Relation of Mitral Valve Surgery Volume to Repair Rate, Durability, and Survival

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 69, NO. 19, 2017 ª 2017 THE AUTHORS. PUBLISHED BY ELSEVIER ON BEHALF OF THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION. THIS IS AN OPEN ACCESS ARTICLE UNDER ISSN 0735-1097 http://dx.doi.org/10.1016/j.jacc.2017.02.026 THE CC BY-NC-ND LICENSE (http://creativecommons.org/licenses/by-nc-nd/4.0/). Relation of Mitral Valve Surgery Volume to Repair Rate, Durability, and Survival Joanna Chikwe, MD,a,b Nana Toyoda, MD,a Anelechi C. Anyanwu, MD,a Shinobu Itagaki, MD, MSC,a Natalia N. Egorova, PHD,c Percy Boateng, MD,a Ahmed El-Eshmawi, MD,a David H. Adams, MDa ABSTRACT BACKGROUND Degenerative mitral valve repair rates remain highly variable, despite established benefits of repair over replacement. The contribution of surgeon-specific factors is poorly defined. OBJECTIVES This study evaluated the influence of surgeon case volume on degenerative mitral valve repair rates and outcomes. METHODS A mandatory New York State database was queried and 5,475 patients were identified with degenerative mitral disease who underwent mitral valve operations between 2002 and 2013. Mitral repair rates, mitral reoperations within 12 months of repair, and survival were analyzed using multivariable Cox modeling and restricted cubic spline function. RESULTS Median annual surgeon volume of any mitral operations was 10 (range 1 to 230), with a mean repair rate of 55% (n ¼ 20,797 of 38,128). In the subgroup of patients with degenerative disease, the mean repair rate was 67% (n ¼ 3,660 of 5,475), with a range of 0% to 100%. Mean repair rates ranged from 48% (n ¼ 179 of 370) for surgeons with total annual volumes of #10 mitral operations to 77% (n ¼ 1,710 of 2,216) for surgeons with total annual volumes of >50 mitral operations (p < 0.001). Higher total annual surgeon volume was associated with increased repair rates of degenerative mitral valve disease (adjusted odds ratio [OR]: 1.13 for every additional 10 mitral operations; 95% confidence interval [CI]: 1.10 to 1.17; p < 0.001); a steady decrease in reoperation risk until 25 total mitral operations annually; and improved 1-year survival (adjusted hazard ratio: 0.95 for every additional 10 operations; 95% CI: 0.92 to 0.98; p ¼ 0.001). For surgeons with a total annual volume of #25 mitral operations, repair rates were higher (63.8%; n ¼ 180 of 282) if they operated in the same institution as a surgeon with total annual mitral volumes of >50 and degenerative mitral valve repair rates of >70%, compared with surgeons operating in the other institutions (51.3%; n ¼ 580 of 1,130) (adjusted OR: 1.79; 95% CI: 1.24 to 2.60; p < 0.001). CONCLUSIONS This study suggests that individual surgeon volume is a determinant of not only mitral repair rates, but also freedom from reoperation, and survival. The data from this study support the guideline’s concept of reference referral to experienced mitral surgeons to improve outcomes in patients with degenerative mitral valve disease. (J Am Coll Cardiol 2017;69:2397–406) © 2017 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M itral valve repair is favored over valve prolapse (1,2). Both U.S. and European guidelines replacement for the treatment of severe strongly recommend valve repair whenever possible, mitral valve regurgitation in patients who and they also emphasize the importance of a durable have degenerative valve disease with mitral valve and long-lasting repair, particularly when intervening From the aDepartment of Cardiovascular Surgery, Icahn School of Medicine at Mount Sinai, New York, New York; bDepartment of Surgery, Stony Brook University Hospital, Stony Brook, New York; and the cDepartment of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, New York. The Icahn School of Medicine at Mount Sinai receives royalty Listen to this manuscript’s payments from Edwards Lifesciences and Medtronic for intellectual property related to Dr. Adams’ involvement in the devel- audio summary by opment of 2 mitral valve repair rings and 1 tricuspid valve repair ring. None of the sponsoring organizations had any role in the JACC Editor-in-Chief design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or Dr. Valentin Fuster. approval of the manuscript; or decision to submit the manuscript for publication. Dr. Adams is the National Co-Principal Investigator of the CoreValve United States Pivotal Trial, which is supported by Medtronic. Dr. Chikwe received speaker honoraria from Edwards Lifesciences. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Marc Gillinov, MD, served as Guest Editor for this paper. This paper was simultaneously presented at the American Association for Thoracic Surgery Centennial, Boston, Massachusetts, in May 2017. Manuscript received July 8, 2016; revised manuscript received January 12, 2017, accepted February 1, 2017. 