CARDIA

C

ADULT Isolated Mitral Valve Surgery Risk in 77,836

Patients From The Society of Thoracic SurgeonsDatabaseSubhasis Chatterjee, MD, J. Scott Rankin, MD, James S. Gammie, MD,Shubin Sheng, PhD, Sean M. O’Brien, PhD, J. Matthew Brennan, MD, MPH,John C. Alexander, MD, Vinod H. Thourani, MD, Paul J. Pearson, MD, PhD, andRakesh M. Suri, MD, DPhilAffinity Health System, Appleton, Wisconsin; Vanderbilt University, Nashville, Tennessee; University Maryland School of Medicine,Baltimore, Maryland; Duke Clinical Research Institute, Durham, North Carolina; NorthShore University HealthSystem, Evanston,Illinois; Emory University School of Medicine, Atlanta, Georgia; and Mayo Clinic, Rochester, Minnesota

Background. Understanding the operative outcomes ofmitral valve (MV) surgery across the spectrum of pre-dicted risk of mortality (PROM) is necessary to determinethe best use of novel percutaneous approaches.

Methods. The Society of Thoracic Surgeons AdultCardiac Surgery Database was utilized to study isolatedMV operations during two time periods: era 1 (2002 to2006, n [ 37,743) and era 2 (2007 to 2010, n [ 40,093). Ineach era, four PROM groups were defined: low risk(PROM 0% to <4%); intermediate risk (PROM 4% to<8%); high risk (PROM 8% to <12%); and extreme risk(PROM ‡12%). In each risk group, mortality rates andobserved to expected mortality ratios were comparedacross eras.

Results. A total of 63,645 cases (82%) were classified aslow risk, 8,032 (10%) as intermediate risk, 2,765 (4%) ashigh risk, and 3,394 (4%) as extreme risk. Sixty-seven

Accepted for publication June 6, 2013.

Presented at the Forty-ninth Annual Meeting of The Society of ThoracicSurgeons, Los Angeles, CA, Jan 26–30, 2013.

Address correspondence to Dr Chatterjee, Affinity Heart LungVascular Center, 1506 Oneida St, Appleton, WI 54915; e-mail: schatter@affinityhealth.org.

� 2013 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

percent of MV repairs (n [ 30,488) and 18% of MVreplacements (n [ 5,749) had a PROM less than 1%.PROM less than 4% was seen for 93% of MV repairs (n [42,196) and 66% of replacements (n [ 21,449). Across thetwo eras, the MV repair rate increased from 54.8% to61.8% (p [ 0.0017); and a significant reduction in opera-tive mortality was observed in high risk and extreme riskcohorts (p < 0.05).

Conclusions. The frequency with which MV repair forisolated MV disease is performed has increased over timeand is associated with very low early mortality. A sig-nificant reduction in mortality among patients at highestrisk has occurred, and must be considered as patients areevaluated for percutaneous therapies.

(Ann Thorac Surg 2013;96:1587–95)� 2013 by The Society of Thoracic Surgeons

itral valve (MV) repair is generally considered the

Mtreatment of choice for severe mitral regurgitation(MR), as reflected in current American College of Cardi-ology/American Heart Association [1] and EuropeanSociety of Cardiology [2] guidelines. Although arandomized comparison of repair versus replacementhas not been conducted, retrospective series [3],propensity-matched groups [4], and metaanalyses [5]have consistently demonstrated superior early and latesurvival with repair. As a result, MV repair isperformed in more than 60% of patients with primaryregurgitant lesions according to national databases [6, 7]Despite a clear benefit of surgical intervention overmedical management, as many as half of patients whomeet current indications for MV surgery do not undergo

surgery because of either perceptions of excessivesurgical risk or inadequate knowledge of current guide-lines [8, 9].This study used the national Society of Thoracic Sur-

geons Adult Cardiac Surgery Database (STS-ACSD) toevaluate contemporary operative results of isolated mitralvalve surgery in the United States, and to assess the utilityof the STS predicted risk of operative mortality (PROM)model for risk prediction among patients selected formitral valve operations in contemporary surgical practice.

Patients and Methods

Data SourceThe Society of Thoracic Surgeons has maintained anobservational database of patients having cardiothoracicsurgery in the United States since 1989. The 2012

Dr Gammie discloses a financial relationship withCardioNet.

