Increased mortality at low-volume orthotopic heart transplantation centers: should current standards change?

Abstract

The Centers for Medicare and Medicaid Services (CMS) mandate that orthotopic heart transplantation (OHT) centers perform 10 transplants per year to qualify for funding. We sought to determine whether this cutoff is meaningful and establish recommendations for optimal center volume using the United Network for Organ Sharing (UNOS) registry.
We reviewed UNOS data (years 1999 to 2006) identifying 14,401 first-time adult OHTs conducted at 143 centers. Stratification was by mean annual institution volume. Primary outcomes of 30-day and 1-year mortality were assessed by multivariable logistic regression (adjusted for comorbidities and risk factors for death). Sequential volume cutoffs were examined to determine if current CMS standards are optimal. Pseudo R2 and area under the receiver operating curve assessed goodness of fit.
Mean annual volume ranged from 1 to 90. One-year mortality was 12.6% (n = 1,800). Increased center volume was associated with decreased 30-day mortality (p < 0.001). Decreased center volume was associated with increases in 30-day (odds ratio [OR] 1.03, 95% confidence interval [CI]: 1.02 to 1.03, p < 0.001) and 1-year mortality (OR 1.01, 95% CI: 1.01 to 1.02, p = 0.03--censored for 30-day death). The greatest mortality risk occurred at very low volume centers (<or= 2 cases = 2.15 times increase in death, p = 0.03). Annual institutional volume of fewer than 10 cases per year increased 30-day mortality by more than 100% (OR 2.02, 95%CI: 1.46 to 2.80, p < 0.001) and each decrease in mean center volume by one case per year increased the odds of 30-day mortality by 2% (OR 1.02, 95% CI: 1.01 to 1.03, p < 0.001]. Additionally, centers performing fewer than 10 OHTs per year had increased cumulative mortality by Cox proportional hazards regression (hazard ratio 1.35, 95% CI: 1.14 to 1.60, p < 0.001). Sequential multivariable analyses suggested that current CMS standards may not be optimal, as all centers performing more than 40 transplants per year demonstrated less than 5% 30-day mortality.
Annual center volume is an independent predictor of short-term mortality in OHT. These data support reevaluation of the current CMS volume cutoff for OHT, as high-volume centers achieve lower mortality.

Full text

Page 1

Increased Mortality at Low-Volume Orthotopic
Heart Transplantation Centers: Should Current
Standards Change?
Eric S. Weiss, MD, Robert A. Meguid, MD, MPH, Nishant D. Patel, BA,
Stuart D. Russell, MD, Ashish S. Shah, MD, William A. Baumgartner, MD, and
John V. Conte, MD
Department of Surgery, Division of Cardiac Surgery, and Department of Medicine, Division of Cardiology, The Johns Hopkins Medical
Institutions, Baltimore, Maryland
Background. The Centers for Medicare and Medicaid
Services (CMS) mandate that orthotopic heart transplan-
tation (OHT) centers perform 10 transplants per year to
qualify for funding. We sought to determine whether
this cutoff is meaningful and establish recommendations
for optimal center volume using the United Network for
Organ Sharing (UNOS) registry.
Methods. We reviewed UNOS data (years 1999 to 2006)
identifying 14,401 first-time adult OHTs conducted at 143
centers. Stratification was by mean annual institution
volume. Primary outcomes of 30-day and 1-year mortality
were assessed by multivariable logistic regression (ad-
justed for comorbidities and risk factors for death).
Sequential volume cutoffs were examined to determine if
current CMS standards are optimal. Pseudo R2 and area
under the receiver operating curve assessed goodness of
fit.
Results. Mean annual volume ranged from 1 to 90.
One-year mortality was 12.6% (n ? 1,800). Increased
center volume was associated with decreased 30-day
mortality (p < 0.001). Decreased center volume was
associated with increases in 30-day (odds ratio [OR] 1.03,
95% confidence interval [CI]: 1.02 to 1.03, p < 0.001) and
1-year mortality (OR 1.01, 95% CI: 1.01 to 1.02, p ?
0.03— censored for 30-day death). The greatest mortality
risk occurred at very low volume centers (< 2 cases ? 2.15
times increase in death, p ? 0.03). Annual institutional
volume of fewer than 10 cases per year increased 30-day
mortality by more than 100% (OR 2.02, 95%CI: 1.46 to
2.80, p < 0.001) and each decrease in mean center volume
by one case per year increased the odds of 30-day
mortality by 2% (OR 1.02, 95% CI: 1.01 to 1.03, p < 0.001].
Additionally, centers performing fewer than 10 OHTs
per year had increased cumulative mortality by Cox
proportional hazards regression (hazard ratio 1.35, 95%
CI: 1.14 to 1.60, p < 0.001). Sequential multivariable
analyses suggested that current CMS standards may not
be optimal, as all centers performing more than 40
transplants per year demonstrated less than 5% 30-day
mortality.
Conclusions. Annual center volume is an independent
predictor of short-term mortality in OHT. These data
support reevaluation of the current CMS volume cut-
off for OHT, as high-volume centers achieve lower
mortality.
(Ann Thorac Surg 2008;86:1250 – 60)
© 2008 by The Society of Thoracic Surgeons
M
technique for orthotopic heart transplantation (OHT) [1].
When OHT was first applied to humans, only certain
centers with specially trained surgeons and substantial
resources could undertake the procedure and care for the
complex posttransplant patient. As time has passed with
improvements in immunosupression, intensive care unit
care, organ preservation, and surgical technique, OHT
has evolved to become the gold standard treatment for a
variety of types of end-stage heart disease and is avail-
ore than 45 years have passed since Norman
Shumway and Richard Lower described the initial
able in many centers worldwide. With more than 140
centers performing OHT in the United States alone, a
wide disparity exists in the number of cases performed
between centers. Because many procedures have vol-
ume-based outcomes, a similar analysis asking if center
volume correlates with outcomes in OHT should be
performed.
Hospital volume has been shown to affect surgical
outcomes for many types of surgery. In fact, over the past
30 years, several studies (examining pancreatic resec-
tions, esophagetcomies, lung lobectomies, and so forth)
have convincingly demonstrated that increased provider
volume is correlated with lower mortality [2–9], shorter
length of hospital stay [10], decreased readmission rates
[11], and decreased costs [10, 12, 13]. This concept is not
only pervasive in general surgery but has also found its
way into specialty surgical fields as well including urol-
Accepted for publication June 13, 2008.
Presented at the Forty-fourth Annual Meeting of The Society of Thoracic
Surgeons, Fort Lauderdale, FL, Jan 28–30, 2008.
Address correspondence to Dr Conte, Division of Cardiac Surgery, The
Johns Hopkins Hospital, 600 N Wolfe St, Blalock 618, Baltimore,
MD 21287-4618; e-mail: jconte@csurg.jhmi.jhu.edu.
© 2008 by The Society of Thoracic Surgeons
Published by Elsevier Inc
0003-4975/08/$34.00
doi:10.1016/j.athoracsur.2008.06.071
ADULT CARDIAC