2398 JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes ABBREVIATIONS on an asymptomatic patient (1,2). Mitral valve ventricle and patients with prior replacement or AND ACRONYMS replacement unfortunately remains relatively repair of any valve, prior coronary revascularization, common in patients with degenerative valve prior heart transplantation, and prior ventricular disease, and studies have highlighted wide assist device placement were also excluded. Addi- variation in mitral surgical volume and repair tionally, we excluded patients with out-of-state resi- rates when looking at “all-comer” study dency to ensure complete follow-up using the groups (3–10). The influence of mitral surgical Statewide Planning and Research Cooperative Sys- volume on perioperative mortality and repair tem. Patients who did not have identifiable surgeons rates is becoming established, but less is were also excluded (Online Figure 1, Online Table 1 CI = confidence interval HR = hazard ratio ICD-9-CM = International Classification of Diseases-Ninth Revision-Clinical Modification OR = odds ratio known regarding the impact of volume on 1-year sur- [ICD-9-CM codes]). vival and the need for early reoperation. Most large identified using Baseline comorbidities database reports have also included a broad range of codes from the index hospitalization and all di- causes of mitral valve disease. agnoses from hospitalizations before the index hos- present-on-admission were diagnosis pitalization (Online Table 2). We compared repair SEE PAGE 2407 rates, long-term survival, and risk of post-repair reoperation in the subgroup of patients with degen- To understand the effect of case volume on clinical erative disease according to total annual surgeon outcomes in patients likely to have isolated degen- volume, which was defined as any mitral valve erative disease more clearly, we analyzed a cohort operation for any cause during the study period (see drawn from all patients undergoing mitral valve definitions). This study was approved by the Data operations in New York State. We sought to deter- Protection Review Board of the New York State mine the effect of surgeon-specific factors on repair Department of Health, as well as by the Program for rates, survival, and long-term freedom from repeat Protection of Human Subjects at the Icahn School of mitral valve surgery. Medicine at Mount Sinai in New York City. The METHODS approval included a waiver of informed consent. PATIENTS. This study was an analysis of adult pa- tients, 18 years of age or older, who underwent primary mitral valve operations in New York State between January 1, 2002 and December 31, 2013. Patients were identified using the Statewide Planning and Research Cooperative System, an all-payer, administrative database that prospectively collects data on every hospital discharge, ambulatory surgery, and emergency department visit in New York State, with longitudinal data available from 1995 to 2014. We reviewed all available pre-operative patient data to identify patients’ baseline characteristics and followed up patients for at least 12 months postoperatively. Patients undergoing mitral valve replacement were identified using the International Classification of Diseases-Ninth Revision-Clinical DEFINITIONS. Total annual surgeon volume for the surgeon who performed the operation on each patient was calculated as the number of mitral valve operations for any causes (repair or replacement) performed by the patient’s operating surgeon in the 365 days immediately before surgery, to reflect the recent experience of each surgeon at the time of surgery. Next, the experience of each surgeon across the entire study period was quantified as mean total annual surgeon volume. Degenerative mitral valve repair rate was calculated as the number of repairs divided by the total number of operations for degenerative mitral valve disease a surgeon performed. Surgeons were considered eligible for the study when they practiced for at least 365 days. ENDPOINTS. The study endpoints were Modification (ICD-9-CM) procedure codes 35.23 and STUDY 35.24; patients undergoing mitral valve repair were all-cause mortality, degenerative repair rates, and identified using ICD-9-CM codes 35.12 and 35.33. post-repair mitral valve reoperation. Reoperation was Through exclusion of other causes, we identified a defined subgroup of patients with presumed degenerative replacement or repair, on subsequent admissions. mitral valve disease. Exclusion criteria were other Patients with no documented reoperation were possible mitral valve causes, including any history of censored on December 31, 2014. Deaths were identi- coronary artery disease, myocardial infarction, rheu- fied using the full national Social Security Death matic valve disease, infective endocarditis, congen- Master File (current as of May 29, 2015) and by ital heart disease, and cardiomyopathy. Patients who searching all hospital