Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement

Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement

Accepted Manuscript Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement – An Analysis of the Society...

489KB Sizes 0 Downloads 2 Views

Accepted Manuscript Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement – An Analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database Thomas A. Schwann, MD, MBA, Robert H. Habib, PhD, Rakesh M. Suri, MD, J. Matthew Brennan, MD, Xia He, MA, Vinod H. Thourani, MD, Milo Engoren, MD, Gorav Ailawadi, MD, Brian R. Englum, MD, Mark R. Bonnell, MD, James S. Gammie, MD PII:

S0012-3692(16)48651-4

DOI:

10.1016/j.chest.2016.04.015

Reference:

CHEST 445

To appear in:

CHEST

Received Date: 26 November 2015 Revised Date:

2 February 2016

Accepted Date: 1 April 2016

Please cite this article as: Schwann TA, Habib RH, Suri RM, Brennan JM, He X, Thourani VH, Engoren M, Ailawadi G, Englum BR, Bonnell MR, Gammie JS, Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement – An Analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database, CHEST (2016), doi: 10.1016/j.chest.2016.04.015. 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.

ACCEPTED MANUSCRIPT

SC M AN U

Thomas A. Schwann, MD, MBA1, Robert H.Habib, PhD2, Rakesh M. Suri, MD3, J. Matthew Brennan, MD4, Xia He, MA4, Vinod H. Thourani, MD5, Milo Engoren, MD6, Gorav Ailawadi, MD7, Brian R. Englum, MD4, Mark R. Bonnell, MD1, James S.Gammie, MD8

RI PT

Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement – An Analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database

EP

TE D

Institutional Affiliations: 1. University of Toledo, Department of Surgery, Toledo, OH. 2. American University of Beirut, Beirut, Outcomes Research Unit, Beirut, Lebanon. 3. Cleveland Clinic, Department of Surgery, Cleveland, OH. 4. Duke University and Duke Clinical Research Institute, Durham, NC. 5. Emory University, Department of Surgery, Atlanta, GA. 6. University of Michigan, Department of Anesthesiology, Ann Arbor, MI. 7. University of Virginia, Department of Surgery, Charlottesville, VA. 8. University of Maryland School of Medicine, Department of Surgery, Baltimore, MD.

AC C

Disclosures: none for all authors. Key Words: Mitral Valve Replacement, Warfarin Utilization, Society of Thoracic Surgeons Database. Corresponding Author: Thomas A. Schwann, University of Toledo, Toledo, OH 43614. [email protected] Total Words: 2,900 Study supported by institutional funds.

1

ACCEPTED MANUSCRIPT

Glossary of Abbreviations:

AC C

EP

TE D

M AN U

SC

RI PT

ACC – American College of Cardiology ADP - adenosine diphosphate AHA – American Heart Association ASA - Aspirin ASD – atrial septal defect AVR – aortic valve replacement BMVR - bioprosthetic mitral valve replacement CHF – congestive heart failure CI – confidence interval CVA – cerebrovascular accident GEE - generalized estimating equation IABP – intra aortic balloon pump GI – gastrointestinal MI - myocardial infarction NW – No warfarin OR – odds ratio RIND – reversible ischemic neurologic defect STS ACSD - Society of Thoracic Surgeons Adult Cardiac Surgery Database TE – thromboembolic TIA –transient ischemic attack TVR - tricuspid valve repair/replacement W - Warfarin

2

ACCEPTED MANUSCRIPT

Abstract:

RI PT

Background: Anticoagulation with warfarin following bioprosthetic mitral valve replacement (BMVR) is recommended by multiple practice guidelines. We assessed practice variability and patient characteristics associated with warfarin prescription following BMVR.

SC

Methods: 7,637 patients in the Society of Thoracic Surgeons Database (1/1/20086/30/2011) who were discharged following isolated, primary, non-emergent BMVR were analyzed. Patients requiring pre-op warfarin, those with pre-op atrial fibrillation or contra-indication to warfarin were excluded. The association between patient, hospital, and surgeon characteristics and warfarin prescription were evaluated.

TE D

M AN U

Results: 58.0% of this cohort (median age, 66; female, 58.7%), was prescribed warfarin. Warfarin patients were older (67vs65yrs, p<0.0001), were less likely to have preoperative: stroke (9.3%vs12.1%, p<0.001), CHF (51.4%vs54.1%, p<0.02) or dialysis (4.9%vs9.0%, p<0.001) and had a longer post-operative length of stay (8.0vs7.0 days, p<0.01). Warfarin was prescribed less often for patients with post-operative gastrointestinal events (44.4%vs55.6%, p< 0.001) but more often for patients with a postoperative myocardial infarction (75.8%vs24.2%, p<0.001), new atrial fibrillation (68.0%vs32.0%, p<0.001) and those requiring blood transfusions (intra-op (55.7%vs44.3%, p<0.001), post-op (57.0%vs43.0%, p<0.03). Similar rates of warfarin prescription were observed in patients requiring reoperation for bleeding (54.9%vs45.1%, p=0.20) and those with a post-operative stroke (53.6 %vs46.4 %, p=0.30). After adjusting for patient characteristics, significant surgeon and hospital variation in warfarin prescription at hospital was observed.

