CC BY-NC-ND
0Accepted Manuscript
A phase IV study of thromboembolic and bleeding events following hip and knee arthroplasty using oral factor Xa inhibitor
David Gomez, MD, PhD, Helen Razmjou, PhD, Andrea Donovan, MD, Vikas Bansal, MD, Jeffrey Gollish, BASc, MD, John J. Murnaghan, MD, MSc, MA
THE JOURNAL OF
ARTHROPLASTY
PII: S0883-5403(16)30664-7
DOI: 10.1016/j.arth.2016.09.021
Reference: YARTH 55413
To appear in: The Journal of Arthroplasty
Received Date: 3 February 2016 Revised Date: 8 September 2016 Accepted Date: 21 September 2016
Please cite this article as: Gomez D, Razmjou H, Donovan A, Bansal V, Gollish J, Murnaghan JJ, A phase IV study of thromboembolic and bleeding events following hip and knee arthroplasty using oral factor Xa inhibitor, The Journal of Arthroplasty (2016), doi: 10.1016/j.arth.2016.09.021.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A phase IV study of thromboembolic and bleeding events following hip and knee arthroplasty using oral factor Xa inhibitor
David Gomez, MD, PhDab, Helen Razmjou, PhDc, Andrea Donovan, MDe, Vikas Bansal, MDf, Jeffrey Gollish, BASc, MDgh, John J. Murnaghan, MD, MSc, MAb'gh
a) Division of General Surgery, Department of Surgery, University of Toronto, Canada
b) Sunnybrook Research Institute, Sunnybrook Health Sciences Centre, Toronto, Canada
c) Department of Rehabilitation, Sunnybrook Health Sciences Centre, Toronto, Canada
d) Musculoskeletal Research, Sunnybrook Health Sciences Centre, Toronto, Canada
e) Department of Diagnostic Imaging, Sunnybrook Health Sciences Centre, Toronto, Canada
f) Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, Canada.
g) Division of Orthopaedics, Department of Surgery, University of Toronto, Canada
h) Sunnybrook Health Sciences Centre, Holland Orthopaedic and Arthritic Center, Toronto, Canada
Address all correspondence to: John J Murnaghan, MD, 43 Wellesley Street East, Suite 319, Toronto, ON, M4Y 1H1. j ohn. m urn a gh an @su nnybrook.c a Phone: 416 9678778
Mailing Address
David Gomez, MD, PhD, c/o Avery Nathens, MD, Chief of Surgery, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON M4N 3M5.
Helen Razmjou, PhD, Holland Orthopaedic and Arthritic Centre, 43 Wellesley St East, Toronto, ON, M4Y 1H1.
Andrea Donovan, MD, Dept of Diagnostic Imaging, M-Ground, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Toronto, ON, M4N 3M5.
Vikas Bansal, MD, Dept of Medicine, Holland Orthopaedic and Arthritic Centre, 43 Wellesley St East, Toronto, ON M4Y 1H1.
Jeffrey Gollish, BASc, MD, Ste 315, Holland Orthopaedic and Arthritic Centre, 43 Wellesley St East, Toronto, ON M4Y 1H1.
John J. Murnaghan, MD, Ste 319, Holland Orthopaedic and Arthritic Centre, 43 Wellesley Ste East, Toronto, ON M4Y 1H1.
1 A phase IV study of thromboembolic and bleeding events following hip and knee arthroplasty
2 using oral factor Xa inhibitor
3 Abstract
5 Background: Multiple randomized controlled trials have documented the effectiveness of rivaroxaban in
6 the prevention of venous thromboembolism up to 1-month following total joint arthroplasty. However,
7 the effectiveness and safety of rivaroxaban in the real world setting, outside of the strict protocols utilized
8 by randomized clinical trials, is unknown.
10 Methods: Prospective, observational, non-interventional, phase IV study of 3,914 consecutive patients
11 whom underwent total joint arthroplasty from June 2010 to December 2012. Patients were treated with
12 rivaroxaban 10 mg by mouth daily starting post-operative day 1 and continued for 15 days. Participants
13 were followed up in clinic at 6 weeks and contacted by telephone at 12 weeks. The primary outcome of
14 interest was symptomatic venous thromboembolism; secondary outcomes included bleeding events,
15 transfusion requirements, and death.