admissions and ambulatory or underwent concomitant surgery on the aortic valve, emergency department visits for patients’ deaths. For pulmonary valve, coronary arteries, or the left 1-year mortality and reoperation, time to events was as any mitral valve operation, either JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes F I G U R E 1 Institutional Repair Rates for Degenerative Mitral Valve Operations Between 2002 and 2013 at All Institutions in New York State 100 90 80 Degenerative Repair Rate (%) 70 60 50 40 30 20 10 41 35 36 37 38 39 40 31 32 33 34 25 26 27 28 29 30 21 22 23 24 15 16 17 18 19 20 7 8 9 10 11 12 13 14 5 6 1 2 3 4 0 Individual Institutions in New York State Wide variation in degenerative mitral repair rates is observed among all 41 institutions reporting mitral valve operations in New York State. censored at 1 year after surgery for patients who did surgeons and adjusting for covariates and total not have events by then. annual surgeon volume treated as a continuous variable. Baseline characteristics (age, sex, race or STATISTICAL ANALYSIS. Continuous variables were ethnicity, type of admission, year of mitral surgery, reported as means with SDs. Categorical variables hypertension, diabetes, peripheral vascular diseases, were expressed as proportions. Differences in base- cerebrovascular disease, chronic heart failure, atrial line characteristics and comorbidities among patients fibrillation, chronic obstructive pulmonary disease, treated by surgeons with different volumes were chronic kidney disease, liver disease, malignant dis- assessed using analysis of variance for normally ease, coagulation, platelet disorders, and pulmonary distributed Pearson’s hypertension) were included in the model as cova- chi-square test for categorical variables. To describe riates. For the primary endpoint analysis of survival, the difference in patients’ baseline demographics and a separate model with procedure type (repair or comorbidities according to total annual surgeon replacement) was also created. This model included volume, were the baseline covariates listed in the preceding text created: #10; 11 to 24; 25 to 50; and $51 mitral valve and procedure type. The probability of mitral valve procedures/year. repair was assessed by fitting multivariable logistic 4 continuous groups variables with arbitrary and cutoffs Survival after mitral valve repair or replacement regression models with generalized estimating and post-repair reoperation were assessed by fitting equations to control for clustering of patients oper- multivariable Cox proportional hazard models with a ated on by the same surgeons, where repair proced- robust sandwich variance estimator to control for ure was a dependent variable and the baseline clustering of patients operated on by the same characteristics listed earlier were included as 2399 2400 JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes T A B L E 1 Demographics of Patients With Degenerative Mitral Valve Disease Stratified by Surgeon Volume Groups #10 Operations/Yr (n ¼ 370) 11–24 Operations/Yr (n ¼ 1,042) 25–50 Operations/Yr (n ¼ 1,847) $51 Operations/Yr (n ¼ 2,216) 59.3  13.9 59.0  14.3 60.3  14.0 58.8  14.2 177 (47.8) 505 (48.5) 875 (47.4) 1128 (50.9) p Value Demographics Age, yrs Male Race 0.005 0.14 <0.001 White (non-Hispanic) 227 (61.4) 649 (62.3) 1,263 (68.4) 1,636 (73.8) African American (non-Hispanic) 44 (11.9) 119 (11.4) 173 (9.4) 126 (5.7) Hispanic 18 (4.9) 75 (7.2) 163 (8.8) 80 (3.6) Other/unknown 81 (21.9) 199 (19.1) 248 (13.4) 374 (16.9) Urgent admission 98 (26.5) 231 (22.2) 309 (16.7) 345 (15.6) <0.001 Hypertension 182 (49.2) 549 (52.7) 973 (52.7) 1,015 (45.8) <0.001 Diabetes mellitus 37 (10.0) 105 (10.1) 168 (9.1) 138 (6.2) <0.001 <10 (<2.7)* 12 (1.2) 15 (0.8) 15 (0.7) 0.10 Cerebrovascular disease 12 (3.2) 42 (4.0) 67 (3.6) 50 (2.3) 0.02 Congestive heart failure 159 (43.0) 440 (42.2) 698 (37.8) 595 (26.9) <0.001 Atrial fibrillation 141 (38.1) 356 (34.2) 654 (35.4) 716 (32.3) 0.07 Chronic obstructive pulmonary disease 52 (14.1) 159 (15.3) 289 (15.7) 236 (10.7) <0.001 Comorbidities Peripheral vascular disease Chronic kidney disease Liver disease Cancer Coagulation/platelet disorders Pulmonary hypertension Mitral repair (repair rate) 25 (6.8) 52 (5.0) 106 (5.7) 50 (2.3) <0.001 <10 (<2.7)* 49 (4.7) 48 (2.6) 55 (2.5) 0.003 37 (10.0) 79 (7.6) 180 (9.8) 221 (10.0) 14 (3.8) 60 (5.8) 84 (4.6) 114 (5.1) 0.35 <10 (<2.7)* 24 (2.3) 34 (1.8) 36 (1.6) 0.61 179 (48.4) 581 (55.8) 1,190 (64.4) 1,710 (77.2) <0.001 0.15 Values are mean  SD or n (%). *Small cell sizes (number <10) are suppressed according to our data use agreement with the Statewide Planning and Research Cooperative System. independent variables. Risk-adjusted probabilities of adjusted HR by 2 groups was reported. The associa- mitral valve repair with corresponding 95% confi- tion between total annual surgeon volume and sur- dence intervals (CIs) were plotted at