0003-4975/$36.00http://dx.doi.org/10.1016/j.athoracsur.2013.06.051

1588 CHATTERJEE ET AL Ann Thorac SurgISOLATED MITRAL VALVE SURGERY RISK 2013;96:1587–95

ADULTCARDIA

C

STS-ACSD includes the outcomes of 3031 surgeons prac-ticing in 1,069 hospitals, representingmore than 90% of UScardiac surgery programs and more than 4.8 million pa-tient records [10]. Harvested data are warehoused andanalyzed by the Duke Clinical Research Institute, incompliance with the Health Insurance Portability andAccountability Act of 1996. Variables include informationabout patient demographics, medical history, surgicalprocedures, cancer staging, and outcome. The DukeUniversity School of Medicine Institutional Review Boardapproved the analysis presented here.

PopulationA retrospective review was performed of all isolated MVoperations entered into the STS-ACSD from January 1,2002, to December 30, 2010. Cases were excluded if theyhad (1) incomplete data for variables in the predicted riskof operative mortality (PROM) risk model; (2) prior valvesurgery; or (3) any concomitant procedure (ie, coronaryartery bypass grafting, aortic/tricuspid valve surgery,aortic or atrial fibrillation surgery) at the time of indexMV operation (Fig 1). To facilitate evaluation of trends inoperative outcomes, patients were divided into twocohorts: era 1 (2002 to 2006, n ¼ 37,743) and era 2 (2007to 2010, n ¼ 40,093). For consistency with othercontemporary series [11], four groups were definedbased on STS-PROM score: low risk (PROM 0% to<4%); intermediate risk (PROM 4% to <8%); high risk(PROM 8% to <12%); and extreme risk (PROM �12%).

Statistical AnalysisThe distributions of patient characteristics and outcomeswere summarized by median and interquartile range forcontinuous variables and percentage distribution forcategorical variables. Comparisons of subgroups wereperformed using Wilcoxon rank sum tests (continuous

Fig 1. Mitral valve (MV) operations studycohort after exclusion criteria. (PROM ¼predicted risk of mortality.)

variables) or c2 tests (categorical variables). The primaryendpoint, operative mortality, was defined by the STS asdeath within the same hospitalization as the mitraloperation or after discharge but within 30 days of theprocedure. For each patient, the PROM was calculatedbased on the published STS 2008 risk model for isolatedvalve surgery [12]. This model calculates risk based on thetype of operation performed (ie, repair, replacement) andpreoperative risk factors. A recalibrated version of thismodel is used for ongoing analyses by the STS analyticcenter. For the present analysis, the model was used inits original form, without recalibration, for consistencywith the published algorithm and the STS online riskcalculator.Analyses compared the predicted mortality data, as

described by the PROM, with actual observed mortality.Comparisons of observed versus predicted mortality rateswere performed for both the overall isolated MV surgerycohort and within subgroups based on year of operation(2002 to 2006 versus 2007 to 2010), type of operation(repair versus replacement), and category of predictedrisk (low, intermediate, high, extreme risk). First, the as-sociation between mortality and PROM as a continuousvariable was nonparametrically estimated using smooth-ing splines and plotted in a graph. Second, to quantify thecalibration of the model, the observed mortality rate wasdivided by the average PROM to obtain an observed-to-expected (O:E) ratio. Comparisons of O:E ratios acrosssubgroups were performed using Z tests, treating theobserved mortality as random and the expected mortalityas constant. Calibration graphs were generated separatelyfor each era, both overall and separately for the MV repairand replacement subgroups. All analyses were performedusing SAS 9.2 (SAS Institute, Cary, NC) and R version2.11 (R Foundation for Statistical Computing, Vienna,Austria).

1589Ann Thorac Surg CHATTERJEE ET AL2013;96:1587–95 ISOLATED MITRAL VALVE SURGERY RISK

ADULTCARDIA

C

Results

Study PopulationFrom 2002 to 2010, 77,836 patients underwent an isolatedmitral valve operation in STS network surgical centers,including 37,743 patients in era 1 (2002–2006) and 40,093in era 2 (2007–2010) (Fig 1). Demographics andcomorbidities for the full cohort stratified by treatmentera are presented in Table 1. Compared with patientstreated in era 1, patients treated in the more recent era2 had a higher prevalence of chronic obstructivepulmonary disease, cerebrovascular disease, age morethan 80 years, and surgery for MR. Across thetwo eras, MV repair rate increased from 54.8% to 61.8%(p ¼ 0.0017).

In the overall cohort, 82% were classified as low risk,10% intermediate risk, 4% high risk, and 4% extreme risk(Table 2). Patients undergoing mitral valve repair had alower baseline PROM than patients undergoing mitralvalve replacement (Fig 2). Overall, a PROM score less than1% was observed in 67% of all MV repairs (n ¼ 30,488) and18% of MV replacements (n ¼ 5,749).