AC C

EP

Conclusions: Although patient characteristics and post-operative events may be associated with the prescription of warfarin following BMVR, substantial surgeon and hospital variability remains. This variability largely ignores the established practice guidelines, and warrants further study to define the optimal anticoagulation strategy in BMVR patients.

3

ACCEPTED MANUSCRIPT

Introduction

Despite lack of randomized evidence supporting the use of post-operative anticoagulation following bioprosthetic mitral valve replacement (BMVR), multiple

RI PT

practice guidelines (1-5) recommend (Class IIa) a two to three month post-operative

warfarin course as a means of minimizing thromboembolic (TE) complications. The goal of peri-operative warfarin therapy is to minimize thrombus formation on the raw surgical suture lines and exposed, non endothelialized prosthetic hardware as well mitigate the

SC

general peri-operative hypercoagulable state – elements that are thought to contribute to an elevated risk of early post-operative risk of TE (6-8). Although this rationale is

M AN U

logically appealing, the data supporting the efficacy of warfarin therapy in minimizing TE and the published guidelines are weak (Level of Evidence C). In light of this and given the increased risk of warfarin associated hemorrhagic complications, clinicians’ adherence to such guidelines is variable (9, 10). Moreover, the available reports on the benefit of peri-operative warfarin therapy are almost always based on relatively older, small, single institutional, retrospective analyses, plagued by lack of consistent

TE D

definitions of TE, non-comparable methodologies, variable follow up and lack meticulous adjustment for confounding factors. Thus, in the absence of compelling evidence based recommendations, surgeons default to empiric practice patterns in choosing appropriate anticoagulation strategy frequently resulting in therapeutic

EP

equipoise between warfarin use and non use (11,12). How, and if, the recent intriguing report of structural bioprosthetic aortic valve leaflet abnormalities and possible

AC C

subclinical leaflet thrombosis associated with the lack of peri-operative warfarin therapy in trans-catheter aortic valve replacement might impact the issue of peri-operative anticoagulation in surgical BMVR will deserve careful future attention. (13) In this study, we evaluated warfarin prescription practice patterns at hospital

discharge in a nationally representative, contemporary cohort of BMVR patients from the Society of Thoracic Surgeons Adult Cardiac Surgery Database (STS ACSD). Specifically, in patients undergoing BMVR we sought to 1) determine the rate of warfarin use in patients without absolute indications for or contra-indications against this treatment; 2) define patient characteristics associated with warfarin prescription and 3)

4

ACCEPTED MANUSCRIPT

determine the degree to which surgeon and hospital variation persisted after adjustment for differences in patient level covariates.

RI PT

Materials and Methods

The Society of Thoracic Surgery Adult Cardiac Surgery Database (STS ACSD) was used as the basis for this report. From a total of 65,711 patients undergoing mitral

SC

valve replacement [45,512 bioprosthetic (BMVR)], between January 1, 2008 and June 30, 2011, a subcohort of 7,637 patients undergoing isolated BMVR was used in the

M AN U

analysis. (Data Collection Form Version 2.61). Exclusion criteria were as follows: mechanical mitral valve replacement (20,199), emergent or salvage status (1,606), reoperative procedures (6,680), concurrent cardiac or non cardiac surgical procedures except for tricuspid valve repair/replacement (TVR) and atrial septal defect (ASD) repair (25,210), pre-operative (within 30 days of surgery) atrial fibrillation (2,595), warfarin use within 24 hours of surgery (528), absolute contraindication to warfarin use (737) and

TE D

missing patient data (87). Since the study endpoint was warfarin use at hospital discharge, all patients who died during the index hospitalization (422, 5.2%) were also excluded from the analysis.

EP

The primary endpoint was warfarin prescription at hospital discharge.

Variable Definitions

AC C

Warfarin prescription was defined at hospital dismissal consistent with the STS

Database definition (“Indicate whether the patient was discharged from the facility on Coumadin, or if it was contraindicated or not indicated”.) and the study population was divided into a warfarin (W) and no warfarin (NW) groups. The definitions for all data elements within the database followed standard STS definitions (version 2.61) and are available at: http://www.sts.org.

5

ACCEPTED MANUSCRIPT

Data Distributions of preoperative, intraoperative and postoperative patient variables, hospital factors including annual total cardiac surgical case volume, annual mitral valve

RI PT

procedures case volume (both repairs and replacements), annual isolated mitral valve replacement case volume, geographic region and academic program status were

summarized for both the W and NW groups. Similarly, distribution of surgeon factors

SC

including annualized surgeon specific cardiac surgery case volume, annualized surgeon

specific mitral valve procedures case volume (repairs and replacements) and annualized

M AN U

surgeon specific mitral valve replacement case volumes were recorded for both groups. For surgeons and hospitals that did not participate for any full year during the entire study period of 3.5 years, the data were imputed by dividing the cases of interest by the total months participated and then multiplying by 12. Wilcoxon rank-sum test and Pearson’s

TE D

χ2 test were used to compare the distributions between two groups for continuous and categorical variables, respectively.