each total vival or probability of repair was linear: adjusted HR annual surgeon volume. for survival and adjusted OR for repair were calcu- Total annual surgeon volume was first included lated by 10-case volume increments and also by the into each model as a continuous variable, and the arbitrary cutoffs defined earlier. Cumulative inci- linear association between volume and each outcome dence function curves of post-repair reoperation were was tested using restricted cubic spline functions constructed using risk competing analysis. The pro- with 3 knots at total annual surgeon volumes of 10, portional hazards assumption was valid in all Cox 25, and 50 operations/year (11,12). Linearity was models. All tests were 2-tailed, and an alpha level of assessed by linear hypothesis testing, and Akaike in- 0.05 was considered statistically significant. All sta- formation criteria were used for the best model se- tistical analyses were performed using SAS software lection. The volume association with outcomes was version 9.4 (SAS Institute, Cary, North Carolina). shown either as an adjusted hazard ratio (HR) for survival and reoperations or as an adjusted odds ratio RESULTS (OR) with its 95% CI for assessing probability of repair. The association between total annual surgeon STUDY POPULATION. A total of 5,475 adults with volume and post-repair reoperation was nonlinear. degenerative mitral valve disease, 18 years of age or By using a multivariable Cox regression model with older, who underwent primary mitral valve opera- restricted cubic spline function, the HR of reoperation tions performed by cardiac surgeons in New York within 1 year after repair was plotted against total State from January 1, 2002 to December 31, 2013, were annual surgeon volume (11). The observed change identified: 3,660 (66.8%) patients underwent mitral point of the slope of the HR curve was used as a cutoff valve repair, and 1,815 patients (33.2%) underwent value. The same Cox regression model, with surgeon mitral valve replacement. Median follow-up time was volume as a categorical variable, was created, and the 6.8 years (range 0 to 13.4 years). JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes C ENT R AL I LL U STRA TI O N Current Status of Degenerative Mitral Valve Repair in New York State Chikwe, J. et al. J Am Coll Cardiol. 2017;69(19):2397–406. Key findings: 1) the median annual surgeon mitral volume is 10 operations; 2) wide variability of repair rates is observed among surgeons; 3) higher volume is associated with higher repair rates and better outcomes; 4) reoperation after repair is less common for patients operated on by surgeons with $25 mitral valve operations a year; and 5) low-volume surgeons (<25 operations/year) in institutions where high-volume, high-repair-rate surgeons (>50 operations/year and >70% repair rate) are present have improved repair rates. A total of 313 surgeons from 41 institutions met the In the cohort of 5,475 patients with degenerative study eligibility criteria. Surgeons in New York State disease, surgeons with a total annual surgeon performed a median of 10 mitral valve operations/ volume <25 operations carried out 25% of operations year (range 1 to 230). The repair rate for primary (n ¼ 1,412). These patients were significantly more mitral valve procedures for any causes was 55% likely to present as urgent admissions (27% vs. 16%; (n ¼ 20,797 of 38,128); the median annual institu- p < 0.001), and they were more likely to have major tional mitral valve volume was 59 mitral valve oper- comorbidities, such as congestive heart failure (43% ations, ranging from a minimum of 6 to a maximum of vs. 27%; p < 0.001), chronic kidney disease (7% vs. 2%; 310 operations. Repair rates for primary mitral valve p < 0.001), or chronic airway disease (14% vs. 11%; operations for any cause at all 41 institutions varied p < 0.001) than were patients operated on by from 15% to 83%, and repair rates for degenerative surgeons with higher total annual surgeon volumes mitral valve operations varied from 25% to 100% (Table 1). (Figure 1). Of 313 surgeons, 231 operated on at least 1 patient in the cohort with degenerative disease who MITRAL VALVE REPAIR RATE. We observed a sig- was included in further analysis. nificant association between lower total annual 2401 JACC VOL. 69, NO. 19, 2017 Chikwe et al. 2402 MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes valve repair (adjusted OR: 3.18; 95% CI: 2.02 to 5.00; F I G U R E 2 Predicted Probability of Mitral Repair for Degenerative Operations p < 0.001) (Table 2). According to Total Annual Surgeon Mitral Valve Volume Predicted Probability of Degenerative Repair We observed a possible influence of a high-volume 1.0 and high-repair surgeon on low-volume surgeons in 0.9 the same institution. Among the 1,412 patients with degenerative mitral valve disease who were operated 0.8 on 0.7 by surgeons with a total annual surgeon volume <25, 20% (n ¼ 282) underwent mitral valve 0.6 surgery at centers where there was also an individual 0.5 surgeon carrying out >50 mitral valve operations annually with a >70% degenerative disease repair 0.4 rate. The repair rate of the 49 surgeons who 0.3 performed <25 operations annually at such centers 0.2 was 63.8%, compared with 51.3% at the remaining 0.1 centers (adjusted OR: 1.79; 95% CI: 1.24 to 2.60; OR 1.13 (95% CI 1.10-1.17), p<0.001 0.0 0 25 50 Predicted Probability of Repair 75 100 125 150 175 Total Annual Surgeon Mitral Volume Upper 95% Confidence Interval 200 225 p ¼ 0.002). 