Operative MortalityIn the more recent era 2 (2007 to 2010), overall operativemortality after MV surgery was 2.9%, which, as comparedwith era 1, was a significant reduction in both theobserved mortality (from 3.2%) and the O:E mortalityratio. In era 2, observed mortality was 1.3% among

Table 1. Total Cohort and Comparison of Era 1 and Era 2

Variable Level

Age Median/mean, years25th/75th percentile, years�80�70 and <80�60 and <70<60

Sex FemaleRisk factors Renal failure

COPDPVDAny CVDCHFNYHA class IVPreoperative AF

Previous CV interventions Prior CABGPrevious other cardiac surgery

Hemodynamics and cath. Median/mean EFModerate to severe MRMitral stenosisModerate-severe TI

Operative status Procedure status: emergent/urgent

Data are percent (%) unless otherwise indicated.

AF ¼ atrial fibrillation; CABG ¼ coronary artery bypass graft surgery;chronic obstructive pulmonary disease; CV ¼ cardiovascular; CVDregurgitation; NYHA ¼ New York Heart Association; PVD ¼ periphe

patients classified as low risk by the PROM, 6.6% amongthe intermediate risk, 9.2% among the high risk, and20.1% among the extreme risk patients. In era 2, observedrisk was significantly higher than predicted risk amongpatients in the intermediate risk group (O:E 1.18, 95%confidence interval [CI]: 1.04 to 1.31), but significantlylower than predicted among patients in the extreme riskgroup (O:E 0.90, 95% CI: 0.82 to 0.99). Both the observedoperative mortality rates (Fig 3) and O:E ratios (Table 3)were significantly lower in era 2 (versus era 1) amonghigh-risk and extreme-risk patients, whereas no differ-ences were observed across low-risk and intermediate-risk patients.Among all current era mitral valve repair patients, the

overall mortality rate was 1.4%. The observed risk ofoperative mortality was similar to predicted, among allbut those at extreme risk (O:E 0.71, 95% CI: 0.54 to 0.89).Although the observed mortality rates were numericallylower among the era 2 (versus era 1) MV repair patients(Fig 3), neither the observed mortality rates nor the O:Eratios (Table 4) were significantly improved in this mostrecent era.Finally, among all current era mitral valve replacement

patients, the observed mortality rate was 5.4%. Theobserved risk of operative mortality was greater thanexpected for low-risk patients (O:E 1.20, 95% CI: 1.04 to1.36) and intermediate-risk patients (O:E 1.23, 95% CI:1.06 to 1.39), whereas the observed risk was similar topredicted among high-risk and extreme-risk patients.

Totaln ¼ 77,836

Era 1(n ¼ 37,743)

Era 2(n ¼ 40,093)

p ValueEra 1 vs Era 2

60/60 60/60 61/60 0.005850/71 50/71 51/717.4 6.9 7.9 <0.000120.7 21.4 20.124.4 23.3 25.447.5 48.4 46.748.1 48.8 47.3 <0.00015.3 6.5 4.3 <0.000119.2 17.7 20.6 <0.00015.4 5.4 5.5 0.57910.0 9.3 10.6 <0.000144.4 44.5 44.3 <0.000110.4 11.0 10.0 <0.000117.0 18.0 16.0 <0.00015.9 6.2 5.6 0.00062.5 3.0 2.0 <0.0001

58/55 55/55 60/55 <0.000191.5 89.4 93.6 <0.000112.7 14.5 11.0 <0.000110.7 8.1 13.4 <0.000123.3 21.6 24.5 <0.0001

cath. ¼ catheterization; CHF ¼ congestive heart failure; COPD ¼¼ cerebrovascular disease; EF ¼ ejection fraction; MR ¼ mitralral vascular disease; TI ¼ tricuspid insufficiency.

Table 2. Overall Mitral Valve Surgery Distribution by Predicted Risk of Mortality Group

STS-PROM

Combined 2002–2010 Era 1, 2002–2006 Era 2, 2007–2010

p Value Era2 vs Era 1

All MVSurgery

MVRepair (%)

MVReplacement

(%)All MVSurgery

MVRepair (%)

MVReplacement

(%)All MVSurgery MV Repair (%)

MVReplacement

(%)