EP

Statistical Analysis

Histograms of hospital and surgeon specific warfarin utilization rates were created.

AC C

To assess the association between warfarin utilization and patient characteristics,

a logistic regression model estimated by a generalized estimating equation (GEE) was created in conjunction with robust standard error estimates for coefficients. The model accounted for the within center correlations of patient characteristics. Patient variables within the STS ACSD were selected based on the entire clinically relevant patient data set included in the Data Collection Form Version 2.61, which are summarized in the e-

6

ACCEPTED MANUSCRIPT

Table 1. Sets of logistic models were fitted to estimate the association of various hospital and surgeon characteristics and warfarin utilization. Each of the models included one hospital or surgeon factor and all patient covariates. SAS version 9.2 and R version

RI PT

2.15.1 were used for all analysis.

Statements of Protocol Review and Investigator Responsibility. The Duke University

SC

School of Medicine Institutional Review Board granted a waiver of informed consent and

M AN U

authorization for this study. All authors have read and agree to the manuscript as written.

TE D

Results

Warfarin was prescribed at hospital discharge in 4,430 (58.0%) of 7,637 BMVR study patients (median age, 64.6; female, 58.7%) at 886 US hospitals, involving 1,774

EP

surgeons.

AC C

Patient and Operative Characteristics Associated with Warfarin Use Selected patient demographics and comorbidities are summarized in Table 1.

Operative characteristics and outcomes are shown in Table 2. Warfarin was prescribed in a similar proportion of women as men (57.9%vs58.1%, p=0.86). Compared to the NW group, the W group was slightly older (67vs65 years, p<0.001) and less likely to have a history of: smoking (19.7%vs24.5%, p<0.001), CVA (9.3%vs12.1%, p<0.001),

7

ACCEPTED MANUSCRIPT

cerebrovascular disease (14.3%vs17.1%, p<0.001), CHF (51.4%vs54.1%, p<0.02) or dialysis (4.9%vs9.0%, p<0.001). Pre-operative anticoagulation in the form of heparin or other agents was used in 13.7% of the study cohort and was similar in the W and NW

RI PT

groups. Aspirin or adenosine diphosphate (ADP) platelet receptor inhibitors were

prescribed at discharge in 76.8% and 8.8% of patients, respectively, and their use was

higher in the NW group (aspirin: 83.9%vs71.6%, p<0.001; ADP inhibitor: 14.2%vs4.9%,

SC

p<0.001). 515 (6.7%) patients received neither warfarin nor aspirin. Except for higher intra operative blood transfusions in the NW group, operative parameters were only

M AN U

slightly different between the two groups (Table 2).

Post Operative Factors

The in hospital mortality for the entire cohort was 5.2% (422 of 8,059). Post-

TE D

operative complications showed no significant differences between the W and NW groups except for the rate of MI and new onset post-op atrial fibrillation which were significantly higher in the in the W group [(W;2.2%, NW;1.0%, p<0.001), (W;41.3%,

EP

NW;26.9%, p<0.001), respectively]. (Table 2) Warfarin was prescribed less often in patients who had experienced postop GI

AC C

events (44.4%vs55.6%, p< 0.001), but was prescribed more commonly in patients who had experienced a postop MI (75.8%vs24.2%, p<0.001), new atrial fibrillation (68.0%vs32.0%, p<0.001) and who required blood transfusions (intraoperative (55.7%vs44.3%,p<0.001), postop (57.0%vs43.0%,p=0.03). There was no difference in warfarin prescription among patients who required reoperations (54.9%vs45.1%, p=0.20)

8

ACCEPTED MANUSCRIPT

or those with a postop stroke (53.6 %vs46.4 %, p=0.30). The median post-operative

RI PT

length of stay was longer in the warfarin group (8.0vs7.0 days, p<0.01).

Hospital and Surgeon Characteristics Associated with Warfarin Use

SC

The variation in discharge warfarin use was sizeable and significant among

hospitals (Fig. 1) and surgeons (Fig. 2), with a bimodal distribution noted in both when

M AN U

all institutions and surgeons were considered. Warfarin prescription varied regionally (Table 3) and was highest in the Northeast (68.9%, p=0.12) and lowest the South (49.7%, p=0.003) compared to the West. There was no difference among warfarin prescription in hospitals with (57.4%) or without (58.2%) a residency program (p=0.80). After adjusting for patient characteristics, neither hospital nor surgeon total

TE D

surgical case volumes nor mitral valve case volumes were associated with warfarin

EP

prescription (e-Table 2).