250 REOPERATION AFTER MITRAL REPAIR. We observed Lower 95% Confidence Interval a significant association between low surgeon volume and increased risk of mitral valve reoperation within After adjustment for pre-operative risk factors, degenerative repair probability is significantly associated with total annual mitral valve surgeon volume. CI ¼ confidence interval; OR ¼ odds ratio. 12 months of follow-up after mitral valve repair (p ¼ 0.04). This relationship was nonlinear, and the curve of the HR of reoperation changed its slope at a total annual surgeon volume of 25 operations (Figure 3). On the basis of these data, patients were surgeon volume and lower mitral valve repair rates in stratified into 2 volume groups (<25 operations/year the cohort with degenerative disease. The overall and $25 operations/year). The cumulative incidence mitral valve repair rate in the degenerative disease of reoperation at 12 months was 1.3% (95% CI: 1.0% to cohort was 66.8% (n ¼ 3,660 of 5,475). The observed 1.8%) for patients operated on by a surgeon with a degenerative disease repair rate according to mean total annual surgeon volume of $25 operations total annual surgeon volume is shown in the Central compared with 3.6% (95% CI: 2.4% to 5.0%) for pa- Illustration. After multivariable adjustment, total tients operated on by a surgeon with total annual annual surgeon volume was independently associ- surgeon volume of <25 operations (adjusted HR: 0.45; ated with the probability of mitral valve repair; the 95% CI: 0.26 to 0.76; p ¼ 0.003) (Figure 4). The sig- chance of repair increased by 13% for every 10-case nificant difference in reoperation rates persisted to 12 increment in total annual surgeon volume (adjusted years (9.5%; 95% CI: 6.9% to 12.6% vs. 6.2%; 95% CI: OR: 1.13/10-case increment; 95% CI: 1.10 to 1.17; 5.0% to 7.7%; p < 0.001) (Online Figure 2). p < 0.001) (Figure 2). Compared with patients oper- SURVIVAL. We observed an association between ated on by surgeons with a total annual surgeon higher total annual surgeon volume and improved volume of #10 operations, patients operated on by survival after mitral valve operations controlling for surgeons with a total annual surgeon volume of >50 pre-operative risk factors. Total annual surgeon vol- operations were >3 times as likely to undergo mitral ume was independently associated with improved 1-year survival in the degenerative disease cohort (adjusted HR: 0.95/10-case increment; 95% CI: 0.92 to T A B L E 2 Adjusted OR of Mitral Valve Repair According to 0.98; p ¼ 0.001). The actuarial survival after 1 year of Surgeon Volume Groups, p < 0.001 repair or replacement of patients with degenerative Volume Category (Operations/Yr) mitral valve disease operated on by surgeons per- Adjusted OR (95% CI) Observed Repair Rate, % #10 Reference 48.4 97.1 to 98.4) compared with 94.1% (95% CI: 91.1 to 11–24 1.22 (0.89–1.70) 55.8 96.0) for patients operated on by surgeons perform- 25–50 1.77 (1.26–2.49) 64.4 ing #10 operations a year (Table 3). To evaluate $51 3.18 (2.02–5.00) 77.2 whether and how the effect of surgeon volume was CI ¼ confidence interval; OR ¼ odds ratio. forming >50 operations a year was 97.8% (95% CI: confounded by repair rates, we added repair versus replacement to the previous model. Mitral repair JACC VOL. 69, NO. 19, 2017 2403 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes (vs. replacement) was significantly associated with better survival (adjusted HR: 0.80; 95% CI: 0.68-0.94; F I G U R E 3 Adjusted HR of Mitral Reoperation After Repair Within 1 Year, Plotted According to Total Annual Surgeon Mitral Valve Volume p ¼ 0.006), but total annual surgeon volume still 1.5 1.4 increment; 95% CI: 0.95 to 0.98; p < 0.001), a finding implying that better patient survival by higher-volume surgeons was explained not simply by their higher repair rate alone, but also by the effect of their higher volumes. In those patients who underwent mitral valve repair, total annual surgeon volume was an independent predictor of late death (adjusted HR: 0.96/10-case increment; 95% CI: 0.94 to 0.98; p < 0.001). DISCUSSION Hazard Ratio of Degenerative Mitral Reoperation remained a significant independent predictor, despite some attenuated effects (adjusted HR: 0.96/10-case 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 This New York State–wide