Page 2

ogy [14], gynecology [15, 16], and vascular [17] and
cardiovascular surgery [18–20]. These findings have been
advocated by health quality consortiums, such as the
Leapfrog Group, and discussed in the lay press [21].
The volume-outcome relationship appears to be par-
ticularly important for highly complex procedures that
require a significant commitment of resources and highly
specialized teams [22]. Orthotopic heart transplantation
is a clear example of this type of procedure. In fact,
several studies have demonstrated a center effect
whereby low-volume OHT centers have increased short-
term mortality rates [23–25].
In response to this early work and the growing success
of OHT after the advent of cyclosporine, the Heath Care
Financing Administration (HCFA)—whose name was
changed in 2001 to the Centers for Medicare and Medic-
aid services (CMS)—publicly decided in 1986 that Medi-
care would fund OHT at HCFA-designated institutions
that met strict outcome standards and performed a
minimum volume of transplantations per year. Initially,
this minimum number was set at 12 cases per year.
Current CMS standards require only 10 cases per year for
approval [26].
It has now been 20 years since this initial mandate, and
although the importance of center volume appears clear,
most studies addressing this issue for OHT were con-
ducted in the late 1980s and early 1990s, before advances
in immunosuppressive and myocardial protection and
support gained significant acceptance. Furthermore, no
study has been designed to specifically examine CMS
standards directly.
The United Network for Organ Sharing (UNOS) data-
set is a nation-wide, physician-overseen sample that
provides a unique opportunity to address the issue of
center volume while minimizing single-institution bias.
We conducted a retrospective examination of the multi-
institutional UNOS dataset to examine effects of annual
institutional volume on short-term and midterm survival
in patients receiving OHT. We hypothesize that in-
creased center volume will be associated with decreased
short-term mortality independent of patient risk factors,
and further, that the current CMS standard may not be
optimal.
Material and Methods
Data Source
The UNOS provided deidentified patient data (stan-
dard transplant analysis and research [STAR] files)
from the Thoracic Organ Transplant Registry for the
years of 1987 to 2006, with follow-up through Septem-
ber of 2007. Unique center identifier codes were in-
cluded to allow examination of institutional volume.
The data include 433 unique demographic, operative,
and postoperative variables collected for all United
States patients undergoing thoracic organ transplanta-
tion and reported to the Organ Procurement Network
during the time period. As individual patients are not
identified in this multicenter registry report, the need
for consent and Institutional Review Board approval is
waived at our institution.
Study Design and Patient Population
A retrospective review of the UNOS dataset from January
1999 to December 2006 with follow-up through January
2007 was performed. These time points were chosen to
identify a modern cohort of patients who underwent
OHT with current management techniques. All first-time
orthotopic heart transplant patients between 18 and 80
years of age were included.
Institutional Volume
Using the unique center identifiers present in the UNOS
dataset, the variable of annual institutional volume was
derived from the extant data. In this way, each individual
patient outcome was linked to a hospital volume mea-
surement. We then primarily stratified by volume to
“low-volume centers” (performing fewer than 10 OHTs
per year) and “high-volume centers” (performing 10 or
more OHTs per year). To further assess the effects of
volume for centers with either very low or very high
volume, additional volume cutoffs were defined and
Fig 1. Distribution of center volumes for 143
centers included in the United Network for
Organ Sharing dataset. Sixty-four centers fall
below the requirement for Centers for Medi-
care and Medicaid Services funding of 10 or-
thotopic heart transplantations per year
(based on Organ Procurement and Transplan-
tation Network data, January 2007).
1251
Ann Thorac Surg
2008;86:1250–60
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
ADULT CARDIAC

Page 3

included fewer than 2, fewer than 5, 5 to 9, 10 to 40, and
more than 40 OHTs per year.
Variables Examined and Measures of Outcome
Of the 433 variables present in the UNOS dataset, perti-
nent clinical factors were compared between groups.
Specifically studied were baseline recipient demographic
factors (age, sex, and race), comorbidities (hypertension,
diabetes mellitus, body mass index, and preoperative
creatinine levels), and transplant variables (donor age,
ischemic time, anastamotic technique, human leukocyte
antigen mismatch, and year of transplant). In addition,
we examined markers of clinical acuity including
whether the patient was hospitalized or in an intensive
care unit before transplant, use of intra-aortic balloon
counterpulsation before transplant, and UNOS status.
Finally, hemodynamic variables before transplant were
identified, including mean pulmonary artery pressure,
pulmonary vascular resistance, cardiac index, and
transpulmonary gradient. The primary endpoint was
30-day mortality. Other endpoints included 1-year and
cumulative mortality.
Analysis
Comparisons of baseline characteristics between study
groups were performed using Student’s t test for contin-
uous variables and the chi-square test for categorical
variables. Cumulative survival was modeled using the
Kaplan-Meier method, with statistical differences between
survival curves assessed by use of the log-rank (Mantel-
Cox)test.Inaddition,modelingbasedon1-yearconditional
survival was employed to assess long-term survival inde-
pendent of early mortality. Mortality was first assessed for
all risk factors using a univariate model. Analysis was
conducted by use of multivariable logistic regression for
short-term (30-day and 1-year) mortality and by use of a
Cox proportional hazards regression model for cumulative
mortality. Significant univariate predictors of either short-
term or cumulative mortality were incorporated into the
multivariable model to assess the effect of volume on
Table 1. Baseline Characteristics of Patients Undergoing Orthotopic Heart Transplantation (OHT) at Low-Volume (? 10 OHT
per Year) Versus High-Volume (? 10 OHT per Year) Centers
Low Volume
(? 10 OHT/Year) n ? 1,608
No. (%) or Mean (SD)
High Volume
(? 10 OHT/Year) n ? 12,793
No. (%) or Mean (SD)
p Valuea
Demographics
Mean center volume
Mean age, years
Age ? 65 years
Female
Black, Hispanic, Native American, Asian/Pacific Islanderb
Comorbidities and acuity
History of hypertension
Diabetes mellitus
Preoperative creatinine
Body mass index
Hospitalized before transplant
ICU before transplant
IABP before transplant
Inotropes before transplant
Ventilated before transplant
UNOS status 1c
Days on wait list
Bicaval anastamosis
Donor age (years)
Ischemic time (hours)
Hemodynamic variables
Mean pulmonary artery pressure
Pulmonary vascular resistanced
Cardiac index
Transpulmonary gradiente
5.9 (2.4)
49.8 (12.8)
6.7 (108)
377 (23.4)
426 (26.5)
33.2 (22.0)
52.2 (11.8)
11.9 (1529)
3,054 (23.8)
3,091 (24.1)
? 0.001
? 0.001
? 0.001
0.75
0.04
600 (39.2)
311 (19.6)
1.4 (0.9)
26.4 (4.9)
810 (51.6)
437 (27.8)
66 (4.1)
557 (36)
56 (3.5)
1,166 (72.5)
239 (391)
484 (31.7)
31.1 (12.4)
3.04 (1)
4,858 (39.7)
2,689 (22.5)
1.4 (0.8)
26.4 (4.7)
6,469 (51.1)
3934 (31.1)
687 (5.3)
5,388 (46)
362 (2.8)
9,501 (74.3)
223 (371)
5,807 (48.0)
31.6 (12.6)
3.15 (1)
0.29
0.08
0.8
0.84
0.76
0.01
0.03
? 0.001
0.14
0.12
0.1
? 0.001
0.14
?0.001
28.1 (10.4)
2.2 (2)
2.3 (0.8)
9.0 (6.3)
28.4 (10.0)
2.4 (2)
2.3 (0.8)
9.5 (5.6)
0.06
0.07
0.61
0.01
aThe p value is based on comparison between two groups by either ?2or Student’s t test.
dataset.
pulmonary artery pressure minus pulmonary capillary wedge pressure divided by cardiac output.
pulmonary artery pressure minus pulmonary capillary wedge pressure.
bBoth race and ethnicity were variables present in the
dPulmonary vascular resistance defined by mean
eTranspulmonary gradient defined by mean
cUNOS status 1 refers to patients listed as status 1a, status 1b, or older UNOS status 1.
IABP ? intra-aortic balloon pump;ICU ? intensive care unit;UNOS ? United Network for Organ Sharing.
1252
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
Ann Thorac Surg
2008;86:1250–60
ADULT CARDIAC