0–4 63,645 42,196 (66.3) 21,447 (33.7) 30,943 19,310 (62.4) 11,633 (37.6) 32,702 22,886 (70.0) 9,814 (30.0) <0.00014–8 8,032 2,177 (27.1) 5,855 (72.9) 3,888 952 (24.5) 2,936 (75.5) 4,144 1,225 (29.6) 2,919 (70.4) <0.00018–12 2,765 566 (20.5) 2,199 (79.5) 1.268 226 (17.8) 1,042 (82.2) 1,497 340 (22.7) 1,157 (77.3) <0.000112þ 3,394 519 (15.3) 2,873 (84.7) 1,644 210 (12.8) 1,434 (87.2) 1,750 309 (17.7) 1,439 (82.2) 0.021Total 77,836 45,458 (58.4) 32,374 (41.6) 37,743 20,968 (54.8) 17,045 (45.2) 40,093 24,760 (61.8) 15,333 (38.2) 0.0017

MV ¼ mitral valve; STS-PROM ¼ The Society of Thoracic Surgeons predicted risk of mortality.

Table 3. Observed and Expected Mortality by Predicted Risk of Mortality Group, All Mitral Valve Operations

All MVOperationsSTS-PROM

Era 1, 2002–2006 Era 2, 2007–2010

p Value O:EEra 2 vs Era 1

% Total MVSurgery

ObservedMortality %

ExpectedMortality %

O:E Ratio(95% CI)

% Total MVSurgery

ObservedMortality %

ExpectedMortality %

p Value ObservedMortality

Era 2 vs Era 1O:E Ratio(95% CI)

0–4 82.0 1.3 1.2 1.08 (0.97–1.18) 81.6 1.3 1.2 0.652 1.08 (0.98–1.19) 0.9324–8 10.3 6.8 5.6 1.22 (1.08–1.36) 10.3 6.6 5.6 0.707 1.18 (1.04–1.31) 0.6758–12 3.4 12.4 9.7 1.28 (1.09–1.46) 3.7 9.2 9.8 0.008 0.94 (0.79–1.09) 0.00712þ 4.4 23.6 22.2 1.07 (0.97–1.16) 4.4 20.1 22.2 0.013 0.90 (0.82–0.99) 0.012Total 100.0 3.2 2.8 1.13 (1.06–1.19) 100.0 2.9 2.9 0.029 1.02 (0.97–1.08) 0.018

CI ¼ confidence interval; MV ¼ mitral valve; O:E ¼ observed to expected; STS-PROM ¼ The Society of Thoracic Surgeons predicted risk of mortality.

1590CHATTERJEEETAL

AnnThorac

Surg

ISOLATED

MITRALVALVESU

RGERY

RISK

2013;96:1587–95

ADULTCARDIAC

Fig 2. The Society of Thoracic Surgeons predicted risk of mortality(STS-PROM) distribution by incremental PROM for combined eras 1and 2, for all (A) mitral valve (MV) operations; (B) MV repair; and(C) MV replacement.

Fig 3. Continuity plot of observed and expected mortality for (A) allmitral valve (MV) operations, (B) MV repair, and (C) MV replace-ment. Green lines indicate era 1 (2002 to 2006); red lines indicate era 2(2007 to 2010).

1591Ann Thorac Surg CHATTERJEE ET AL2013;96:1587–95 ISOLATED MITRAL VALVE SURGERY RISK

ADULTCARDIA

C

Both the observed operative mortality rates (Fig 3) andO:E ratios (Table 5) were lower among high-risk andextreme-risk patients treated with mitral replacement inera 2 versus era 1.

Table 5. Observed and Expected Mortality by Predicted Risk of Mortality Group, Mitral Valve Replacements

MVRSTS-PROM

Era 1, 2002–2006 Era 2, 2007–2010

p Value O:EEra 2 vs Era 1

% TotalMVR

ObservedMortality %

ExpectedMortality %

O:E Ratio(95% CI)

% TotalMVR

ObservedMortality %

ExpectedMortality %

p Value ObservedMortality

Era 2 vs Era 1O:E Ratio(95% CI)

0�4 68.3 2.0 1.8 1.11 (0.97–1.26) 64.0 2.2 1.8 0.281 1.20 (1.04–1.36) 0.4694�8 17.2 6.7 5.6 1.19 (1.03–1.35) 19.0 7.0 5.7 0.715 1.23 (1.06–1.39) 0.7718�12 6.1 13.0 9.7 1.33 (1.12–1.54) 7.6 9.6 9.8 0.013 0.98 (0.81–1.15) 0.01112þ 8.4 23.9 22.2 1.08 (0.98–1.18) 9.4 20.8 21.9 0.041 0.95 (0.85–1.04) 0.065Total 100.0 5.3 4.6 1.14 (1.07–1.22) 100.0 5.4 5.0 0.719 1.07 (1.00–1.14) 0.159

CI ¼ confidence interval; MVR ¼ mitral valve replacement; O:E ¼ observed to expected.