multicenter analysis sug- 0.1 10 gests that individual surgeons’ mitral valve case HR plot volume has a significant impact on early- and long- 20 30 60 40 50 70 Total Annual Surgeon Mitral Volume 95% Upper Confidence Interval 80 90 95% Lower Confidence Interval term patients’ outcomes after mitral valve surgery. We observed incremental benefits in terms of repair Adjusted hazard ratio (HR) of mitral reoperation after repair within 1 year was plotted rates, survival, and reoperation with increasing against total annual surgeon mitral volume, with 10 mitral operations/year as a refer- surgeon volume. Thus our data provide additional strong support to the calls for systematically ence. The curve of the hazard ratio of reoperation changed its slope at a total annual surgeon volume of 25 operations. CI ¼ confidence interval. focusing experience in mitral valve surgery (13–17). Encouraging targeted referral, with the goal of concentrating surgical volume, should help 100 case volumes. The proportion of patients undergoing to urgent surgery was also significantly higher for lower- address the wide variation in mitral valve repair volume surgeons. This leads to a double jeopardy, rates described in the United States and elsewhere where sicker patients are adversely affected by the (3–10), a variation that persists despite the signifi- lower repair rates and poorer outcomes seen with cant benefits of mitral valve repair over replacement lower-volume surgeons, and it underscores the need (1,2). We found that a total annual surgeon volume to refer the highest-risk patients to high-volume of <25 operations was associated not only with surgeons. lower mitral valve repair rates, but also with Importantly, we observed improved mitral valve increased 1-year mortality and mitral valve reoper- repair rates and survival in the patients of the lower- ation rates, and that improvements in repair rates, volume surgeons if they operated at institutions survival, and freedom from reoperation continued where there was a surgeon performing >50 total with increasing surgeon case volumes. The median mitral operations annually with a >70% degenera- number of mitral valve operations performed annu- tive repair rate. One potential explanation for this ally by individual surgeons in the United States was finding is the direct benefit of reviewing and oper- 5, according to an analysis of The Society of Thoracic ating on patients who require a technically complex Surgeons database (3); similarly, in New York State, repair strategy with a more experienced surgeon and most surgeons actually performed <1 mitral opera- surgical team. A second reason for this finding may tion/month. Our data indicate that there may be be that centers where mitral valve surgeons perform several factors contributing to poorer outcomes in a higher number of mitral valve repairs are more patients operated on by the lower-volume surgeons likely to have cardiology, imaging, and critical care in this analysis. teams optimally equipped to evaluate and manage We observed significant differences in the charac- these patients. The demonstrated ability of experi- teristics of patients across surgeons’ case volume enced centers to field multidisciplinary teams expe- groups. For example, the prevalence of congestive rienced in the guideline-based assessment and heart failure was significantly higher in patients management of complex heart valve disease is the operated on by surgeons with lower annual case main rationale for managing such patients in heart volumes, compared with surgeons with higher annual valve referral centers (1,2). New and low-volume 2404 JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes no data on volume-outcome relationships for mitral F I G U R E 4 Cumulative Incidence of Mitral Valve Reoperation After Repair Within valve surgery were available (13). In the United States, 1 Year by 2 Volume Groups large health care purchasers, payers, and professional organizations have already established minimum 5 Cumulative Incidence of Mitral Reoperation (%) HR = 0.45 (95% CI 0.26-0.76, p = 0.003) volume standards for several surgical procedures, but mitral valve surgery is not among these (18). Our data 4 suggest that focusing surgeon experience may help to facilitate improvements in mitral valve repair rates 3 and long-term outcomes. Until then, consensus guideline standards (which require a high likelihood 2 of a durable repair for patients undergoing surgery for degenerative mitral valve disease) (1,2) may best be 1 served by concentrating referrals among surgeons performing $25 mitral valve operations annually. 