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.