17 Results: The incidence of symptomatic deep venous thrombosis at 3 months was 0.5% (n=18). Only one
18 deep venous thrombosis event occurred within 7 days of surgery. The incidence of symptomatic
19 pulmonary embolism at 3 months was 0.7% (n=28). Thirteen pulmonary embolisms (46%) occurred
20 within 7 days of surgery. The rate of major bleeding while on prophylaxis was 0.1%. Only 5% of patients
21 received a blood transfusion. No deaths were attributed to thromboembolic events.
23 Conclusions: This prospective observational phase IV study demonstrates that rivaroxaban appears to
24 protect patients against symptomatic pulmonary embolism and is not associated with major bleeding 2 5 events when used in a real world setting as described.
27 Key words: Arthroplasty, Xa inhibitor, venous thromboembolism, bleeding, transfusion.
29 Introduction
30 Four randomized controlled trials have documented the effectiveness of rivaroxaban compared
31 with enoxaparin in the prevention of radiographic deep venous thrombosis (DVT) or a composite
32 outcome of any DVT and non-fatal pulmonary embolism (PE) with acceptable levels of bleeding
33 following total hip arthroplasty (THA)1'2 and total knee arthroplasty (TKA)3'4. However, the effectiveness
34 of rivaroxaban in the real world setting, outside of the strict protocols utilized by randomized clinical
3 5 trials, is unknown. Furthermore, by design, these randomized controlled trials document radiographic and
36 not necessarily symptomatic rates of DVT and PE; thus, not reflecting real-world symptom driven
3 7 investigations and rates of venous thromboembolism.
38 Our purpose was to document the efficacy (i.e. symptomatic thromboembolic events) and safety
39 (i.e. bleeding events as well as deaths) following primary and revision THA and TKA in a real world
40 surgical setting using a locally developed 15-day dosing of rivaroxaban. The shorter duration of treatment
41 for THA was utilized because patients are mobilized weight-bearing as tolerated on postoperative day
42 (POD) 1 in most cases and there is less soft tissue dissection with smaller incisions.
44 Materials and Methods
45 This was a phase IV prospective, observational, non-interventional study. Phase IV studies are an
46 essential complement to tightly controlled randomized control trials as they supplement the previously
47 published efficacy data and characterize the true safety profile of medication in a naturalistic setting.
48 All patients underwent surgery in a high volume, academic arthroplasty centre in XXXX. All
49 patients 18 years or older scheduled to undergo either primary or revision THA and TKA between June
50 16, 2010 and December 14, 2012 were screened for contraindications for the use of rivaroxaban
51 according to manufacturer's guidelines. All eligible patients were approached and informed consent was
52 obtained. The study protocol was approved by the institutional Research Ethics Board.
54 Patient Data
55 Demographic data was abstracted from charts prospectively and included age, gender, body mass
56 index (BMI), and risk factors for thromboembolic disease (i.e. history of previous VTE, current platelet
57 count >500,000, BMI >30, major surgery within 3 months, cancer treatment within 3 months and being
58 wheelchair bound). Anaesthetic and surgical details including type of anaesthetic, use of regional blocks,
59 procedure time from surgical start to surgical closure, estimated blood loss (i.e. total suction volume plus
60 fluid from weighed sponges less irrigation volume), time to first dose anticoagulant, perioperative blood
61 values and date of discharge were collected.
63 Thromboprophylaxis
64 Patients were treated with rivaroxaban 10 mg by mouth daily starting POD 1 and continued for 15
65 days. Patients were not recruited for this study if they were pregnant, breast feeding, had active bleeding,
66 substantial liver disease (i.e. INR greater than 1.3 without anticoagulation), severe renal impairment (i.e.
67 glomerular filtration rate less than 30), concomitant use of protease inhibitors for the treatment of human
68 immunodeficiency virus infection or were on long term anticoagulation (i.e. for a condition that
69 rivaroxaban was not licensed in XXXX).