Page 4

mortality. Only well-represented variables (less than 33%
missing in the registry) were included in multivariable
analysis.
To determine if the current CMS standard of 10 OHTs
per year is optimal, we performed a series of sequential
multivariable logistic regressions at changing volume
thresholds examining 30-day mortality as an endpoint.
Common independent variables (significant on univari-
ate analysis) were included along with the sequentially
changing independent variable of dichotomized annual
OHT volume. This allows for comparison of outcomes
from hospitals with less than versus those with greater
than or equal to, the changing volume cutoff. To illustrate
this concept, the first multivariable regression compared
centers with a mean annual center volume of 1 or more
and centers performing a mean of fewer than 1 per year.
The second regression compared those at a mean annual
center volume of 2 or more with those averaging fewer
than 2 per year. Sequential analyses were performed in
this manner until a volume threshold was tested for all
center volumes. Various measures of “goodness of fit”
were used examined for each comparison. These in-
cluded area under the receiver operating curve (AUC)
and McFadden’s Pseudo R2[27–29]. The latter result is
normalized and presented as a percentage.
For all analyses, a p value of less than 0.05 (two-tailed)
was considered significant, and all odds ratios (OR) and
regression coefficients are presented with 95% confi-
dence intervals (CI). All statistical analysis was per-
formed with the aid of STATA software (version 9.2;
StataCorp LP, College Station, Texas).
Results
During the 8-year study period, UNOS data revealed 143
unique OHT centers, ranging in mean annual institutional
volume from 1 to 90 cases per year, with a median volume
Fig 2. Individual center unadjusted percent mortality as plotted by
mean annual center orthotopic heart transplantation volume. Both
(A) 30-day and (B) 1-year mortality censored for 30-day mortality
are included. Data trend lines are provided, demonstrating the sub-
stantial effect that center volume has on 30-day mortality (based on
Organ Procurement and Transplantation Network data, January
2007).
Table 2. Multivariable Logistic Regression Model for 30-Day
Mortality
Risk FactorOR (95% CI)
p Valuea
Volume (continuous)
Volume ? 10b
Volume ? 5
Volume ? 2
Volume ? 2
Volume ? 40
Age
Sex (female versus male)
Hypertension
Diabetes mellitus
Creatinine
Panel reactive antibody (continuous) 1.01 (0.99–1.01)
UNOS status 1
Donor age
HLA mismatch (HLA ? 4 versus 5
or 6)
Ischemic time
Cardiac index
Pulmonary vascular resistance
Bicaval anastamosis
Mechanical ventilation before
transplant
Intensive care unit before transplant 0.98 (0.71–1.34)
Transplantation year
0.97 (0.97–0.98)
2.02 (1.46–2.80)
1.86 (1.04–3.32)
2.15 (1.02–4.56)
1.66 (0.38–7.2)
0.35 (0.22–0.56)
1.01 (1.00–1.02)
0.84 (0.62–1.14)
1.32 (1.02–1.71)
0.86 (0.63–1.18)
1.20 (1.11–1.30)
? 0.001
? 0.001
0.04
0.03
0.48
? 0.001
0.14
0.26
0.03
0.35
? 0.001
0.14
0.001
0.05
0.27
1.87 (1.30–2.69)
1.01 (1.0–1.02)
0.87 (0.67–1.12)
1.27 (1.14–1.43)
0.95 (0.74–1.21)
1.04 (0.97–1.10)
0.93 (0.70–1.23)
4.33 (2.56–7.33)
? 0.001
0.70
0.2
0.61
? 0.001
0.9
0.02 0.90 (0.83–0.98)
aThe p value is based on multivariable logistic regression analysis, using
factors significant on univariate analysis.
number of OHT cases/year.
bVolume defined as mean
Multivariable model included the following variables: mean annual
center orthotopic heart transplant (OHT) volume (continuous), age, sex,
history of hypertension, history of diabetes mellitus, preoperative creat-
inine, panel reactive antibody level, UNOS status 1, donor age, HLA
mismatch, ischemic time, preoperative cardiac index, preoperative pul-
monary vascular resistance, bicaval anastamotic technique, need for
mechanical ventilation before undergoing transplantation, and transplan-
tation year.
CI ? confidence interval;
odds ratio;
HLA ? human leukocyte antigen;
UNOS ? United Network for Organ Sharing.
OR ?
1253
Ann Thorac Surg
2008;86:1250–60
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
ADULT CARDIAC