AC C

Multivariate Analysis

On multivariate analysis certain baseline patient characteristics and perioperative

outcomes were associated with warfarin prescription (Table 4). Perioperative MI, new onset post-op atrial fibrillation and beta blocker prescription at discharge were, among others, significantly associated with an increased odds of warfarin use. Conversely, pre operative dialysis, intra operative blood transfusion, intra operative IABP use, post operative GI events and aspirin at discharge were, among others, significantly associated

9

ACCEPTED MANUSCRIPT

with decreased warfarin prescription. Neither TVR nor ASD procedures were associated with of warfarin prescription (Table 4).

RI PT

Discussion:

Despite multiple practice guidelines (1-5) recommending perioperative

anticoagulation following bioprosthetic mitral valve replacement (BMVR) and older data

SC

suggesting a protective effect of warfarin against TE (8), the results of this analysis

indicate that less than sixty percent of eligible patients are prescribed warfarin therapy at

M AN U

discharge from North American hospitals. Of those patients discharged without warfarin, 34% were discharged on aspirin and 14% were discharged on ADP inhibitors. Although a number of patient and operative factors were weakly associated with warfarin prescription, the practice variability at both surgeon and hospital level were pronounced,

TE D

suggesting that physician and institution specific practice patterns predominantly influenced the decision to prescribe warfarin. Such disparate patterns of perioperative warfarin prescription are likely driven by the lack of convincing evidence for or against

EP

warfarin use in the current practice guidelines. The potential protective effects of peri-operative anticoagulation in BMVR

AC C

patients in minimizing thromboembolic (TE) risks must be weighed against its wellknown increased hemorrhagic complications as well as the increased complexity and cost of effective warfarin use. Wide variations in surgeon and institution specific warfarin practice patterns have been reported in relatively small, single institution reports, with rates ranging from 23% to 100% (6, 10, 14-17) with no appreciable impact on outcomes. More recently, Thourani and colleagues reported a 55% warfarin utilization rate

10

ACCEPTED MANUSCRIPT

following mitral valve repair or BMVR. Importantly they also found no statistically significant differences in the acute perioperative complications or long-term survival in patients undergoing mitral vale repair or BMVR with or without warfarin therapy in this

RI PT

single institutional retrospective analysis (18). Our finding of a 58% warfarin prescription is in line with these observations. Following adjustment for baseline patient factors, we

identified a similar bimodal “very few or almost all” approach among surgeons, and thus

SC

hospitals, to anti coagulation following BMVR as was previously reported in mitral valve repair patients (11). In patients receiving an aortic bioprostheses, Brennan (12), reported

M AN U

only a 35% warfarin utilization rate in a large series of contemporary patients over the age of 65, with no difference in mortality, hemorrhage or thrombotic complications between patients treated with aspirin only compared to warfarin only within 3 months of surgery. Relative to the patients discharged on aspirin only, patients discharged on

of bleeding.

TE D

warfarin and aspirin, had a lower risk of mortality and thromboembolism, but higher risk

Despite the concern regarding thromboembolic complications, the TE risk

EP

following BMVR has not been well defined. Multiple small, retrospective studies, using variable anticoagulation regimens, reported a TE risk following BMVR between 0 to

AC C

3.4%/patient year (8,9,15,17,19,20). Most, but not all, of these same investigators also found that the TE risk was higher among BMVR patients with a history of TE, atrial fibrillation, a hypercoagulable state, increased age, hypertension, left ventricular dysfunction and increased left atrial size and it is in these patients that warfarin utilization maybe of maximum value. In line with these observations, we also identified that

11

ACCEPTED MANUSCRIPT

increasing age, lower ejection factor and post operative atrial fibrillation are associated with increased warfarin use at hospital discharge. It is commonly assumed that the hazard function for TE post any MV operation is

RI PT

greatest in the immediate post-operative period and declines thereafter coincident with the endothelialization of fresh suture lines and resolution of the inflammatory and the

hypercoagulable peri-operative milieu. Logic dictates therefore, that the protective effects

SC

warfarin would be most prominent during this same vulnerable period. Indeed, such a time dependent TE risk has been confirmed (6-8), but whether perioperative warfarin

M AN U

therapy has a protective effect remains unclear, as Nunez et al reported very low TE rates among BMVR patients treated exclusively with aspirin (17).

Our analysis revealed that only 76.8% of the entire BMVR cohort received ASA, with a significantly higher ASA use in the NW group (NW;83.9%, W;71.6%, p< 0.001).