Table 4. Observed and Expected Mortality by Predicted Risk of Mortality Group, Mitral Valve Repairs

MV RepairsSTS-PROM

Era 1, 2002–2006 Era 2, 2007–2010

p ValueO:E Era 2vs Era 1

% Total MVRepairs

ObservedMortality %

ExpectedMortality %

O:E Ratio(95% CI)

% Total MVRepairs

ObservedMortality %

ExpectedMortality %

p Value ObservedMortality

Era 2 vs Era 1O:E Ratio(95% CI

0–4 93.3 0.9 0.9 1.04 (0.88–1.19) 92.4 0.9 0.9 0.827 0.98 (0.85–1.12) 0.5814–8 5.0 7.1 5.5 1.31 (1.01–1.61) 5.0 5.8 5.5 0.210 1.06 (0.82–1.30) 0.2108–12 1.4 9.7 9.6 1.01 (0.61–1.42) 1.4 7.9 9.7 0.466 0.82 (0.52–1.12) 0.45112þ 1.3 21.4 21.9 0.98 (0.72–1.23) 1.3 16.8 23.5 0.194 0.71 (0.54–0.89) 0.097Total 100.0 1.5 1.4 1.08 (0.96–1.20) 100.0 1.4 1.5 0.533 0.93 (0.83–1.03) 0.063

CI ¼ confidence interval; MV ¼ mitral valve; O:E ¼ observed to expected; STS-PROM ¼ The Society of Thoracic Surgeons predicted risk of mortality.

1592CHATTERJEEETAL

AnnThorac

Surg

ISOLATED

MITRALVALVESU

RGERY

RISK

2013;96:1587–95

ADULTCARDIAC

1593Ann Thorac Surg CHATTERJEE ET AL2013;96:1587–95 ISOLATED MITRAL VALVE SURGERY RISK

ADULTCARDIA

C

STS PROM Risk Model PerformanceThe STS-PROM was well calibrated across the spectrumof predicted risk in patients undergoing isolated mitralvalve operations, including the subgroups of patientsundergoing mitral valve repair and replacement opera-tions. On average, however, the STS-PROM over-estimated risk in patients from era 2 compared with era1 at the highest predicted risk groups (Fig 3).

Comment

This analysis is the first to provide a nationally repre-sentative, contemporary benchmark for operative out-comes after mitral valve surgery in the United States. Theresults indicate that (1) the vast majority of MV operationsare performed in low-risk patients with excellent results;(2) operative outcomes for both repair and replacementcontinue to improve among patients at highest risk foradverse events; and (3) the STS-PROM risk calculator iswell calibrated across the spectrum of patient risk in themost recent cohort of patients selected to undergo iso-lated mitral valve operations.

The high proportion of low-risk patients presenting forMV surgery suggests that, unlike other valvular heartdisease, symptomatic mitral valve disease often presentsin younger patients and in isolation from other seriouscomorbidities. This observation also may support currentguideline recommendations for the early treatment ofselect patients with asymptomatic, severe mitral valvedisease, particularly in the setting of pulmonary hyper-tension or atrial fibrillation. Certainly, it has been estab-lished that mitral valve surgery among patients withadvanced symptoms has been associated with a sub-stantially higher operative mortality and a lower repairrate than in less symptomatic patients [13, 14].

Over the past decade, operative mortality rates haveremained stable for low- and intermediate-risk patients;however, substantial reductions in operative mortalityhave been observed among high- and extreme-riskpatients—especially, among those undergoing mitralvalve replacement. That may be due to increasingchordal-preservation techniques, fewer replacements af-ter failed initial repair attempts, enhanced care methods,or referral to more experienced centers. These continuedimprovements should be acknowledged and encouraged.

In this series, MV repair rate increased from 55% in era1 to 62% to era 2, a trend that was observed among each ofthe four PROM risk groups. This finding suggests thatrather than isolated improvement in the low-risk group,MV repair is broadly increasing across all risk groups.This trend toward increasing mitral valve repair mayhave resulted from a combination of factors, includingsimplified or improved mitral repair techniques, ongoingtraining in the broader surgical community, or regional-ization to centers with higher rates of repair. It is inter-esting to note the higher prevalence of MR and lowerprevalence of mitral stenosis in era 2 compared with era 1.This trend may also contribute to the increasing repairrates observed. Regardless of the mechanism, observed

benefits associated with mitral valve repair (versusreplacement) support continued educational efforts topromote the application of mitral valve repair as a first-line therapy among patients with suitable valvemorphology. Although recent results highlight the similarsafety and efficacy of minimally invasive [15] and robotic[16] platforms, the survival benefit [17] is primarilyassociated with MV repair, and thus, the focus of thesurgical community should be continued improvementin MV repair rates.In this series, only 8% of patients undergoing isolated