Page 5

of 10 cases a year (Fig 1). At these 143 centers, 17,131
patients underwent OHT, and after exclusion of patients
with previous heart transplant (n ? 601) and patients less
than 18 years of age (n ? 2,128), 14,401 patients were
included in the final analysis.
Primary stratification revealed that 11% (n ? 1,608) of
patients received OHT at centers performing an average of
fewer than 10 transplants per year (64 centers ? 45%). An
additional 31 centers, who averaged 10 or more OHTs per
year, failed to achieve 10 transplants in at least 1 year of the
study. Thus, the total number of centers failing to achieve at
least 10 OHTs in all 8 years of the study was 95 (66%).
Baseline Demographics and Acuity
Patients transplanted at low- and high-volume centers
differed slightly in their baseline characteristics (Table 1).
Sex was equally distributed, although high-volume cen-
ters had a slightly greater number of underrepresented
minorities and transplanted patients who were slightly
older. In addition, patients at low- and high-volume
centers had similar rates of comorbidities, including
hypertension and diabetes mellitus. High-volume cen-
ters appeared to have slightly higher levels of acuity as
assessed by rates of intensive care unit admission, ino-
trope use, and intra-aortic balloon pump before trans-
plant. Rates of hospitalization before transplantation and
rates of listing as UNOS status 1, however, were similar
between groups. Hemodynamic variables were also sim-
ilar between groups. High-volume centers did have a
higher rate of performing OHT with the bicaval anasta-
motic technique, which may reflect that increased expe-
rience is necessary for proficiency with the multiple
anastamoses.
Center Volume and Unadjusted Mortality
During the study period, cumulative mortality was 21%,
with a 30-day mortality rate of 6%. Thirty-day and 1-year
mortality were both higher at low-volume centers on
unadjusted analysis: 8.5% versus 5.6% (p ? 0.001), and
16.2% versus 12.2% (p ? 0.001), respectively. Upon addi-
tional stratification, very low volume centers (fewer than
5 cases per year on average) demonstrated an unadjusted
30-day mortality rate of 9.2%, and those performing 0 to
1 transplants per year on average had a 30-day mortality
Fig 3. Kaplan-Meier estimates of 5-year sur-
vival for patients who underwent orthotopic
heart transplantation (OHT) at high-volume
(black line) and low-volume (gray dashed
line) centers. A survival table is provided, and
the number of patients at risk is given in pa-
rentheses. The p value corresponds to Mantel-
Cox log-rank test results (based on Organ
Procurement and Transplantation Network
data, January 2007).
Fig 4. Kaplan-Meier estimates of 5-year sur-
vival censored for patients who died before
1-year follow-up and stratified by mean cen-
ter volume. A total of 10,347 patients survived
beyond 1 year. The only group that varied
significantly from others was those centers
performing 5 or fewer OHTs per year. *p ?
0.001 for this volume group when compared
with the others using the Mantel-Cox log-
rank test (based on Organ Procurement and
Transplantation Network data, January 2007).
1254
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
Ann Thorac Surg
2008;86:1250–60
ADULT CARDIAC

Page 6

of 15.2% (Fig 2). One-year mortality rates were also
increased at very low volume centers (18.2% versus
12.4%, p ? 0.001, for centers performing fewer than 5
OHT per year compared with 5 or more per year (22%
versus 12.6%, p ? 0.001, for centers performing 0 to 1
OHT per year compared with 2 or more per year). It is
noteworthy that all centers performing more than 40
OHTs per year demonstrated less than 5% 30-day mor-
tality (Fig 2).
Multivariable Analysis
On multivariable analysis, decreased center volume was
significantly associated with an increase in short-term
mortality. Specifically, each decrease in mean center
volume by 1 case per year increased the odds of 30-day
mortality by 2% (OR 1.02, 95% CI: 1.01 to 1.03, p ? 0.001].
In addition, falling below the CMS standard of 10 cases
per year led to a 100% increase in risk of 30-day mortality
(Table 2). Other significant predictors of death on multi-
variable logistic regression included ischemic time, base-
line creatinine level, hypertension, transplantation year,
listing as UNOS status 1, donor age, and the requirement
of mechanical ventilation before transplantation (Table
2). Centers performing fewer than than 5 OHTs per year
and those performing 2 or fewer OHTs per year also had
significant increases in the odds of 30-day mortality on
multivariable logistic regression. Performing more than
40 OHTs per year was associated with a 65% decrease in
the odds of 30-day mortality. The effect of volume was not
restricted to 30-day mortality, as logistic regression of
1-year mortality censored for 30-day death revealed
increased mortality for low-volume centers (OR for death
at low-volume [? 10 OHT per year] centers ? 1.50, 95%
CI: 1.06 to 2.02, p ? 0.02).
Cumulative Survival
Unadjusted cumulative survival was first evaluated by
the Kaplan-Meier method. Although cumulative survival
was lower at centers performing fewer than 10 cases per
year (Fig 3), censoring for 1-year mortality eliminated the
effect, indicating that differences in long-term survival for
those centers performing fewer than 10 OHTs per year
are largely related to increases in early mortality (Fig 4).
Upon further stratification, however, those centers per-
forming fewer than 5 OHTs per year did show a signifi-
cant difference in cumulative Kaplan-Meier survival even
after censoring for 1-year mortality (p ? 0.001). High-
volume centers (more than 40 OHTs per year) did not
show a cumulative survival benefit when censored for
early death. On Cox proportional hazards regression
analysis, however, low-volume center status (fewer than
10 OHTs per year) was associated with a significant
increase in cumulative mortality (hazard ratio ? 1.35,
95% CI: 1.14 to 1.60, p ? 0.001] when compared with
high-volume centers (10 or more OHTs per year). In
addition, centers performing fewer than 5 OHTs per year
Fig 5. Various models of goodness of fit: area
under the receiver operating curve (AUC
[black]) and McFadden’s pseudo R2as a per-
centage (gray) of the sequential multiple lo-
gistic regression analyses performed at se-
quential volume cutoffs. Black and gray
arrows point to goodness of fit for comparison
between centers performing 10 or more cases
per year versus fewer than 10 cases per year.
This plot demonstrates that the current Cen-
ters for Medicare and Medicaid Services stan-
dard of 10 OHTs per year was not the opti-
mal fit in this model. The dotted line
represents the baseline AUC for the model
with no volume variable, which was 0.6415.
Table 3. Multivariable Cox Proportional Hazards Regression
Model for Cumulative Mortality
Risk FactorHR (95% CI)
p Valuea
Volume (continuous)
Volume ? 10b
Volume ? 5
Volume ? 2
Volume ? 2
Volume ? 40
0.99 (0.98–0.99)
1.35 (1.14–1.60)
1.90 (1.45–2.51)
1.78 (1.08–2.86)
1.73 (0.89–3.40)
0.71 (0.59–0.85)
?0.001
?0.001
?0.001
0.02
0.1
?0.001
aThe p value is based on multivariable Cox proportional hazards regres-
sion analysis, using factors significant on univariate analysis.
ume defined as mean number of OHT cases per year.
bVol-
Multivariable model included the following variables: mean annual
center orthotopic heart transplant (OHT) volume (continuous), age, sex,
history of hypertension, history of diabetes mellitus, preoperative creat-
inine, panel reactive antibody level, UNOS status 1, donor age, human
leukocyte antigen mismatch, ischemic time, preoperative cardiac index,
preoperative pulmonary vascular resistance, bicaval anastamotic tech-
nique, need for mechanical ventilation before undergoing transplanta-
tion, and transplantation year. Additional significant predictors of cumu-
lative mortality include history of diabetes, history of hypertension, panel
reactive antibody level (continuous), creatinine, mechanical ventilation
before transplant, donor age, and ischemic time. Transplant year and
UNOS status 1 did not emerge as significant predictors of cumulative
mortality but were predictors of 30-day mortality (Table 2).
CI ? confidence interval;
Network for Organ Sharing.
HR ? hazard ratio;UNOS ? United
1255
Ann Thorac Surg
2008;86:1250–60
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
ADULT CARDIAC

