty: a double-blind, randomised controlled trial. Lancet 2008; 372: 31–9. Lassen MR, Ageno W, Borris LC, Lieberman JR, Rosencher N, Bandel TJ, Misselwitz F, Turpie AG. Rivaroxaban versus enoxaparin for thromboprophylaxis after total knee arthroplasty. N Engl J Med 2008; 358: 2776–86. Turpie AG, Lassen MR, Davidson BL, Bauer KA, Gent M, Kwong LM, Cushner FD, Lotke PA, Berkowitz SD, Bandel TJ, Benson A, Misselwitz F, Fisher WD. Rivaroxaban versus enoxaparin for thromboprophylaxis after total knee arthroplasty (RECORD4): a randomised trial. Lancet 2009; 373: 1673–80. Eriksson BI, Borris LC, Dahl OE, Haas S, Huisman MV, Kakkar AK, Misselwitz F, Muehlhofer E, Kalebo P. Dose-escalation study of rivaroxaban (BAY 59-7939)–an oral, direct Factor Xa inhibitor–for the prevention of venous thromboembolism in patients undergoing total hip replacement. Thromb Res 2007; 120: 685–93. Turpie AG, Fisher WD, Bauer KA, Kwong LM, Irwin MW, Kalebo P, Misselwitz F, Gent M. BAY 59-7939: an oral, direct factor Xa inhibitor for the prevention of venous thromboembolism in patients after total knee replacement. A phase II doseranging study. J Thromb Haemost 2005; 3: 2479–86. Eriksson BI, Borris L, Dahl OE, Haas S, Huisman MV, Kakkar AK, Misselwitz F, Kalebo P. Oral, direct Factor Xa inhibition

17

18

19

20

21 22

23

24

25

26

27

28

29 30 31

32

33

34

35

with BAY 59-7939 for the prevention of venous thromboembolism after total hip replacement. J Thromb Haemost 2006a; 4: 121–8. Eriksson BI, Borris LC, Dahl OE, Haas S, Huisman MV, Kakkar AK, Muehlhofer E, Dierig C, Misselwitz F, Kalebo P. A once-daily, oral, direct Factor Xa inhibitor, rivaroxaban (BAY 59-7939), for thromboprophylaxis after total hip replacement. Circulation 2006b; 114: 2374–81. Eriksson BI, Borris LC, Friedman RJ, Haas S, Huisman MV, Kakkar AK, Bandel TJ, Beckmann H, Muehlhofer E, Misselwitz F, Geerts W. Rivaroxaban versus enoxaparin for thromboprophylaxis after hip arthroplasty. N Engl J Med 2008; 358: 2765–75. Gomez-Outes A, Terleira-Fernandez AI, Suarez-Gea ML, Vargas-Castrillon E. Dabigatran, rivaroxaban, or apixaban versus enoxaparin for thromboprophylaxis after total hip or knee replacement: systematic review, meta-analysis, and indirect treatment comparisons. Br Med J 2012; 344: e3675. Nallamothu BK, Hayward RA, Bates ER. Beyond the randomized clinical trial: the role of effectiveness studies in evaluating cardiovascular therapies. Circulation 2008; 118: 1294–303. Black N. Why we need observational studies to evaluate the effectiveness of health care. Br Med J 1996; 312: 1215–8. Statistics Canada. Estimates of population, by age group and sex. Canadian Socioeconomic Database. 2013. http://www5.statcan.gc.ca/cansim/home-accueil?lang=eng. Accessed 26 August 2013. Levy AR, O’Brien BJ, Sellors C, Grootendorst P, Willison D. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003; 10: 67–71. Arnason T, Wells PS, van Walraven C, Forster AJ. Accuracy of coding for possible warfarin complications in hospital discharge abstracts. Thromb Res 2006; 118: 253–62. van Walraven C, Mamdani M, Cohn A, Katib Y, Walker M, Rodger MA. Risk of subsequent thromboembolism for patients with pre-eclampsia. Br Med J 2003; 326: 791–2. Jha P, Deboer D, Sykora K, Naylor CD. Characteristics and mortality outcomes of thrombolysis trial participants and nonparticipants: a population-based comparison. J Am Coll Cardiol 1996; 27: 1335–42. Austin PC. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat Simul Comput 2009; 38: 1228–34. Newcombe RG. Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat Med 1998; 17: 857–72. Cummings P. The relative merits of risk ratios and odds ratios. Arch Pediatr Adolesc Med 2009; 163: 438–45. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–81. Dean AG, Sullivan KM, Soe MM. OpenEpi: open source epidemiologic statistics for public health. Version 3.01. www.OpenEpi.com. Accessed 4 June 2013. Turpie AG, Haas S, Kreutz R, Mantovani LG, Pattanayak CW, Holberg G, Jamal W, Schmidt A, van Eickels M, Lassen MR. A non-interventional comparison of rivaroxaban with standard of care for thromboprophylaxis after major orthopaedic surgery in 17,701 patients with propensity score adjustment. Thromb Haemost 2014; 111: 94–102. Ontario Ministry of Finance. Ontario Population Projections Update 2012–2036. Toronto, ON, Canada: Queen’s Printer for Ontario, 2013. Tagalakis V, Kahn SR. Determining the test characteristics of claims-based diagnostic codes for the diagnosis of venous thromboembolism in a medical service claims database. Pharmacoepidemiol Drug Saf 2011; 20: 304–7. Becattini C, Agnelli G, Schenone A, Eichinger S, Bucherini E, Silingardi M, Bianchi M, Moia M, Ageno W, Vandelli MR, © 2014 International Society on Thrombosis and Haemostasis

Rivaroxaban vs. LMWH after orthopedic surgery 1635 Grandone E, Prandoni P. Aspirin for preventing the recurrence of venous thromboembolism. N Engl J Med 2012; 366: 1959– 67. 36 Brighton TA, Eikelboom JW, Mann K, Mister R, Gallus A, Ockelford P, Gibbs H, Hague W, Xavier D, Diaz R, Kirby A, Simes J. Low-dose aspirin for preventing recurrent venous thromboembolism. N Engl J Med 2012; 367: 1979–87.

© 2014 International Society on Thrombosis and Haemostasis

37 Anderson DR, Dunbar MJ, Bohm ER, Belzile E, Kahn SR, Zukor D, Fisher W, Gofton W, Gross P, Pelet S, Crowther M, MacDonald S, Kim P, Pleasance S, Davis N, Andreou P, Wells P, Kovacs M, Rodger MA, Ramsay T, et al. Aspirin versus lowmolecular-weight heparin for extended venous thromboembolism prophylaxis after total hip arthroplasty: a randomized trial. Ann Intern Med 2013; 158: 800–6.