TE D

Although aspirin use carries a Class IIa, recommendation (Level of Evidence B) in the 2014 AHA/ACC Guidelines (1), Brennan et al reported a significantly higher rate of post operative bleeding complications with combined warfarin and aspirin use compared to

EP

only warfarin use following bioprosthetic aortic valve replacement (12). The increased adverse effects of ASA therapy compared to warfarin in the elderly (20,21) may possibly

AC C

have impacted the hesitancy to prescribe ASA more frequently by STS surgeons in patients undergoing BMVR. 6.7% of the study group was discharged with neither aspirin nor warfarin. Colli (16) reported a 13% rate of neither antiplatelet agent nor warfarin use following BMVR with no appreciable difference in outcomes. In contradistinction, Aramendi (18) identified a lower TE rate with exclusive antiplatelet therapy compared to warfarin. The reasons behind these diverse outcomes are unclear.

12

ACCEPTED MANUSCRIPT

In the current study, we found that neither surgeon case volume nor the presence of an institutional residency program were associated with warfarin prescription, while statistically significant regional variation in warfarin prescription was noted with the

RI PT

highest warfarin use in the Northeast (66.4%) and the lowest in the South (47.8%).

Regional warfarin prescription variability was also reported among patients undergoing isolated mitral valve repair (11), highlighting the complex regional biases underlying

SC

these practice patterns.

To better understand the reasons behind the differential warfarin prescription

M AN U

patterns, we evaluated patient demographics and peri-operative factors that may impact warfarin prescription. Certain patient and operative characteristics were indeed correlated with warfarin use, albeit weakly. Despite the increased challenges of warfarin management in the elderly, warfarin was prescribed to BMVR patients that were

TE D

significantly older (67vs65,p<0.001) than their NW counterparts, a finding reported we previously (14). Increasing operative complexity as evidenced by concurrent tricuspid valve procedures or atrial septal defect surgery, was not associated with increased

EP

warfarin prescription. New onset perioperative atrial fibrillation, whether episodic, recurrent or persistent, developed in 32.5% of study patients. Given the association of

AC C

atrial fibrillation with stroke, predictably, warfarin was prescribed more frequently in this group (68.0%vs32.0%, p<0.001), and was associated with warfarin prescription (OR 2.06, p=0.004). Suri et al reported a similar observation following mitral valve repair (11). Expectedly, the incidence of anti arrhythmic use was higher in our warfarin group (36.1% vs 27.6%, p<0.001). The STS ACSD is not granular enough to allow determination whether the atrial fibrillation was paroxysmal or persisted to hospital

13

ACCEPTED MANUSCRIPT

discharge - a clear indication for warfarin use. Thus at least a portion of warfarin prescription post BMVR, maybe driven by arrhythmic considerations rather than concerns of bioprosthesis associated TE risk.

RI PT

Whether the differential warfarin use meaningfully impacts long-term outcomes is beyond the scope of the current report, given that STS ACSD focuses on short term outcomes. Furthermore, given the inability of the STS database to capture the

SC

chronologic relationship between outcomes and initiation of warfarin use, we are unable to ascribe differences in clinically important outcomes to the variability in warfarin

M AN U

prescription. Such an assessment is urgently needed to guide clinicians’ optimal warfarin use following BMVR.

The limitations of this analysis relate to its retrospective nature and the inability to establish a chronologic relationship between outcomes and initiation of warfarin use.

TE D

Although the STS database has been well validated, warfarin prescription has not been a data field that has been routinely audited for accuracy. It is also unknown what number of patients started on warfarin post–operatively developed a complication requiring its

EP

discontinuation prior to discharge. Thus our analysis likely underestimates the true warfarin intention to treat rate. Although multivariate analysis was performed to identify

AC C

factors associated with warfarin use, confirmatory modeling of predictors of warfarin utilization by splitting the study group into isolated “derivation” and “validation” cohorts may be an additional valuable analysis, but was not performed due to the relatively small size of the entire study group. Finally, it is conceivable that by excluding patients that did not survive the index hospitalization, the true warfarin utilization rate was either over or underestimated.

14

ACCEPTED MANUSCRIPT

Conclusion: Our analysis demonstrates that post operative warfarin therapy is highly variable reflecting the uncertainty of the established benefit of warfarin use in patients with

RI PT

isolated BMVR, with practice patterns at both surgeon level and hospital level reflective of diametrically opposite “two schools of thought” regarding warfarin therapy resulting in its prescription for most or very few BMVR patients with an overall warfarin

SC

utilization rate of only 58%. Our data reveal that warfarin prescription is linked to what

appears to be institutional/surgical protocols more than any specific patient characteristic.

M AN U

Randomized trials aimed to clarify the impact of warfarin therapy on post-operative outcomes are needed to inform appropriate anticoagulation therapy and optimize postoperative outcomes. Alternatively, modifying the data collection tool to allow determination of the chronologic relationship between peri-operative warfarin use and

TE D

complication maybe useful in establishing the optimal anticoagulation regimen. The results of this analysis establishing significant warfarin prescription variability and identifying certain patient characteristics that are weakly associated with warfarin use,

EP

might be helpful in formulating a well defined prospective study.