Page 7

and those performing 2 or fewer OHTs per year had
significantly lower cumulative survival when compared
with higher volume centers (Table 3).
Optimal Center Volume
Sequential multivariable logistic regressions at chang-
ing volume thresholds were conducted, examining
30-day mortality as an endpoint. At each sequential
volume threshold, low-volume centers were associated
with a significantly increased risk of adjusted 30-day
mortality when compared with higher volume centers.
The goodness of fit of the sequential models as as-
sessed by the AUC varied from a nadir of 0.64 to a
maximum of 0.665. Performing more than 10 cases per
year led to an AUC of 0.6551, which did not account for
the greatest variability in the model, and thus indicates
that achieving a volume of 10 does not predict mortal-
ity with highest accuracy in our multivariable analysis.
The results of the McFadden Pseudo R2mirrored that
of the AUC calculations (Fig 5).
Comment
In our analysis of cardiac transplant centers, we confirm
that a reduced volume of OHTs is associated with an
increase in short-term mortality. Our intention was to
provide a thorough treatise on the “center effect” in OHT
and determine whether current CMS standards are in-
deed appropriate. The large multi-institutional UNOS
dataset was used to provide an account of the effect of
volume on OHT during the years 1999 to 2006 to focus on
outcomes unaffected by major advances in the field of
heart transplantation. The CMS-defined cutoff of 10 has
been criticized by some as being too small. However, it is
noteworthy that despite this volume threshold being
required by CMS to receive federal funding, 65% of
centers studied failed to achieve this level during at least
one year of the study period.
The data provided in this analysis clearly point to a
volume effect whereby increased center volume leads to
a reduction in short-term mortality. This effect is most
pronounced at low-volume centers performing fewer
than 10 OHTs per year, as these low-volume centers have
a 100% increase in the risk of 30-day mortality when
compared with centers performing at least 10 OHTs per
year. It is also clear that having an OHT at a center
performing 2 or fewer OHTs per year is associated with a
115% increase in 30-day mortality. Some individual very
low volume centers have alarmingly high 30-day mortal-
ity rates. For example, there were five centers performing
an average of fewer than 1 OHT per year that demon-
strated 30-day mortality rates of more than 20%. Con-
versely, achieving high volumes was associated with
superior outcomes. All centers performing more than 40
OHTs per year demonstrated less than 5% 30-day unad-
justed mortality. Furthermore, it is clear that substantial
intercenter variance is reduced once a volume of 20
OHTs per year is achieved.
It is important to note that not all low-volume centers
have poor outcomes, and there is not a level of OHT
volume that ensures a superior outcome. Thirteen cen-
ters that averaged fewer than 1 OHT per year demon-
strated a 0% 30-day mortality rate. These data support
the concept that increased institutional experience and
activity is one factor associated with improved OHT
outcomes.
It is noteworthy that high-volume centers had im-
proved outcomes despite arguably increased acuity (as
measured by greater number of patients in an intensive
care unit before transplantation and preoperative intra-
aortic balloon pump and inotrope use). Furthermore,
high-volume centers transplanted a greater number of
older patients (older than 65 years). In this analysis, we
chose to include all adult patients (up to age 80). We did
this to limit biasing the results in favor of improved
survival at high-volume institutions (who treat a greater
number of older patients). Although age 65 is generally
considered a cutoff for performing OHT, a substantial
body of evidence now exists supporting acceptable out-
comes for older patients who undergo OHT [30]. Al-
though the data are not presented here, exclusion of
patients older than age 65 did not alter the results of this
analysis.
Although volume effects were most significant for
30-day mortality, long-term survival was also influenced
by volume status. One-year mortality was higher for
low-volume centers even after censoring for 30-day mor-
tality. Long-term survival proved less dependent on
center volume status, as censoring for 1-year mortality
led to similar 5-year survival by Kaplan-Meier analysis.
This indicates that the majority of low-volume programs
have their mortality within the first year. The notable
exception to this trend was for centers performing fewer
than 5 OHTs per year. These centers demonstrated
decreased cumulative survival even with censoring for
1-year mortality. Cox proportional hazards regression
also indicated that decreased volume led to increases in
cumulative mortality. The reasons for this trend are not
apparent from this dataset, but we can speculate that
they may relate to processes of care, such as limited
resources and expertise needed to appropriately follow
complex posttransplant patients at low-volume institu-
tions over time.
Importance of Volume in OHT
In addition to the findings of this study, substantial
literature has now provided ample support to the notion
that increased center and provider volume is important
for improved outcomes after surgical procedures [2–4, 6].
This effect is particularly true for transplantation where
patients and procedures are complex, and significant
hospital resources are required [22]. Several reports have
now confirmed that this “center effect” exists for OHTs as
well [23–25]. In two early reports utilizing the Interna-
tional Society of Heart and Lung Transplantation
(ISHLT) registry, published by Heck and colleagues [25]
and Evans and coworkers [23], low-volume OHT centers
demonstrated increased mortality rates. In an early study
conducted from 1984 to 1986 and based primarily on
survey data, Laffel and colleagues [31] demonstrated that
1256
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
Ann Thorac Surg
2008;86:1250–60
ADULT CARDIAC