0 0 6 12 Months Since Surgery 1-24 Operations/Year 25 or More Operations/Year Number at Risk 1-24 Operations/Year 760 25 or More Operations/Year 2,900 726 2,842 709 2,819 Considering that there was an incremental improvement in survival and probability of repair with increasing volume over 25 operations, one could make the argument that a minimum volume target of 50, or even more, operations would be optimal. Patients operated on by higher-volume surgeons ($25 operations/year) had lower risk of Developing more very high-volume surgeons experi- reoperation at 1 year than patients operated on by lower-volume surgeons (1 to 24 enced in mitral valve repair would likely be particu- operations/year). Abbreviations as in Figure 3. larly beneficial for patients with complex, but repairable mitral valve disease and for asymptomatic patients whose repair feasibility would optimally surgeons may be mentored and technically sup- approach 100% (19). ported, thus allowing them to perform degenerative The main strength of this study is the ability to mitral valve repair while successfully transforming analyze freedom from reoperation and mortality rates to high-volume surgeons. Even among high-volume in a large, multicenter, contemporary cohort of surgeons, there was an observed variability of patients undergoing surgery for degenerative disease degenerative disease repair rates, ranging from 19% by using a statewide mandatory database. to nearly 100%. This finding reflects that surgeon volume is not the only factor for better outcomes, STUDY LIMITATIONS. For this analysis, we used an and it emphasizes the need for more transparency of administrative dataset subject to well-recognized surgeon-related factors and outcomes of degenera- limitations that can affect the reliability of informa- tive mitral valve surgery for patients and referring tion obtained from it. Data may be entered by non- cardiologists. clinicians and are subject to inaccurate coding of The role of volume thresholds in designating referral centers and surgeons has not patients’ diagnoses and procedures. Additionally, been ICD-9-CM codes do not perfectly distinguish degen- established. A volume threshold of 25 mitral valve erative from ischemic mitral valve disease, nor do repairs/year for surgeons performing mitral surgery they provide detailed information on symptom sta- was originally proposed in a consensus statement on tus, severity of mitral valve disease, ventricular best practices in the United Kingdom, at a time when function, pulmonary hypertension, or severity of comorbidities (e.g., coronary disease). Because pre- T A B L E 3 Adjusted HR of 1-Yr Mortality of Patients With operative echocardiographic data or surgical reports Degenerative Mitral Valve Disease According to Surgeon were not available, our creation of a subgroup of pa- Volume Groups, p < 0.001 tients with degenerative disease relied on an exclu- Volume Category (Operations/Yr) sionary process. We believe we were able to validate a Adjusted HR (95% CI) 1-Yr Survival (95% CI) #10 Reference 94.1% (91.1–96.0) disease with high specificity, but this was at the 11–24 0.72 (0.42–1.22) 95.8% (94.4–96.8) expense of sensitivity. For example, we excluded 25–50 0.52 (0.31–0.85) 97.0% (96.1–97.7) certain patients with degenerative mitral valve dis- $51 0.46 (0.28–0.76) 97.8% (97.1–98.4) ease and concomitant coronary disease from our CI ¼ confidence interval; HR ¼ hazard ratio. method of identifying patients with degenerative analysis to minimize the inclusion of patients with ischemic mitral regurgitation in our degenerative JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes disease group. We also excluded nearly 10% of pa- significantly worse survival and were significantly tients who lived out of New York State. Although the more likely to undergo reoperation in long-term use of a statewide, rather than a single-center, data- follow-up than patients operated on by higher- base improved our ability to detect reoperations, we volume surgeons. Incremental benefits, in terms of could not identify patients with residual or recurrent repair rates, reoperation, and survival, continued mitral regurgitation, patients who underwent valve with increasing surgeon volume, up to the highest- replacement for immediate failure of repair during volume surgeons in the study cohort. The presence the same operation, or patients who subsequently of a high-volume surgeon was associated with better migrated out of state, thus potentially causing us to repair rates achieved by low-volume surgeons oper- underestimate the rate of repair failure. We excluded ating at the same institution. Our data support 2.5% of operations performed by surgeons where we concentrating surgeons’ experience in mitral valve were unable to calculate their annual surgical vol- surgery and the concept of centers of excellence in umes; this could potentially contribute to under- mitral valve repair. estimating the effect of volume on repair rates and long-term outcomes. Each surgeon’s cumulative ADDRESS FOR CORRESPONDENCE: Dr. David H. experience was not analyzed because cumulative Adams, volume could not be accurately calculated for sur- Mount Sinai Medical Center, 1190 Fifth Avenue, New