AC C

ACKNOWLEDGMENTS;

Thomas A. Schwann; Responsible for the integrity of data, study design and results Robert H. Habib; Responsible for the integrity of data, study design and results The following investigators were responsible for editing, reviewing of data and generating final manuscript/ Rakesh M. Sur; J. Matthew Brennan; Xia He; Vinod H. Thourani;

15

ACCEPTED MANUSCRIPT

RI PT

Milo Engoren; Gorav Ailawadi; Brian R. Englum Mark R. Bonnell James S.Gammie

References:

M AN U

SC

1. AHA/ACC 2014 Guidelines for the management of patients with valvular heart disease. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. JACC 2014;63:e57-185. 2. Whitlock, R.P, Sun JC, Fremes SE, Rubens FD, Teoh KH. Antithrombotic and thrombolytic therapy for valvular disease: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 2012;141(2 Suppl):e576S600S.

TE D

3. The Task Force on the Management of Valvular Heart Disease of the European Society of Cardiology. Guidelines on the management of valvular heart disease. European Heart Journal 2007;28:230-268. 4. Asopa S, Patel A, Dunning J. Is short-term anticoagulation necessary after mitral valve repair. Best evidence topic. Interact CardioVasc Thorac Surg 2006;5:76166.

AC C

EP

5. Butchart EG, Gohlke-Barwolf C, Antunes MJ, Tornos P, et al on behalf of the Working Groups on Valvular Heart Disease, Thrombosis, and Cardiac Rehabilitation and Exercise Physiology, European Society of Cardiology. Recommendations for the management of patients after heart valve surgery. European Heart Journal 2005;26:2463-71. 6. Russo A, Grigioni F, Avierinos JF, Freeman WK, Suri R, Michelena H, Brown R, Sundt TM, Enriquez-Sarano M. Thromboembolic complications after surgical correction of mitral regurgitation: incidence, predictors, and clinical implications. J Am Coll Cardiol 2008;51:1203-1211. 7. Orszulak TA., Schaff HV, Pluth JR, Danielson GK, Puga FJ, Ilstrup DM, Anderson BJ. The risk of stroke in the early post-operative period following mitral valve replacement. Eur J Cardiothorac Surg 1995; 368:1005-11.

16

ACCEPTED MANUSCRIPT

8. Heras M, Chesebro JH, Fuster V, Penny WJ, Grill DE, Bailey KR, Danielson GK, Orszulak TA, Pluth JR, Puga FJ, Schaff HV, Larsonkeller JJ. High risk of thromboemboli early after bioprosthetic cardiac valve replacement. J Am Coll Cardiol 1995; 25:1111-9.

RI PT

9. Edmunds Jr LH. Thrombotic and bleeding complications of prosthetic heart valves. Ann Thorac Surg 1987;44:430-445.

SC

10. Vaughan P, Waterworth PD. An audit of anticoagulation practice among UK cardiothoracic consultant surgeons following valve replacement/repair. J Heart Valve Dis 2005;14:576-582.

M AN U

11. Suri RM, Thourani VH, He X, Brennan JM, O’Brien SM, Rankin JS, Schaff HV, Gammie JS. Variation in warfarin thromboprophylaxis after mitral valve repair: does equipoise exist and is a randomized trial warranted. Ann Thorac Surg 2013; 95:1991-1999. 12. Brennan JM, Edwards FH, Zhao Y, O’Brien S, Booth ME, Dokholyan RS, Douglas PS, Peterson ED, on behalf of the DEcIDE AVR Research Team. Early anticoagulation of bioprosthetic aortic valves in older patients. J Am Coll Cardiol 2012;60:971-7.

TE D

13. Makkar RR, Fontana G, Jilaihawi H, Chakravarty T, Kofoed KF, de Backer O, Asch Fm, Ruiz CE et al. Possible subclinical leaflet thrombosis in bioprosthetic aortivc valves. N Engl J Med 2015; 373;2015-24.

EP

14. Schwann TA, Engoren M, Bonnell M, Clancy C, Khouri S, Kabour A, Jamil T, Habib RH. Mitral valve repair and bioprosthetic replacement without perioperative anticoagulation does not increase the risk of stroke or mortality. Eur J Cardiothorac Surg 2013;44:24-31.

AC C

15. Ruel M, Masters RG, Rubens FD, Bedard PJ, Pipe AL, Goldstein WG, Hendry PJ, Mesana TG. Late incidence and determinants of stroke after aortic and mitral valve replacement. Ann Thorac Surg 2004;78:77-84.

16. Colli A, Verboye JP, Leguerrier A, Giberli T. Anticoagualtion or antiplatelet therapy of bioprosthetic heart valve recipients: an unresolved issue. Eur J Cardiothorac Surg 2007;31:573-577. 17. Nunez L, Aguado MG, Larrea JL, Celemin D, Oliver J. Prevention of thromboembolism using aspirin after mitral valvereplacement with porcine bioprosthesis. Ann Thorac Surg 1984;37:84-87.