Page 8

while volume plays a role in outcome after OHT, the
primary factor is the learning curve that takes place, as
supported by centers reporting progressively improved
outcomes over time.
The importance of this learning curve was refuted in a
more recent definitive study by Hosenpud and associates
[24] who failed to show improvement in outcomes over
the history of a program’s experience. This study simi-
larly utilized the UNOS dataset and examined outcomes
by center volume from 1987 to 1991. Centers were dichot-
omized into low (fewer than 9 OHT a year) and high-
volume centers (9 or more OHT a year), and 30-day and
1-year survival was lower in the low-volume group. The
authors made note of the fact that more than 50% of
centers were in the low-volume group.
More than 14 years have passed since this description
of the volume effect in OHT, and the currently used
definition of high-volume center remains similar. Al-
though we still believe volume to be important, reexam-
ination of this issue in the context of current surgical,
immunosuppressive, and critical care advances was
warranted.
Volume cutoffs are often arbitrarily defined and not
based on strict outcome standards. We attempted to
determine whether the current CMS standard of 10 cases
per year was indeed the “optimal” volume threshold by
analyzing the amount of variability in the data accounted
for by this volume cutoff and comparing it to other
possible cutoffs. We used sequential statistical modeling
centered on volume and compared the resultant “good-
ness of fit” from each model. Based on this technique,
performing 10 OHTs per year did not account for the
greatest variability in the model. An important finding
from this analysis, however, is that volume differences
account for no more than 1% of the variance in the
sequential multivariable models. This apparent paradox
whereby volume is clearly important but not sufficient to
fully explain mortality after OHT highlights the complex-
ity of these patients and underscores the importance of
identifying additional factors that affect outcomes be-
yond volume alone.
Processes of Care
Thus, the findings of the current and previous studies
lead us to speculate that center volume is one factor that
is highly important for short-term survival in OHT. This
notion, however, must be taken in the context of recent
work focusing on institutional factors affecting postoper-
ative outcomes. Many recent studies have focused on
“processes of care” that affect outcome, including the
presence of dedicated intensive care unit providers [32,
33], implementation of patient safety measures [34], a
multidisciplinary team approach, and implementation of
critical pathways [5, 35, 36]. Most of this work has been
examined using general surgery patients. However, it is
likely that these processes of care are equally important
for transplantation, where patients are complex and
utilize a variety of hospital resources. It is likely that
volume may be serving as a surrogate for additional
unidentified processes unable to be quantified using the
multi-institutional UNOS dataset. This last point under-
scores the importance of identifying these unknown
processes of care to help improve all centers—local,
regional, and national alike—to improve care for heart
transplant patients. To what degree of importance sur-
geon experience affects the outcomes above and beyond
these other processes cannot be quantified in this anal-
ysis and should be the subject of further study.
Access to Care and Organ Utilization
Restricting complicated procedures such as OHT to “cen-
ters of expertise” may impose unique access to care
challenges to patients, families, and payers. This is a
potential burden that cannot be ignored. However, it is
also noteworthy that high-risk patients are frequently not
accepted as recipients at lower volume centers owing to
fear of their impact on center mortality results. Access to
care is rarely examined from this perspective. However,
it is possible that limiting OHT to high-volume centers of
expertise may in fact increase access to care for high-risk
patients. Furthermore, our review indicates that high-
volume centers have better outcomes for high-risk
patients.
Organ utilization may also be improved by limiting
OHT to high-volume centers. Low-volume centers are
less likely to accept “marginal” donors owing to the same
mortality concerns mentioned and are therefore less
likely to have a listed recipient appropriate and available
for any given donor. For these reasons, discussion of
limiting OHT to centers of expertise must also include
mention of these important issues of access to care as
well as organ utilization.
Limitations
As in all retrospective studies, this analysis has limita-
tions. The UNOS database provides limited variables.
Had other variables been included in this database, the
observed results may have been different. Specifically,
the UNOS dataset offers limited information on immu-
nosuppressive regimens, type of rejection, and processes
of care already mentioned. Furthermore, follow-up is
limited. This study is retrospective and, by nature, cannot
account for inherent undocumented differences in pa-
tient characteristics. Although multivariable analysis par-
tially controls for this bias, one fundamental limitation of
any retrospective study is a lack of control of all potential
patient confounders. Nationwide clinical registries, in-
cluding the UNOS database, rely upon accurate coding of
information. We acknowledge that the data present may
not have necessarily been entered by individuals with
clinical expertise, and it is further unknown if errors were
made in the coding of information. However, errors of
this nature are likely to be equally distributed throughout
the data and should not lead to significant bias in the
results. Finally, from a mathematical perspective, it is
difficult to draw strong conclusions about outcomes at
centers with very low volume. Despite the size of the
dataset, centers performing fewer than 1 OHT per year
will have variable outcomes that are difficult to compare.
It is difficult to know whether a high mortality rate at a
1257
Ann Thorac Surg
2008;86:1250–60
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
ADULT CARDIAC