70 Only patients who had an epidural anesthetic (n=127, 3.2%) were treated with dalteparin or
71 enoxaparin while the epidural catheter was in-situ and started on rivaroxaban the day after the catheter
72 was removed, usually POD 2 or 3. No intermittent compressive devices or thromboembolic stockings
73 were used postoperatively. No intraoperative unfractionated heparin was used. Aspirin was not used
74 routinely. Patients were mobilized POD 1 weight-bearing as tolerated for primary arthroplasties. A small
75 number of cases were partial weight-bearing due to osteotomy, greater trochanter stabilization or with
76 elevated BMI following THA.
78 Surgery
79 Nine arthroplasty surgeons contributed cases to this study. Seven of the nine surgeons used a
80 direct lateral approach for their primary THA, one used an anterolateral approach and one used a posterior
81 approach. All cases done with lateral positioning. No intraoperative or postoperative blood salvage
82 techniques were used. Type of anaesthetic, surgical time and estimated blood loss is reported. All nine
83 surgeons used anteromedial incision for TKA. One surgeon used a lateral parapatellar approach for severe
84 valgus knees. All surgeons used a tourniquet. Five of nine surgeons deflated the tourniquet prior to
85 closure. All surgeons used a light compressive dressing with tensor bandages and a knee immobilizer.
86 There was no routine use of tranexamic acid or drains.
88 Follow up
89 Participants were followed up in clinic at 6 weeks where a self-report questionnaire was
90 completed. Patients were contacted by telephone at 12 weeks and asked the same questions as were on the
91 questionnaire at 6 weeks. For those who were lost to follow-up, searches of electronic medical records at
92 our institution and of a web-based application that gathers patients' results and information from various
93 other institutions in our region were undertaken. An online search for obituaries was undertaken for
94 patients whose questionnaires or electronic medical records were not available.
96 Outcomes
97 The primary outcome of interest was the incidence of symptomatic, proximal DVT and/or
98 symptomatic PE by the end of postoperative week 12. Doppler ultrasound (DUS) was used to diagnose
99 symptomatic, proximal DVT (i.e. occlusion or lack of flow in popliteal, femoral, and/or iliac veins).
100 Spiral computed tomography angiogram (CTA), angiogram, or ventilation-perfusion (V/Q) scans were
101 used to diagnose symptomatic PE (i.e. central or segmental filling defects). Sub-segmental filling defects
102 were not included in the definition of a positive PE. All DUS, CTA, angiogram, and V/Q scans were
103 reviewed by a radiologist to verify findings. All thromboembolic events detected were managed
104 according to local guidelines. Incidence of symptomatic DVT and PE was calculated from postoperative
105 weeks 0 to 12 and was displayed graphically in cumulative incidence plots. Incidence rates were stratified
106 by type of surgery.
107 Secondary outcomes of interest included major and non-major bleeding, transfusion and death.
108 Bleeding events were monitored for the full duration of follow-up. Categorization of bleeding events was
109 more stringent compared to the RECORD trials as the need for interventional radiology was considered a
110 major bleeding event. Bleeding events categorized as not clinically relevant are not reported. Previously
111 defined major and non-major bleeding events were reviewed by a hospitalist to confirm the severity of the
112 bleeding episode (Figure 1). Bleeding complications were documented as 'on prophylaxis' when they
113 occurred between 2 hours after the first dose of anticoagulant therapy until 24 hours after the 15 th dose.
114 Bleeding complications were categorized as surgical site and non-surgical site bleeding. In the case of
115 death, autopsy reports were obtained where available.
117 Statistical analysis
118 Means and standard deviations (SD) were calculated for continuous, normally distributed
119 variables. Medians and interquartile ranges (IQR) were calculated for continuous variables with a non-
120 normal distribution. Absolute and relative frequencies were measured for discrete variables. Baseline
121 patient characteristics were compared using %2 test and non-parametric methods, as appropriate. Relative
122 risk ratios were calculated for potential risk factors. Statistical analysis was performed using SAS®
123 version 9.3 (SAS® Institute, Cary, NC). Statistical results are reported using 2-tailed p values with
124 significance set atp < 0.05.