17

ACCEPTED MANUSCRIPT

18. Thourani VH, Gunter Rl, Hurst S, Kilgo P, Padala M, Puskas JD, Lattouf OM, Halkos ME, Guyton RA. Postoperative warfarin following mitral valve repair or bioprosthetic valve replacement. J Heart Valve Dis 2013;22:716-23.

RI PT

19. Aramendi JL, Agredo J, Llorente A, Larrarte C, Pijoan J. Prevention of throboembolism with ticlopidine shortly after valve repair or replacement with bioprosthesis. J Heart Valve Dis 1998;7:610-14.

SC

20. Eberlein U, von der Emde J, Rein J, Esperer HD. Thromboembolic and bleeding complications after mitral valve replacement. Eur J Cardiothorac Surg 1990; 4:605-612.

M AN U

21. Rash, A., Downes T, Portner R, et al. A randomised controlled trial of warfarin versus aspirin for stroke prevention in octogenarians with atrial fibrillation (WASPO). Age and Ageing, 2007;36(2):151-156.

AC C

EP

TE D

22. Mant, J., Hobbs FDR, Fletcher K, et al., for the BAFTA Investigators, the Midland Research Practices Network (MidReC). Warfarin versus aspirin for stroke prevention in an elderly community population with atrial fibrillation (the Birmingham Atrial Fibrillation Treatment of the Aged Study, BAFTA): a randomised controlled trial. The Lancet;370:493-50.

18

ACCEPTED MANUSCRIPT

Table 1. Patient Demographics and Comorbidities

Warfarin (n=4,430)

%

%

Demographics

Male

3,148

41.2

1,822

Female

4,485

58.7

2,605

Risk Factors Body Surface Area, m2* Chronic Lung Disease Cerebrovascular Disease CVA TIA Preoperative Arrhythmia Diabetes Mellitus

1.9 2,141 1,185 799 282

1,712

Ejection Fraction Preoperative IABP

Dialysis

58.8

1,880

58.6

1.9

0.86

0.84

4,430

28

3,207

28

0.84

15.5

635

14.3

550

17.1

0.001

10.5

411

9.3

388

12.1

<0.001

3.7

170

3.8

112

3.5

0.43

5.1

243

5.5

149

4.6

0.10

963

21.7

749

23.4

0.10

22.4

55 (50-63)

55 (50-63)

0.51

70.7

3,134

70.7

2,268

70.7

0.99

195

2.6

111

2.5

84

2.6

0.76

1,659

21.7

873

19.7

786

24.5

<0.001

Severe

5,987

78.4

3,497

78.9

2,490

77.6

0.17

Severe

523

6.8

302

6.8

221

6.9

0.30

4,009

52.5

2,275

51.4

1,734

54.1

0.018

669

8.8

370

8.4

299

9.3

0.14

106

1.4

56

1.3

50

1.6

0.28

505

6.6

217

4.9

288

9

<0.001

AC C

Mitral Insufficiency Tricuspid Insufficiency Congestive Heart Failure Peripheral Vascular Disease Cardiogenic Shock

41.3

5,402

EP

Current or Recent Smoker

1,326

28

55 (50-63)

Hypertension

Pvalue

<0.001

41.1

1.9

TE D

392

%

63.2 (54.075.0)

M AN U

Gender

65.5 (58.076.0)

SC

64.6 (56.075.0)

Age (years)

No Warfarin (n=3,207)

RI PT

Overall (n=7,637)

Preop Medications Beta Blockers

4,389

57.5

2,592

58.5

1,797

56

0.031

ACE/ARB Inhibitors

2,950

38.6

1,722

38.9

1,228

38.3

0.59

Anticoagulants

1,048

13.7

597

13.5

451

14.1

0.47

Aspirin

3,190

41.8

1,815

41

1,375

42.9

0.10

19

ACCEPTED MANUSCRIPT

Table 2. Operative Characteristics and Post Operative Outcomes:

Atrial Septal Defect Repair Tricuspid Valve Procedures Intraoperative IABP Blood Products Used IntraOperatively Red Blood Cells Used IntraOperatively

%

Pvalue

n=4430

n=3207

95 (66-113)

96 (67-115)

94 (65-110)

130 (92153) 287 845 185

3.8 11.1 2.4

131 (93155) 168 0 83

3.8 11.2 1.9

128 (90150) 119 348 102

3.7 10.8 3.2

0.85 0.50 <0.001

4650

60.9

2589

58.4

2061

64.3

<0.001

3955

51.8

2206

49.8

1749

54.5

<0.001

19

0.2

12

0.3

7

0.2

0.65

5

208

4.7

171

5.3

0.1977

1.7 1.4 0.3 0.4 2.6 35.2 2.3

97 59 20 16 88 1830 98

2.2 1.3 0.5 0.4 2 41.3 2.2

31 51 6 17 110 862 79

1 1.6 0.2 0.5 3.4 26.9 2.5

<0.001 0.34 0.051 0.26 <0.001 <0.001 0.46

379 128 110 26 33 198 2692 177

TE D

AC C

Discharge Meds Antiarrhythmics Aspirin ADP Inhibitors Beta Blockers Lipid Agents

No Warfarin

n=7637

10.1 8

EP

POST OPERATIVE OUTCOMES Reoperation for Valvular Dysfunction Reoperation for Bleeding/Tamponade Perioperative MI Permanent Stroke Transient Ischemic Attack Tamponade GI Event Atrial Fibrillation Anticoagulant Event Post-Op Length of Stay Mean Median Blood Products Used PostOperatively Red Blood Cells Used PostOperatively