Page 9

low-volume center is due to natural variance or a clear
trend.
In conclusion, low-volume cardiac transplantation is
correlated with increased mortality. Furthermore, 45% of
UNOS centers fall below CMS standards, performing an
average of fewer than 10 OHTs per year. These findings
should reinforce the notion that center volume, whether
by itself of serving as a surrogate for processes of care, is
important for improved outcomes after OHT. They
should also lead to the reevaluation of the current CMS
volume cutoff of 10 OHTs per year. Appropriate evi-
dence-based volume standards should be established
from multi-institutional data and active consideration of
enforceable penalties should exist for centers repeatedly
falling below acceptable levels.
The authors would like to thank Dianne Alejo, Jenna Pearce, and
Jennifer Neeley for the provision of Johns Hopkins OHT Data.
Doctor Weiss is the Irene Piccinini Investigator in Cardiac
Surgery. This work was supported in part by Health Resources
and Services Administration Contract 231-00-0115 and by a Ruth
L. Kirschstein National Research Service Award (NIH
2T32DK007713-12, to E.S.W. and R.A.M.). The content is the
responsibility of the authors alone and does not necessarily
reflect the views or policies of the Department of Health and
Human Services, nor does mention of trade names, commercial
products, or organizations imply endorsement by the US
government.
References
1. Lower RR, Stofer RC, Shumway NE. Homovital transplan-
tation of the heart. J Thorac Cardiovasc Surg 1961;41:196–
204.
2. Luft HS, Bunker JP, Enthoven AC. Should operations be
regionalized? The empirical relation between surgical vol-
ume and mortality. N Engl J Med 1979;301:1364–9.
3. Dudley RA, Johansen KL, Brand R, Rennie DJ, Milstein A.
Selective referral to high-volume hospitals: estimating po-
tentially avoidable deaths. JAMA 2000;283:1159–66.
4. Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wenn-
berg DE, Lucas FL. Surgeon volume and operative mortality
in the United States. N Engl J Med 2003;349:2117–27.
5. Birkmeyer JD, Sun Y, Goldfaden A, Birkmeyer NJ, Stukel
TA. Volume and process of care in high-risk cancer surgery.
Cancer 2006;106:2476–81.
6. Finlayson EV, Goodney PP, Birkmeyer JD. Hospital volume
and operative mortality in cancer surgery: a national study.
Arch Surg 2003;138:721–6.
7. Dimick JB, Goodney PP, Orringer MB, Birkmeyer JD. Spe-
cialty training and mortality after esophageal cancer resec-
tion. Ann Thorac Surg 2005;80:282–6.
8. Gordon TA, Burleyson GP, Tielsch JM, Cameron JL. The
effects of regionalization on cost and outcome for one gen-
eral high-risk surgical procedure. Ann Surg 1995;221:43–9.
9. Lieberman MD, Kilburn H, Lindsey M, Brennan MF. Rela-
tion of perioperative deaths to hospital volume among
patients undergoing pancreatic resection for malignancy.
Ann Surg 1995;222:638–45.
10. Stavrakis AI, Ituarte PH, Ko CY, Yeh MW. Surgeon volume
as a predictor of outcomes in inpatient and outpatient
endocrine surgery. Surgery 2007;142:887–99.
11. Weller WE, Rosati C, Hannan EL. Relationship between
surgeon and hospital volume and readmission after bariatric
operation. J Am Coll Surg 2007;204:383–91.
12. Harmon JW, Tang DG, Gordon TA, et al. Hospital volume
can serve as a surrogate for surgeon volume for achieving
excellent outcomes in colorectal resection. Ann Surg
1999;230:404–13.
13. Konety BR, Dhawan V, Allareddy V, O’Donnell MA. Asso-
ciation between volume and charges for most frequently
performed ambulatory and nonambulatory surgery for blad-
der cancer. Is more cheaper? J Urol 2004;172:1056–61.
14. Hollenbeck BK, Wei Y, Birkmeyer JD. Volume, process of
care, and operative mortality for cystectomy for bladder
cancer. Urology 2007;69:871–5.
15. Diaz-Montes TP, Zahurak ML, Giuntoli RL II, Gardner GJ,
Bristow RE. Uterine cancer in Maryland: impact of surgeon
case volume and other prognostic factors on short-term
mortality. Gynecolog Oncol 2006;103:1043–7.
16. Schrag D, Earle C, Xu F, et al. Associations between hospital
and surgeon procedure volumes and patient outcomes after
ovarian cancer resection. J Natl Cancer Inst 20061;98:163–71.
17. Hannan EL, Popp AJ, Tranmer B, Fuestel P, Waldman J,
Shah D. Relationship between provider volume and mortal-
ity for carotid endarterectomies in New York State. Stroke
1998;29:2292–7.
18. Hannan EL, Racz M, Ryan TJ, et al. Coronary angioplasty
volume-outcome relationships for hospitals and cardiolo-
gists. JAMA 1997;277:892–8.
19. Peterson ED, Coombs LP, DeLong ER, Haan CK, Ferguson
TB. Procedural volume as a marker of quality for CABG
surgery. JAMA 2004;291:195–201.
20. Nallamothu BK, Wang Y, Cram P, et al. Acute myocardial
infarction and congestive heart failure outcomes at specialty
cardiac hospitals. Circulation 2007;116:2280–7.
21. Comarrow A. Higher volume, fewer deaths. US News and
World Report, July 17, 2007, Sect. 68–70.
22. Lin HM, Kauffman HM, McBride MA, et al. Center-specific
graft and patient survival rates: 1997 United Network for
Organ Sharing (UNOS) report. JAMA 1998;280:1153–60.
23. Evans RW, Dong FB, Manninen DL. The center effect in
heart transplantation. Clin Transplants 1991:4559.
24. Hosenpud JD, Breen TJ, Edwards EB, Daily OP, Hunsicker
LG. The effect of transplant center volume on cardiac trans-
plant outcome. A report of the United Network for Organ
Sharing scientific registry. JAMA 1994;271:1844–9.
25. Heck CF, Shumway SJ, Kaye MP. The Registry of the
International Society for Heart Transplantation: sixth official
report—1989. J Heart Transplant 1989;8:271–6.
26. The Center for Medicare and Medicaid Services. Medicare
program; hospital conditions of participation: requirements
for approval and re-approval of transplant centers to per-
form organ transplants. Federal Register 2007:15197–280.
27. Allison PD. Comparing logit and probit coefficients across
groups. Sociolog Meth Res 1999;28:186–208.
28. Long JS, Freese J. Regression models for categorical and
limited dependent variables. Thousand Oaks, CA: Sage
Publications, 1997.
29. Long JS, Freese J. Regression models for categorical depen-
dent variables using Stata. 2nd ed. College Station, TX: Stata
Press, 2006.
30. Weiss ES, Nwakanma LU, Patel ND, Yuh DD. Outcomes in
patients older than 60 years of age undergoing orthotopic
heart transplantation: an analysis of the UNOS database.
J Heart Lung Transplant 2008;27:184–91.
31. Laffel GL, Barnett AI, Finkelstein S, Kaye MP. The relation
between experience and outcome in heart transplantation.
N Engl J Med 1992;327:1220–5.
32. Angus DC, Shorr AF, White A, Dremsizov TT, Schmitz RJ,
Kelley MA. Critical care delivery in the United States:
distribution of services and compliance with Leapfrog rec-
ommendations. Crit Care Med 2006;34:1016–24.
33. Dimick JB, Cattaneo SM, Lipsett PA, Pronovost PJ, Heit-
miller RF. Hospital volume is related to clinical and eco-
nomic outcomes of esophageal resection in Maryland. Ann
Thorac Surg 2001;72:334–41.
34. Makary MA, Sexton JB, Freischlag JA, et al. Patient safety in
surgery. Ann Surg 2006;243:628–35.
1258
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
Ann Thorac Surg
2008;86:1250–60
ADULT CARDIAC