126 Results
127 From June 16, 2010 to December 14, 2012, 4,900 adult patients underwent total joint arthroplasty
128 and 4,152 (85%) agreed to participate in the study (Figure 2). Two hundred and thirty-eight patients were
129 excluded as they received an anticoagulant other than rivaroxaban. Therefore, data of 3,914 patients was
130 used for analysis. Only 74 patients (1.9%) did not complete the 12-week telephone questionnaire per
131 protocol. Of these, 62 were shown to have contact with the health care system with laboratory or x-ray
132 reports after the 3-month follow-up indicating they were alive. None of the remaining 12 (0.3%) patients
133 appeared in an online search for an obituary and were deemed to be likely alive at the completion of the
134 study period.
136 Study population
137 Baseline patient and operative characteristics across procedure types are displayed in Table 1.
138 Patients who underwent TKA represented 59% of the population (n=2,316) while those who underwent
139 THA represented the remaining 41% (n=1,598). The majority of patients were female (61%, n=2,381).
140 Average age was 65 years (SD 10.7, range 22-94). Almost half of the population had a BMI of 30 or more
141 (median 29.7, IQR 26.2 - 34.2).
142 The most common indication for surgery was osteoarthritis (85%, n=3,323). Spinal anesthesia
143 was the most common anesthetic technique used (94%, n=3,666) for THA and TKA. Duration of surgery
144 as well as estimated blood loss (EBL) was higher for THA than TKA. There were 2,024 primary
145 unilateral TKA (87%), 89 primary bilateral TKA (4%), 201 unilateral revision TKA (9%), and 2
146 unicompartmental TKA (<1%). There were 1,444 primary unilateral THA (90%), 7 primary bilateral
147 THA (<1%), and 147 unilateral revision THA (9%).
148 Half of patients received their first dose of rivaroxaban 22 hours after surgery. The interval to
149 first dose ranged from 6 to 113 hours (median 22, IQR 20 - 24 hours). The first dose of rivaroxaban was
150 given significantly sooner in patients who underwent TKA (median 22, IQR 19 - 24 hours) when
151 compared to those who underwent THA (median 24, IQR 21 - 25 hours, p<0.01).
153 Incidence and timing of symptomatic venous thromboembolism
154 There were a total of 18 symptomatic proximal DVTs at postoperative week 12 (0.46%) (Table
155 2). Only 1 case of DVT was identified within 7 days of surgery. Only 2 symptomatic proximal DVTs
156 (0.05%) were identified while patients were receiving rivaroxaban. The median time to occurrence of
157 DVT was postoperative day 45 (IQR 48 - 67). The majority of DVT events (89%, n=16) occurred after
158 rivaroxaban had been discontinued (Figure 3). Three of six patients (50%) with DVT after primary TKA
159 had a prior history of VTE compared with only 1 of 11 (9%) patients following primary THA. There was
160 a significant association for the occurrence of DVT in patients with previous history of VTE following
161 primary TKA with a RR= 2.39 (95% CI 0.82-6.99). There was no statistically significant association
162 between DVT occurrence and a prior history of VTE in primary THA (p=0.35).
163 There were a total of 28 symptomatic PEs at postoperative week 12 (0.7%). Thirteen PEs (46%)
164 occurred within 7 days of surgery (median POD 10, IQR 5 - 36). Age, sex, BMI, type of anesthesia and
165 duration of surgery did not differ between those that had a thromboembolic event and those who did not.
166 There was a trend between occurrence of PE and having had a major surgery within previous 3 months in
167 the primary THA (p=0.05). In the present study, there were no statistically significant associations
168 between PE and other risk factors in either TKA or THA.
169 The incidence of DVT at postoperative week 12 was significantly higher in THA compared to
170 TKA (0.75% n=12 vs. 0.26% n=6, p<0.03). However, the incidence of PE at postoperative week 12 did
171 not differ between THA and TKA patients (0.75%, n=12 vs. 0.69%, n=16, p=0.83) (Table 2).