Department geons who started their practice before the study York, period or out of New York State. Finally, we were mountsinai.org. New York of 10029. Cardiovascular E-mail: Surgery, david.adams@ unable to adjust for referral bias, which may cause us to overestimate the number of mitral valves PERSPECTIVES amenable to repair seen by low-volume surgeons, as well as the impact of more complex repairs seen by COMPETENCY IN PRACTICE-BASED LEARNING: Based on higher-volume surgeons. observations in the state of New York, a minimum surgeon CONCLUSIONS annual volume of 25 mitral operations is a reasonable target to improve clinical outcomes in patients with degenerative mitral In this 12-year analysis of patients operated on for mitral valve disease in New York State, surgeons performing <25 mitral valve operations/year were significantly more likely to replace, rather than repair, mitral valves in patients with degenerative disease than were surgeons performing $25 mitral valve operations/year. Additionally, patients valve disease. TRANSLATIONAL OUTLOOK: Prospectively acquired national data in patients undergoing mitral surgery for degenerative disease is needed to further define estimates of minimum surgeon volume targets to improve repair rates and clinical outcomes. operated on by the lower-volume surgeons had REFERENCES 1. Nishimura RA, Otto CM, Bonow RO, et al. 2017 4. Gammie BP, 8. Vassileva CM, McNeely C, Spertus J, Markwell S, AHA/ACC Focused Update of the 2014 AHA/ACC Guideline for the Management of Patients With Valvular Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol 2017 Mar 10. [E-pub ahead of print]. Ferguson TB, Peterson ED. Influence of hospital procedural volume on care process and mortality for patients undergoing elective surgery for mitral regurgitation. Circulation 2007;115:881–7. Hazelrigg S. Hospital volume, mitral repair rates, and mortality in mitral valve surgery in the elderly: an analysis of US hospitals treating Medicare feefor-service patients. J Thorac Cardiovasc Surg 2015;149:762–8.e1. 2. Joint Task Force on the Management of Valvular Heart Disease of the European Society of Cardiology (ESC), European Association for Cardio-Thoracic Surgery (EACTS), Vahanian A, Alfieri O, et al. Guidelines on the management of valvular heart disease (version 2012). Eur Heart J 2012;33:2451–96. 3. Bolling SF, Li S, O’Brien SM, Brennan JM, Prager RL, Gammie JS. Predictors of mitral valve repair: clinical and surgeon factors. Ann Thorac Surg 2010;90:1904–12. JS, O’Brien SM, Griffith 5. Anyanwu AC, Bridgewater B, Adams DH. The lottery of mitral valve repair surgery. Heart 2010; 96:1964–7. 6. LaPar DJ, Ailawadi G, Isbell JM, et al. Investigators for the Virginia Cardiac Surgery Quality Initiative. Mitral valve repair rates correlate with surgeon and institutional experience. J Thorac Cardiovasc Surg 2014;148:995–1004. 7. Kilic A, Shah AS, Conte JV, Baumgartner WA, Yuh DD. Operative outcomes in mitral valve surgery: combined effect of surgeon and hospital volume in a population-based analysis. J Thorac Cardiovasc Surg 2013;146:638–46. 9. Vassileva CM, Mishkel G, McNeely C, et al. Long-term survival of patients undergoing mitral valve repair and replacement: a longitudinal analysis of Medicare fee-for-service beneficiaries. Circulation 2013;127:1870–6. 10. Vassileva CM, Boley T, Markwell S, Hazelrigg S. Impact of hospital annual mitral procedural volume on mitral valve repair rates and mortality. J Heart Valve Dis 2012;21: 41–7. 11. Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in 2405 2406 JACC VOL. 69, NO. 19, 2017 Chikwe et al. MAY 16, 2017:2397–406 Surgeon Volume and Long-Term Mitral Surgery Outcomes public health 1037–57. research. Stat Med 2010;29: 12. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989;8: 551–61. 13. Bridgewater B, Hooper T, Munsch C, et al. Mitral repair best practice: proposed standards. Heart 2006;92:939–44. 14. Adams DH, Anyanwu AC. The cardiologist’s role in increasing the rate of mitral valve repair in degenerative disease. Curr Opin Cardiol 2008;23: 105–10. 15. McCarthy PM. When is your surgeon good enough? When do you need a “referent surgeon”? Curr Cardiol Rep 2009;11:107–13. 16. Adams DH, Rosenhek R, Falk V. Degenerative mitral valve regurgitation: best practice revolution. Eur Heart J 2010;31:1958–66. 17. Chikwe J, Adams DH. Megaphone message: discouraging low-volume mitral surgery. J Thorac Cardiovasc Surg 2015;149:769–70. 18. Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for highrisk surgery. N Engl J Med 2011;364:2128–37. 19. Castillo JG, Anyanwu AC, Fuster V, Adams DH. A near 100% repair rate for mitral valve prolapse is achievable in a reference center: implications for future guidelines. J Thorac Cardiovasc Surg 2012; 144:308–12. KEY WORDS mitral valve repair, mitral valve replacement, surgical volume A PPE NDI X For supplemental tables and figures, please see the online version of this paper.