MV surgery had an STS-PROM score greater than 8%(fewer than 800 patients, annually). For this highest riskgroup of patients with MR who may not be surgicalcandidates, nascent percutaneous therapies could have afuture role [11, 18]. While the STS-PROM remains thebest available model for predicting adverse events afterisolated MV surgery, observed results were better thanexpected among patients at high and extreme risk ofoperative mortality, which is important to rememberwhen counseling these patients. In light of this observa-tion, individualized treatment plans should be carefullycrafted by a heart team with both surgical and cardiologyinput for patients with high and extreme predictedoperative risk—especially as percutaneous valve thera-pies become available [19].

Study LimitationsThis study has several limitations consistent with obser-vational studies in general. First, some MV replacementswere performed after attempted MV repair at the samesetting, and thus, were exposed to a higher risk of com-plications due to longer ischemic and bypass times than aprimary MV Replacement. The previous STS model didnot capture intention-to-treat or surgical crossover in thedata collection forms. The proportion of MV re-placements representing failed MV repair was 19% of allMV replacements in one report [13]. The most recentiteration of the STS-ACSD data collection form (version2.73) includes intent-to-treat information, and future an-alyses using this variable will be instructive. Second, byexcluding concomitant coronary artery bypass graft sur-gery, this study likely disproportionately reduced theoverall contribution from ischemic MR patients thatmight have impacted the higher risk groups. Addition-ally, isolated mitral valve surgery in this analysis repre-sents only 27% of all MV procedures performed in theSTS-ACSD. However, the analysis of this experiencemay be relevant for comparison to emerging percuta-neous mitral valve therapies. Further analysis of thebroader MV surgery population will be necessary to un-derstand the complete spectrum of patients selected forsurgical management. Third, because this study wasdesigned to focus exclusively on operative mortality,echocardiographic follow-up, functional outcomes, andlong-term survival were not evaluated. These variablesare critical when comparing the efficacy of percutaneoustreatments to established MV surgery.In conclusion, the majority of MV surgery performed in

the United States is in low-risk patients with a very low

1594 CHATTERJEE ET AL Ann Thorac SurgISOLATED MITRAL VALVE SURGERY RISK 2013;96:1587–95

ADULTCARDIA

C

observed operative mortality. Excellent outcomes shouldcontinue to encourage early referral for MV repair beforethe development of significant comorbidities. Trends to-ward improvement in operative mortality among high-risk and extreme-risk patients should be considered asthese patients are evaluated for percutaneous therapies inthe United States.

References

1. Bonow RO, Carabello BA, Chatterjee K, et al. 2008 focusedupdate incorporated into the ACC/AHA 2006 guidelines forthe management of patients with valvular heart disease: areport of the American College of Cardiology/AmericanHeart Association Task Force on Practice Guidelines. J AmColl Cardiol 2008;52:e1–142.

2. Vahanian A, Alfieri O, Andreotti F, et al. Guidelines on themanagement of valvular heart disease (version 2012). EurHeart J 2012;33:2451–96.

3. Suri RM, Schaff HV, Dearani JA, et al. Survival advantageand improved durability of mitral repair for leaflet pro-lapse subsets in the current era. Ann Thorac Surg 2006;82:819–26.

4. Moss RR, Humphries KH, Gao M, et al. Outcome of mitralvalve repair or replacement: a comparison by propensityscore analysis. Circulation 2003;108:II90–7.

5. Shuhaiber J, Anderson RJ. Meta-analysis of clinical outcomesfollowing surgical mitral valve repair or replacement. Eur JCardiothorac Surg 2007;31:267–75.

6. Gammie JS, Sheng S, Griffith BP, Peterson ED,O’Brien SM, Brown JM. Trends in mitral valve surgery inthe United States: results from The Society of ThoracicSurgeons Adult Cardiac Database. Ann Thorac Surg2009;87:1431–9.

7. Bridgewater B, Kinsman R, Walton P, et al. Sixth nationaladult cardiac surgery database report 2008. Reading, UK:Dendrite Clinical Systems, 2009.

8. Bach DS, Awais M, Gurm HS, Kohnstamm S. Failure ofguideline adherence for intervention in patients with severemitral regurgitation. J Am Coll Cardiol 2009;54:860–5.