173 Bleeding, transfusion requirements, and deaths
174 Thirty-six patients (0.9%) had clinically relevant bleeding events during the study period: 6 major
175 (0.15%) and 30 non-major (0.8%) (Table 2). Twenty-nine bleeding events (4 major and 25 non-major)
176 occurred while taking rivaroxaban (0.7%). The incidence of surgical site bleeding while on prophylaxis
177 was 0.5% (n=20). Three surgical site bleeding events were considered major: one leading to death in a
178 patient with previously undiagnosed Von Willebrand's disease and two leading to re-operation. Surgical
179 site bleeding was equivalent following THA compared with TKA (0.68% vs. 0.39%, p=0.37). The
180 incidence of non-surgical site bleeding while on prophylaxis was 0.28% (n=11); two of these bleeding
181 events were characterized as major. Upper gastrointestinal bleeding accounted for 64% of all non-surgical
182 site bleeding events while on prophylaxis. Non-surgical site bleeding while on prophylaxis did not vary
183 across procedure type.
184 Overall, 182 patients required blood transfusions (4.7%). Transfusion requirements were
185 significantly higher following THA compared to TKA (7% vs. 3%, p<0.01) (Table 2).
186 There were nine deaths during the study period (0.2%). No deaths were attributed to
187 thromboembolic causes. Two deaths occurred during the perioperative period. One cause of death was
188 considered to be undetermined after undergoing autopsy. There was no evidence of DVT, PE, MI, stroke
189 or bleeding. The second death was characterized as diffuse fatal hemorrhage in the setting of previously
190 undiagnosed Von Willebrand's disease (Table 3).
192 Discussion
193 This phase IV prospective, observational, non-interventional study focused on clinically relevant
194 outcome measures that an effective thromboprophylaxis program needs to balance: prevention of
195 symptomatic VTE versus major bleeding events. It would not be an acceptable trade-off for surgeons or
196 their patients if the effective thromboprophylactic drug resulted in excessive bleeding and increased
197 postoperative complications. This study demonstrates that rivaroxaban appears to protect patients against
198 symptomatic pulmonary embolism and is not frequently associated with major bleeding events when used
199 in a real world setting as described.
200 The overall symptomatic proximal DVT rate in 3,914 patients within 3 months of surgery was
201 0.5%. This compares with proximal DVT rates following THA between 0.1%-0.6% at 35 days post-
202 surgery in RECORD 1 and 21'2 and following TKA between 0.3%-1.1% at 15 days post-surgery in
203 RECORD 3 and 4 measured the day after completion of treatment 3'4. Using the same time intervals as the
204 RECORD trials, our DVT rate in THA at approximately 35 days following surgery was 0.3% and for
205 TKA at day 15 post-surgery was 0.04%. The DVT rate in TKA was significantly lower than the reported
206 rates from RECORD 3 and 4 because we were not investigating asymptomatic patients.
207 Treating all 1,444 primary THA patients for 35 days with rivaroxaban may have theoretically
208 prevented 1-3 symptomatic proximal DVT, as one occurred on POD 34 and two more on POD 36. Nine
209 DVT occurred after POD 35 representing 82% of DVT following THA. None of these patients had prior
210 history of VTE.
211 The RECORD 1 and 2 studies reported nonfatal PE rates of 0.1-0.3% for THA at 35 days post-
212 surgery1'2 having received 35 days of treatment on rivaroxaban. Our PE rate in THA at 35 days post-
213 surgery was 0.4% having received 15 days of treatment on rivaroxaban. The RECORD 3 and 4 studies
214 reported nonfatal PE rates of 0-0.4% following TKA at 15 days3,4 . Our PE rate for all TKA at day 15
215 post-surgery was 0.5%. The PE rates for THA and TKA are close to the rates reported in the RECORD
216 Trials. The overall PE rate (0.7%) reported in this study is affected by: (1) a relatively high incidence of
217 symptomatic PE following TKA within 7 days of surgery; and (2) the longer duration of follow-up
218 revealing late occurrence of PE following THA and TKA between 35 days and 90 days.
219 The frequent occurrence of PE following TKA within 7 days of surgery was striking but not
220 unique. Kerr and Linkins5 reported a perioperative PE rate of 4.6% following TKA and 0.4% following
221 THA in a retrospective review of 683 primary and revision THA and TKA using low molecular weight
222 heparin. Clayton et al6 reported 73% of PE events in their study following TKA occurred on or before day
223 5 following surgery. In our study, nearly 50% of the PE events occurred within 7 days of surgery and
224 11/13 (85%) of these PE followed TKA. The diagnosis of PE was only made with central or segmental
225 filling defects. Several authors have questioned whether there is really an increase in PE or has there been
226 an increase in our ability to image the pulmonary circulation with relatively non-invasive tests678.