Page 10

35. Hannan EL, Popp AJ, Feustel P, et al. Association of surgical
specialty and processes of care with patient outcomes for
carotid endarterectomy. Stroke 2001;32:2890–7.
36. Zehr KJ, Dawson PB, Yang SC, Heitmiller RF. Standardized
clinical care pathways for major thoracic cases reduce hos-
pital costs. Ann Thorac Surg 1998;66:914–9.
DISCUSSION
DR NICHOLAS G. SMEDIRA (Cleveland, OH): Mister Chair-
man, members of the Society, and guests. Dr Weiss, I enjoyed
your presentation and I enjoyed reading your concise, extremely
well written, and what I am sure will be an extremely contro-
versial manuscript. And before I ask a few questions and make
a final comment, I would like to recognize Dr Weiss for his
impressive accomplishments.
Dr Weiss obtained his medical degree from and completed his
general surgery training at Johns Hopkins University, and since
2006 he has been the Irene Piccinini Investigator in cardiac
surgery, completing the investigation we just saw and two other
investigations, one on cystic fibrosis, which will be presented at
IHLST, as the lead author, and another one at the American
Association for Thoracic Surgery. Doctor Weiss, congratulations
on your productivity.
Let me focus on the two concerns about outcome assessments,
both of which we heard about in the first talk this morning, and
these are risk adjustment for patient characteristics and the high
variability in outcomes among low-volume centers.
Let’s look at risk adjustment. It appears from the data that
patients in the low-volume centers were at lower risk: they were
younger, fewer intensive care unit patients, fewer intra-aortic
balloon pumps were used. However, some variables that I am
familiar with that impact transplant risk such as active use of
short- or long-term mechanical circulatory support or an under-
lying diagnosis of congenital heart disease did not appear to be
examined. How confident are you that the UNOS data provided
valid and accurate information for risk assessment?
It is known, as was discussed earlier, that patient risk factors
directly impact complication rates, failure to rescue from com-
plications, and mortality. However, hospital characteristics such
as high nurse-to-bed ratios, the presence of residents or a
teaching facility, and large hospital size show a divergence, with
more complications but better recovery from complications and
lower mortality. To help us understand the mechanism for
higher mortality at lower-volume centers, have you thought of
and could you incorporate hospital characteristics or, as you call
them, “processes of care,” into your model?
Let’s focus on variability. As you noted, there is huge variabil-
ity in outcomes in low-volume centers. This was shown in your
talk and the paper. There are many low-volume centers with
outstanding results. As you propose these regulatory changes,
we must keep in perspective that only 10% of the transplants
done were at low-volume centers and nearly 50% of centers
would have fit the low-volume criteria. How will you respond to
the angry words of the low-volume centers with consistently
good or great results when we shut them down for failing to
meet the 10 or 14 patients per year requirement, and can you
foresee any unanticipated consequences, as I can, such as listing
patients that might not necessarily need to be listed just to fill
the list or performing a transplant just meet numbers?
You take great effort to find a scientific way to determine
minimum center volume and recommend going from 10 to 14
based on a 0.2 difference in the C-value. In general, a C-value of
greater than 0.7 is needed for a good fit. You don’t give
confidence intervals for your C-statistic. Can you justify statis-
tically the recommendation that intuitively looks a little bit
silly—increasing the requirement from 10 to 14 transplants per
year?
And finally, I will leave you with my thoughts. I agree in
principle with your thesis and think very low volume centers
should be shut down if they cannot increase their volume and
improve results; however, we need to be very careful defining
centers of excellence based on volume alone. Volume determi-
nations are weak and highly variable surrogates for excellent
care. In addition to volume, I would add observed to expected
mortality, acknowledging the problem with small sample size in
these cases, and utilize morbidity and process measurements
similar to that provided by The Society of Thoracic Surgeons in
conjunction with the National Quality Forum in our nontrans-
plant cases. Finally, we must consider ease of access. Expecting
a patient to travel hundreds of miles for a transplant may in the
end be worse than going to a low-volume center.
I enjoyed your presentation. I think this will be a very
controversial manuscript, and thank you for this opportunity.
DR WEISS: Thank you very much, Dr Smedira, for your
thoughtful critique and also your kind remarks. I certainly
appreciate your comments. Let me answer your questions
individually.
In terms of the robustness of the risk assessment, I agree that
it would be nice to incorporate mechanical circulatory support
as well as diagnoses. Some of these variables are not well
represented in the UNOS database, and so we did the best we
could with what we had. Ventricular assist device support is
unfortunately one of these unrepresented variables. We were
able to examine 75 preoperative variables, of which 13 were
significant predictors of mortality on univariate analysis. We
then incorporated these variables into our multivariable model.
Our C-index for that multivariable analysis was 0.68, which is
consistent with other reports from the literature, and I think
reasonable for a large retrospective multi-institutional database
like this. It is not perfect, but I think we did our best given the
limitations of our data.
I think that processes of care are very important, as you
mentioned. From our analysis, only 2% of the variability in the
data is explained by volume. So other processes of care, such as
the presence of ancillary staff such as respiratory therapists,
specialized providers such as cardiac anesthesiologists, resi-
dents and fellows, nurse-to-patient ratio, dedicated intensive
care unit intensivists, and many other factors are extremely
important. Most of these factors are unfortunately not present in
the data set. We would of course have liked to include them. I
think only through a multi-institutional study where we can
specifically design what variables we want to examine is that
achievable.
Your third question was about how to respond to low-volume
centers. This is a key point and I thank you for bringing it up. A
clear problem is that not all low-volume centers have bad
outcomes, and I think it is important to tease out what factors
contribute to a center having good or bad outcomes. Clearly,
volume is not everything. I think centers that have low volumes
should be monitored, and if there is high mortality, then
enforceable penalties should exist. However, low-volume cen-
ters with low mortality should be allowed to continue to thrive.
1259
Ann Thorac Surg
2008;86:1250–60
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
ADULT CARDIAC

Page 11

I think that identifying additional processes of care will be
important for helping low-volume centers that currently do not
have good outcomes to improve.
And finally, I will just say that in terms of the difference
between 10 and 14, our point was more for proof of concept than
anything else. Our intention was to show that a volume cutoff
derived from statistical modeling is better than one that is
arbitrarily defined. And so we did our best to use a model to try
to define that cutoff of 14. But I wouldn’t put tremendous stock
into it; as you mentioned, the variability differences are very low.
I think it just reinforces the notion that volume is very important
in orthotopic heart transplantation and that centers that perform
a large volume generally have better outcomes.
Thank you very much again for your thoughtful critique.
The Society of Thoracic Surgeons Policy Action Center
The Society of Thoracic Surgeons (STS) is pleased to
announce a new member benefit—the STS Policy Action
Center, a website that allows STS members to participate
in change in Washington, DC. This easy, interactive,
hassle-free site allows members to:
● Personally contact legislators with one’s input on key
issues relevant to cardiothoracic surgery
● Write and send an editorial opinion to one’s local media
● E-mail senators and representatives about upcoming
medical liability reform legislation
● Track congressional campaigns in one’s district—and
become involved
● Research the proposed policies that help—or hurt—
one’s practice
● Take action on behalf of cardiothoracic surgery
This website is now available at www.sts.org/takeaction.
1260
WEISS ET AL
EFFECT OF CENTER VOLUME ON TRANSPLANT OUTCOME
Ann Thorac Surg
2008;86:1250–60
© 2008 by The Society of Thoracic Surgeons
Published by Elsevier Inc
Ann Thorac Surg 2008;86:1260• 0003-4975/08/$34.00
ADULT CARDIAC