%

SC

Perfusion Time (mnts)

Warfarin

M AN U

OPERATIVE FACTORS Aortic Occlusion Time (mnts)

%

RI PT

Total

Variable

9.8 8

<0.001 0.001

10.6 7

<0.001

4626

60.6

2638

59.5

1988

62

0.028

4265

55.8

2420

54.6

1845

57.5

0.012

2485 5866 673 5286 4039

32.5 76.8 8.8 69.2 52.9

1601 3174 217 3151 2397

36.1 71.6 4.9 71.1 54.1

884 2692 456 2135 1642

27.6 83.9 14.2 66.6 51.2

<0.001 <0.001 <0.001 <0.001 0.013

20

ACCEPTED MANUSCRIPT

Table 3. Warfarin Utilization by Region and Residency Program Discharged on Warfarin

Total Number Of BMVR Cases

Warfarin on Discharge, mean % (95% CI, %)

Midwest

257

1,092

1,882

58.0 (55.8 60.3)

Northeast

122

974

1,414

68.9 (66.4 -71.3)

South

316

1,372

2,762

West

191

992

1,579

No

797

3,499

6,014

Yes

89

931

1,623

M AN U 58.2 (56.9 59.4) 57.4 (54.9 59.8)

AC C

EP

TE D

Residency Program

49.7 (47.8 51.6) 62.8 (60.4 65.2)

0.88 (0.63 125) 1.34 (0.93 1.95) 0.62 (0.45 0.85)

SC

Region

OR (95% CI) (vs West)

RI PT

Number of Hospitals

21

0.95 (0.65 1.40)

P value

0.48

0.12

0.003

0.80

ACCEPTED MANUSCRIPT

Table 4. Selected estimated odds ratio (OR) of warfarin prescription at hospital discharge and 95% confidence intervals for patient level factors from the multivariate logistic regression model. Study Variable

OR (95% CI)

RI PT

Factors Associated with Warfarin

P-value

2.06 (1.26-3.36)

0.004

Postop atrial fibrillation

1.90 (1.65-2.19)

<0.001

Postop TIA or RIND

1.44 (0.69-3.05)

0.33

Discharge beta blocker

1.32 (1.16-1.49)

<0.001

Tricuspid valve repair

1.13 (0.89-1.44)

0.30

Age increase by 10 (in < 65 years old)

1.17 (1.08-1.26)

<0.001

Ejection fraction decrease by 5

1.05 (0.83-1.09)

0.04

Female (vs Male)

1.00 (0.95-1.06)

0.99

SC

Perioperative MI

Atrial septal defect repair Reoperation for bleeding/tamponade Tricuspid valve replacement CHF NYHA IV Intraop use of red blood cell Intraop use of blood product Discharge aspirin Intraop IABP

AC C

EP

Postop GI event

TE D

Preop on dialysis

M AN U

Factors Associated with No Warfarin

22

0.97 (0.75-1.26)

0.82

0.86 (0.67-1.10)

0.24

0.86 (0.44-1.65)

0.64

0.77 (0.61-0.97)

0.027

0.78 (0.67-0.90)

<0.001

0.73 (0.59-0.89)

<0.001

0.63 (0.47-0.83)

<0.001

0.43 (0.37-0.49)

<0.001

0.49 (0.33-0.74)

<0.001

0.46 (0.33-0.64)

<0.001

ACCEPTED MANUSCRIPT

Figure 1. Distribution of warfarin use across hospitals; A. All hospitals (n=886) B.

RI PT

Hospitals with over 10 BMVR cases annually (n=245)

Figure 2. Distribution of discharge warfarin rates across surgeons; A. All surgeons

AC C

EP

TE D

M AN U

SC

(n=1774), B. Surgeons with over 10 MVR cases annually (n=172)

23

ACCEPTED MANUSCRIPT

AC C

A.

EP

TE D

M AN U

SC

RI PT

Figure 1. Distribution of warfarin use across hospitals; A. All hospitals (n=886) B. Hospitals with over 10 BMVR cases annually (n=245)

B.

ACCEPTED MANUSCRIPT

AC C

A.

EP

TE D

M AN U

SC

RI PT

Figure 2. Distribution of discharge warfarin rates across surgeons; A. All surgeons (n=1774), B. Surgeons with over 10 MVR cases annually (n=172)

B.