227 A previous meta-analysis has suggested prior history VTE has a strong association with
228 postoperative VTE following THA and TKA with Odds Ratio of 10.7(CI 2.4-47.5, p=0.002)9. The
229 relative risk for development of DVT in primary TKA in this study was 2.39 (CI 0.82-6.99) in patients
230 with previous history of VTE. The relative risk for development of DVT following primary THA was not
231 increased. None of the late occurring DVT between week 6 and 12 had a prior history of VTE. The
232 relative risk for development of PE following THA and TKA were not increased.
233 The interval between the completion of surgery and the first dose of rivaroxaban was statistically
234 longer for THA (median 24, IQR 21 - 25 hours) versus TKA (median 22, IQR 19 - 24 hours, p<0.01).
235 This difference is likely due to local operating room scheduling where a THA is most commonly booked
236 as the first case. There were 2 PE (0.1%) events following THA and 10 PE (0.4%) following TKA within
237 7 days of surgery. The longer interval to first dose did not appear to give rise to more frequent events
238 following THA. It is not clear from this data whether TKA are at higher risk for PE despite having a
239 shorter interval to first dose of rivaroxaban.
240 There were 4 major bleeding events while on rivaroxaban, including 48 hours after the last dose
241 (0.1%). There were 2 major surgical site bleeds following THA (0.1%) and 2 following TKA (0.1%).
242 This compares with the rates of major bleeding reported in the RECORD trials following THA of 0.1243 0.3%1,2 and following TKA of 0.7%3,4. The combination of major and non-major bleeding following THA
244 was 1.1% compared with 3.2-3.3% in RECORD 1 and 2. The combination of major and clinically
245 relevant non-major bleeding following TKA was 0.6% compared with 3-3.2% in RECORD 3 and 4. The
246 incidence of major bleeding was less than previously reported in the RCTs. This may be due to the first
247 dose of rivaroxaban being given the morning after surgery and treatment lasting a total of 15 days for both
248 THA and TKA.
249 The overall transfusion rate in this study was 4.7%. The transfusion rate following primary
250 unilateral THA was 7%. The transfusion rate following THA in the RECORD studies were 40-55%1' 2.
251 The transfusion rate following TKA was 3%. The transfusion rate following TKA in the RECORD
252 studies was 41-51%3' 4. The large discrepancy in transfusion rates likely represents different transfusion
253 criteria and surgical practice. There was no routine use of cell saver, reinfusion drains, tranexamic acid or
254 drains in our cases. Tourniquets were routinely used for TKA and about half of the surgeons deflated the
255 tourniquet prior to wound closure.
256 The overall death rate was 0.3%. None of these deaths were attributed to thromboembolic events.
257 The RECORD trials reported death rates following THA of 0.2-0.3% and following TKA of 0-0.4%.
258 These death rates are not unexpected given the extent of the surgery, the age of the patients and the
259 medical co-morbidities in this surgical population.
261 Limitations
262 This study had a number of limitations. This is a phase IV observational study without a
263 comparison group. We can only comment on the outcome measures relative to previously reported
264 randomized control trials. Despite following up with 99.7% of 3,914 patients, the event rates for DVT,
265 PE, death, and bleeding were very low (<1%) and not suitable for rigorous additional regression analyses.
267 Conclusions
268 This prospective, observational, phase IV study of rivaroxaban at three month follow-up
269 capturing 99.7% of patients has shown: a symptomatic proximal DVT rate of 0.5%; symptomatic PE rate
270 of 0.7%; major bleeding rate of 0.1% while on prophylaxis; death rate of 0.3 %; and an overall
271 transfusion rate of 4.7% following primary and revision THA and TKA. Prior history of VTE was only 2 72 associated with symptomatic proximal DVT following primary TKA in this study. The occurrence of 273 VTE 5-12 weeks following THA may warrant some form of thromboprophylaxis for a longer duration. 2 74 Our data suggests that rivaroxaban appears to protect a real-world cohort of unselected prospective 275 patients from symptomatic VTE and not cause excessive major bleeding events following primary and 2 76 revision THA and TKA when used as described.