9. Toledano K, Rudski LG, Huynh T, B�eïque F, Sampalis J,Morin J-F. Mitral regurgitation: determinants of referral forcardiac surgery by Canadian cardiologists. Can J Cardiol2007;23:209–14.

10. Shahian DM, Jacobs JP, Edwards FH, et al. The Society ofThoracic Surgeons national database. Heart 2013 Jan 18[E-Pub ahead of print].

11. Whitlow PL, Feldman T, Pedersen WR, et al. Acute and 12-month results with catheter-based mitral valve leaflet repair.J Am Coll Cardiol 2012;59:130–9.

12. O’Brien SM, Shahian DM, Filardo G, et al. The Society ofThoracic Surgeons 2008 cardiac surgery risk models: part2—isolated valve surgery. Ann Thorac Surg 2009;88(Suppl):23–42.

13. Gillinov AM, Mihaljevic T, Blackstone EH, et al. Should pa-tients with severe degenerative mitral regurgitation delaysurgery until symptoms develop? Ann Thorac Surg 2010;90:481–8.

14. Stevens LM, Rodriguez E, Lehr EJ, et al. Impact of timingand surgical approach on outcomes after mitral valveregurgitation operations. Ann Thorac Surg 2012;93:1462–8.

15. Gammie JS, Zhao Y, Peterson ED, O’Brien SM, Rankin JS,Griffith BP. Less-invasive mitral valve operations: trends andoutcomes from The Society of Thoracic Surgeons adult car-diac surgery database. Ann Thorac Surg 2010;90:1401–10.

16. Suri RM, Burkhart HM, Daly RC, et al. Robotic mitral valverepair for all prolapsed subsets using techniques identical toopen valvuloplasty: establishing the benchmark againstwhich percutaneous interventions should be judged.J Thorac Cardiovasc Surg 2011;142:970–9.

17. Badhwar V, Peterson ED, Jacobs JP, et al. Longitudinaloutcome of isolated mitral repair in older patients—resultsfrom 14,604 procedures performed from 1991 to 2007. AnnThorac Surg 2012;94:1870–9.

18. Van den Branden BJL, Swaans MJ, Post MC, et al. Percuta-neous edge-to-edge mitral valve repair in high-surgical riskpatients. J Am Coll Cardiol Intv 2012;5:105–11.

19. Treede H, Schirmer J, Rudolph V, et al. A heart team’sperspective on interventional mitral valve repair: percuta-neous clip implantation as an important adjunct to a surgicalmitral valve program for treatment of high-risk patients.J Thorac Cardiovasc Surg 2012;143:78–84.

DISCUSSION

DR HAROLD G. ROBERTS (Fort Lauderdale, FL): I haveenjoyed the study, and find it gratifying to see that the repair ratecontinues to climb. However, I think before we pat ourselves onthe back a bit too vigorously, we should at some point try to gleanhow many of these repairs eventuate in replacements. I havebeen to several valve meetings where videos of complex mitralreconstructions are presented. I can tell you, I have looked at theend result of some of these repairs and I’m thinking, I’m not surethat this is going to do real well. Perhaps some of these patients,particularly those with already impaired LV function, might beperhaps better served with a tissue prosthesis inserted in lessthan half the clamp time needed for some gonzo repair of un-certain quality and durability. I think at times we may be pushingthe limits of reconstruction a little too far.

DR CHATTERJEE: That is an excellent comment, Dr Roberts,and I recognize your leadership in mitral valve surgery. I thinkthat the new iteration of the risk model will be able to gleanfurther information, and the long-term information that will beinteresting in studies like this is looking at long-term data interms of not just freedom from reoperation but freedom fromsignificant mitral regurgitation.

DR EDWARD M. BENDER (Cape Girardeau, MO): I enjoyedyour talk. Two points. Number one, the act of calculating apreoperative risk and placing somebody in the repair versusreplacement category automatically decreases your predictedrisk; it’s almost like a self-fulfilling prophecy. So that brings upthe question, let’s say you could find out those patients whocould go either way, repair versus replacement, filter out thestenoses, look at only the regurgitant patients, of which the vastmajority are repairable, and then compare the observed versusexpected mortality figures for the regurgitant patients whoshould or could be repaired versus those that are not repaired toequalize the preoperative status of those patients. Have youlooked at that, strictly regurgitant patients?