277 References
278 1. Eriksson BI, Borris LC, Friedman RJ, et al. Rivaroxaban versus enoxaparin for
279 thromboprophylaxis after hip arthroplasty. N Engl J Med 2008;358.26:2765-75.
280 2. Kakkar AK, Brenner B, Dahl OE, et al. Extended duration rivaroxaban versus short-term
281 enoxaparin for the prevention of venous thromboembolism after total hip arthroplasty: a double-
282 blind, randomised controlled trial. Lancet 2008;372:31-39.
283 3. Lassen, MR, Ageno W, Borris LC, et al. Rivaroxaban versus enoxaparin for thromboprophylaxis
284 after total knee arthroplasty. N Engl J Med 2008;358:2776-86.
285 4. Turpie AG, Lassen MR, Dacidson BL, et al. Rivaroxaban versus enoxaparin for
286 thromboprophylaxis after total knee arthroplasty (RECORD 4): a randomized trial. Lancet
287 2009;373:1673-80 Epub 2009 May 4
288 5. Kerr J, Linkins L-A. High Incidence of in hospital pulmonary embolism following joint
289 arthroplasty with dalteparin prophylaxis. Thromb Haemost 2010;103:123-8.
290 6. Clayton RAE, Gaston P, Watts AC, Howie CR. Thromboembolic disease after total knee
291 replacement: Experience of 5100 cases. The Knee 2009;16(1): 18-21.
292 7. Winters BS, Solarz M, Jacovides CL, et al. Over diagnosis of pulmonary embolism: evaluation of
293 a hypoxia algorithm designed to avoid this catastrophic problem. Clin Orthop Rel Res
294 2012;470:497-502.
295 8. Carrier M, Righini M, Wells PS, et al. Subsegmental pulmonary embolism diagnosed by
296 computed tomography: incidence and clinical implications. A systematic review and meta
297 analysis of the management outcome studies. J Thromb Haemost 2010;8:1716-22.
298 9. Zhang J, Chen Z, Zheng J, Breusch SJ, Tian J. Risk factors for venous thromboembolism after
299 total hip and total knee arthroplasty: a meta-analysis. Arch Orthop Trauma Surg 2015; 135:759300 772.
Total Knee Arthroplasty (n=2,316) Total Hip Arthroplasty (n=1,598)
Primary Unilateral Primary Bilateral Revision Unicom Primary Unilateral Primary Bilateral Revision
n=2,024 n=89 n=201 n=2 n=1,444 n=7 n=147
Gender
Male, n (%) 735 (36) 47 (52) 79 (39) 1 (50) 606 (42) 1 (14) 64 (44)
Female, n (%) 1,289 (64) 42 (48) 122 (61) 1 (50) 838 (58) 6 (86) 83 (56)
Average (SD) 66 (10) 62 (8) 67 (11) 48 (10) 65 (11) 55 (7) 65 (13)
Range 33-93 38-80 40-94 44-51 22-94 46-66 32-91
Median 65 61 66 47.5 66 52 65
Body Mass Index
Average (SD) 32 (7) 30 (5) 32 (6) 28 (0.23) 29 (6) 30 (7) 28 (5)
Range 17 - 67 19 - 45 18 - 54 28 16 - 57 18 - 39 17 - 51
Median 31 29 31 28 28 29 28
Previous VTE/DVT n=175 93 (5) 1 (1) 15 (7) 0 56 (4) 0 10 (7)
Present DVT, n (%) 3/90 (3) 0/1 0/15 1/53 (2) 0/10
Present PE, n (%) 1/90 (1) 0/1 0/15 0/53 0/10
Type of Anesthesia
General, n (%) 136 (7) 4 (5) 24 (12) 0 93 (6) 3 (43) 37 (26)
Spinal, n (%) 1,903(94) 84 (94) 175(89) 2 1,376(95) 5 (71) 117 (80)
Epidural, n (%) 9(<1) 68 (77) 8(4) 0 13(<1) 4 (57) 25 (17)
Duration of Surgery
Average (SD) 72 (22) 114 (31) 110 (42) 85 (6) 74 (18) 132 (16) 143 (67)
Median 71 108 104 85 73 135 132
Estimated Blood Loss
Average (SD) 124 (87) 155 (79) 195 (215) 100 (0) 374 (229) 579 (180) 927 (819)
Median 100 150 100 100 350 550 650
Pathology
OA, n (%) 1,932 (95) 87 (98) 0 0 1,298 (90) 6 (86) 0
Inflammatory, n (%) 45 (2) 2 (2) 0 0 32 (2) 0 0
Fracture, n (%) 0 0 0 0 10 (<1) 0 0
Table 1: Demographics of the sample including type of anaesthetic, duration of surgery and estimated blood loss by procedure
Total Knee Arthroplasty (n=2,316) Total Hip Arthroplasty (n =1,598)
Primary Primary Revision Unicom Primary Primary Revision
Unilateral Bilateral n=201 n=2 Unilateral Bilateral n=147
n=2,024 n=89 n=1,444 n=7
Deep Venous Thrombosis, n=18
Perioperative (POD <=7) 1 0 0 0 0 0 0
6 weeks (POD 8-42) 2 0 0 0 4 0 1
3 month (POD 43-90) 3 0 0 0 7 0 0
Total 6 (0.3%) 0 0 0 11 (0.8%) 0 1 (0.7%)
Pulmonary Embolism, n=28
Perioperative (POD<=7) 10 0 1 0 1 0 1
6 weeks (POD8-42) 2 1 0 0 4 0 1
3 months (POD43-90) 2 0 0 0 5 0 0
Total 14 (0.7%) 1 (1.1%) 1 (0.5%) 0 10 (0.7%) 0 2 (1.3%)
Death, n=9 3 (0.1%) 0 0 0 6 (0.4%) 0 0
Surgical Bleeding, n=22
Major, n=3 2 0 0 0 0 0 1
Non-major, n=19 5 1 1 0 9 1 2
Non-Surgical bleeding, n=14
Major, n=3 1 0 0 0 1 0 1
Non-major, n=11 6 0 0 0 5 0 0
Transfusion, n=160 42(2%) 15(17%) 10(5%) 0 54(3%) 2(27%) 37(26%)
Table 2: Thromboembolic Events, Deaths, Bleeding and Transfusions by Surgical Procedure.
Table 3. Deaths during study by procedure and adjudicated cause.
Type of Post operative Patient Age
surgery day
Sex BMI Autopsy
THA TKA THA THA TKA THA THA
14 18 38 45 45 61 63
61 65 84 59 67
F M F M F M M
36 26 40
No No No No No No No
Adjudicated cause
Undetermined. No PE, myocardial infarction, bleed or stroke
Fatal hemorrhage, previously undiagnosed Von Willebrand's disease
Drug toxicity
Cardiac failure
Pneumonia
Cardiac death
Cardiac death
Cardiac death
Metastatic cancer
1 Figure 1. Classification of bleeding events
Major bleeding
Non-major bleeding
Non-clinically relevant
Non-fatal
Non-fatal
Bleeding in a critical organ (e.g. retroperitoneal, intracranial, intraocular, intraspinal). Extra-surgical site bleeding that is clinically overt and is associated with a fall in hemoglobin of at least 2 g/dL
Bleeding that requires transfusion of 2 or more units of blood
Bleeding requiring reoperation or invasive intervention (i.e. including interventional radiology)
Bleeding which does not reach threshold for major bleeding
May require transfusion up to 1 unit of blood
Bleeding that does not require reoperation or invasive treatment
Bleeding which is mentioned in interdisciplinary notes but does not reach threshold for major or non-major bleeding
Does not require a blood transfusion
Bleeding that does not interfere with clinical course or require any intervention
3 Figure 2. Study participants and follow-up
Figure 3. Cumulative incidence plots of thromboembolic events by joint replaced (Hip and Knee)
Deep venous thrombosis
•£ 2-
10 20 30 40
Pulmonary embolism
-——f^
40 50 Days
Joint replaced - Hip -Knee