DR CHATTERJEE: I have not. You bring up two excellent pointsand I think the analysis is very perceptive. I will take the secondquestion first. Mitral stenosis represented 12% of the entiremitral valve surgery population. It represented 25% of the mitralreplacement population. So approximately 75% of mitral re-placements were performed in patients without mitral stenosis.Addressing your first question, you are absolutely correct. The

mitral surgery risk model automatically adjusts for mitral valve

1595Ann Thorac Surg CHATTERJEE ET AL2013;96:1587–95 ISOLATED MITRAL VALVE SURGERY RISK

ADULTCARDIA

C

repair versus replacement. So that for the exact same risk profile,for example, a 55-year-old patient who is otherwise healthy with a55% ejection fraction has a 0.3% predicted risk for mitral repairand a 0.6% risk for mitral replacement. So you are correct. Thenext step in the analysis that we are contemplating is to take all ofthese patients and give them the same risk score, say the mitralvalve replacement risk score, and then look back on it accordingly.But the point is correct that for the exact same risk profile, if youenter mitral valve repair, you will get a slightly lower number.

DR BENDER: The reason I bring that up is because, in reality,what you would do as a surgeon is your preoperative riskcalculation is actually done at the very end of the operation whenyou have either replaced the valve or repaired the valve. That isthe way that most surgeons I think go and fill the forms out andthen the data are harvested that way. You have no guarantee thatthese are indeed preoperative risks. They are actually a minute ofpostoperative risks.

DR CHATTERJEE: That’s a fair point.

DR LAURENS PICKARD (Houston, TX): Excellent paper andpresentation. I just wanted to ask how you decided on the timeperiod division between the two eras.

DR CHATTERJEE: We divided it between 2002 to 2006 and thenbetween 2007 to 2010. In 2008 the current risk model wasdeveloped, and it was based upon data from 2002 to 2006. Whatwe looked at is the same 2008 risk model. We analyzed the 2007to 2010 data and then we went back and used the same model toanalyze the 2002 to 2006. In some ways it was arbitrary to be ableto find a time period, but we saw that as a natural break becauseof when the data collection forms transitioned.

arolyn E. Reed Traveling Fellowship

C

It is with honor and sorrow that the Thoracic SurgeryFoundation for Research and Education (TSFRE), inconjunction with Women in Thoracic Surgery (WTS),announces the Carolyn E. Reed Traveling Fellowship.Dr Reed was a thoracic surgeon, educator, and suc-

cessful researcher; she died early and before herretirement. To honor her legacy, TSFRE in conjunctionwith WTS, established this award, which will allow aclinically established woman thoracic surgeon to travelto another institution in order to learn a new skill ortechnology.The reality and success of this award requires a

specialty-wide fundraising effort. Together with you –

Carolyn’s colleagues, CT surgery community members,corporate supporters, and institutional partners – we can

� 2013 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

DR VINAY BADHWAR (Pittsburgh, PA): The laudable intentthat you had for your paper was to set a benchmark againstpotential future therapies other than surgery. If I am inter-preting your data correctly, in the most recent era 2, the high-risk cohort you noted approximately 16% observed mortalityfor mitral valve repair and 20% in mitral replacement. TheCOAPT clinical trial is basically focusing on this high-risksubset of patients. Is your message that if we can get mortalitythat is less than 16%, this should be considered a comparablemarker for percutaneous treatment and any other future en-deavors, or do you foresee that we can decrease that mortalityrate further?

DR CHATTERJEE: That is an excellent question, Dr Badhwar.I would say there are a couple of things there. One is thatwhen looking at, for example, MitraClip or any other alter-native technologies, it is a limited sample in terms of thepotential number of patients that have the appropriate anat-omy to be able to do that. I think there are clearly a group ofpatients at a certain high risk level that are appropriatecandidates for a percutaneous technology.This study indicates that, and we see this in the observed

mortality results, because repair has a better mortality thanreplacement, the more of these patients that we can repair, Ithink, overall our short-term mortality results will improve. Atthis point with this study given the numbers of the study, it isprobably still early.

DR BADHWAR: All answered with good longitudinal follow-up.

DR CHATTERJEE: Yes.

DR BADHWAR: Excellent presentation. Thank you.

celebrate Carolyn’s life and dedication to education andinnovation.There are several ways to contribute to the Carolyn E.

Reed Traveling Fellowship:

� Make a credit card donation online at www.tsfre.org.� Mail your donation to TSFRE, 633 N Saint Clair St,Flr 23, Chicago, IL 60611. Please make checkspayable to TSFRE and note the Carolyn E. ReedFellowship in the memo section of your check.

Please contact Priscilla S. Page, TSFRE ExecutiveDirector, at ppage@tsfre.org with questions. A full de-scription of the Carolyn E. Reed Traveling Fellowship canbe found at www.tsfre.org. Applications for this awardwill be accepted starting in July of 2014.

Ann Thorac Surg 2013;96:1595 � 0003-4975/$36.00