2. The Journal of Thoracic and
Cardiovascular Surgery Rao et al. 39
Volume 112, Number 1
may represent a failure of myocardial protection or Table I. Patient population
difficulty with bypass grafting and is often associated N %
with a higher mortality and prolonged hospital stay. N 4558
A recent randomized clinical trial done at the Sex
University of Toronto by The W a r m H e a r t Investi- Male 3673 81
gators showed no difference in operative mortality Female 885 19
or perioperative myocardial infarction (MI) be- Age (yr) 61.5 _+ 9.6
Timing
tween patients who received w a r m blood or cold Elective 2620 57
blood cardioplegic solution. 2 However, the occurrence Semielective 1130 25
of low cardiac output syndrome was significantly lower Urgent 657 14
in the warm group. Similarly, a large clinical trial that Emergency 151 4
evaluated acadesine as a myocardial protective agent Recent MI 692 15
LVEF
failed to show any difference in the prevalence of >60% 1530 34
perioperative infarction except in a subgroup of pa- 40%-60% 1926 42
tients with high-risk conditions. 3 Because of the low 20%-40% 956 21
rates of operative mortality and perioperative MI, <20% 146 3
studies aimed at developing improved strategies of NYHA class
I 76 2
myocardial protection are unlikely to show a benefit II 654 14
unless they include a large number of patients or limit III 1861 41
their focus to subgroups of patients at high risk. Both IV 1967 43
operative mortality and a new perioperative MI occur CAD
infrequently and these outcome measures are difficult ] vessel 308 7
2 vessel 1083 24
to use in a trial of alternate myocardial protective 3 vessel 3159 69
strategies even in patients at high risk. Left main CAD 808 18
Low cardiac output syndrome is a more frequent Repeat operation 335 7
event and may be modified by improved methods of Diabetes 1086 24
myocardial protection in high-risk subgroups. How- Hypertension 2337 51
PVD 591 13
ever, the syndrome must be carefully defined and its
pathophysiologic components may be better under- LVEF, Left ventricular ejection fraction; NYHA, New York Heart Asso-
ciation; CAD, coronary artery disease; PVD, Peripheral vascular disease.
stood by assessment of the predisposing risk factors.
To make comparisons between studies, the defini-
tion of low cardiac output syndrome must become artery stenoses, the mean arterial pressure was main-
standardized. This study presents the independent tained higher than 70 mm Hg in an attempt to improve
cerebral perfusion. Blood cardioplegic solutions were
preoperative risk factors for the development of low
delivered after oxygenated blood from the bypass circuit
cardiac output syndrome in 4558 consecutive pa- was mixed with a crystalloid solution in a 4:1 ratio. Blood
tients undergoing isolated coronary artery bypass cardioplegic solution was administered either in an ante-
grafting at The Toronto Hospital. grade fashion via the aortic root or in a retrograde fashion
via the coronary sinus.
In all patients, the heart was arrested with an aortic root
Methods
infusion of high potassium (27 mEq/L) blood cardioplegic
Patient population. Preoperative, perioperative, and solution. Cardioplegia was maintained with a low potas-
postoperative data were collected prospectively on all sium formulation (15 mEq/L). Proximal and distal anas-
patients undergoing isolated coronary artery bypass graft- tomoses were constructed during a single prolonged cross-
ing between July 1, 1990, and December 31, 1993, at The clamp period. 4 A left internal thoracic artery graft was
Toronto Hospital. The preoperative characteristics of the used in 3789 (83%) patients. Cardiac output was mea-
4558 consecutive patients are shown in Table I. sured with a thermodilution catheter placed percutane-
Operative technique. After median sternotomy and ously via the internal jugular vein into the pulmonary
heparinization, cardiopulmonary bypass was established artery. 5
with a single two-stage right atrial cannula and an ascend- Patients in whom weaning from cardiopulmonary by-
ing aortic cannula. During bypass, the hematocrit was pass was ditficult or in whom inadequate cardiac perfor-
maintained between 20% and 25%, pump flow rates mance developed in the intensive care unit had an in-
between 2.0 and 2.5 L/min per square meter, and mean traaortic balloon pump (Datascope Corporation, Paramus,
arterial pressures between 50 and 60 mm Hg by use of N.J.) inserted percutaneously via the common femoral ar-
sodium nitroprusside or phenylephrine hydrochloride as tery. Patients with less severe hemodynamic compromise
required. In elderly patients or in patients with carotid received inotropic medication.
3. The Journal of Thoracic and
40 Rao et al. Cardiovascular Surgery
July 1996
Study design. This study represents a retrospective t tests. Logistic models for each outcome variable were
analysis on data gathered in a prospective fashion and constructed with the use of methods described by Hosmer
included in a database registry. Multivariable analyses and Lemeshow. 9 Each prognostic variable was carefully
were used to determine the independent predictors of evaluated by the appropriate univariate test. Variables
outcomes. Comparisons of baseline, operative, and post- were selected for inclusion in a multivariable model if
operative data were made between patients in whom the their univariatep value was less than 0.25 or if the variable
outcome of interest occurred and patients in whom it did was of known clinical importance but failed, univariately,
not. to achieve the critical alpha level.
Definitions of variables. Appendix A gives a definition Models were fit and the best model for each outcome
of all preoperative variables. variable was determined by (1) an examination of the
Study outcomes. Low cardiac output syndrome was Wald statistic for each variable 9 and (2) a comparison of
diagnosed if the patient required an intraaortic balloon each estimated coefficient with the coefficient from the
pump either in the operating room or in the intensive care univariate model that contained only that variable. Coef-
unit because of hemodynamic compromise. Patients who ficients that changed markedly in magnitude indicated
had a balloon pump inserted preoperatively because of that one or more of the excluded variables were important
either ischemic chest pain or hemodynamic dysfunction in the adjustment of the effect of the remaining variables
were believed to have a postoperative low cardiac output in the model and an effort was made to refit the model. 7
syndrome if, in addition to the balloon pump, they also The next step in determining the best multivariable
required inotropic medication. Low cardiac output syn- model was to examine the goodness-of-fit statistic. Good-
drome was also diagnosed if the patient required inotropic ness of fit assessed the effectiveness of the model in
medication to maintain the systolic blood pressure greater describing the outcome variable. In addition, differences
than 90 mm Hg and the cardiac output greater than 2.2 between the observed data and the estimated values (the
L/rain per square meter for at least 30 minutes in the residual) for each covariate pattern were calculated by the
intensive care unit after correction of all electrolyte or Pearson or Hosmer-Lemeshow X2 statistic.9 The null
blood gas abnormalities and after adjusting the preload to hypothesis for goodness of fit claims that there are no
its optimal value. 6 Afterload reduction was also attempted significant differences between the predicted outcomes
when possible. Patients received either dopamine hydro- and the observed data. Therefore a probability greater
chloride, dobutamine hydrochloride, amrinone, or epi- than 0.05 indicates acceptance of the null hypothesis and
nephrine. We believe that prolonged treatment with ino- a valid model.
tropic medication may augment perioperative ischemic The final approach to determining the best model for
injury]' 8 Therefore inotropic medication was avoided each outcome variable was to examine the receiver oper-
when possible and was used only in patients who had mild ator characteristic (ROC) curve for each model. ROC
and transient hemodynamic compromise. An intraaortic curves are usually used to evaluate and compare an
balloon pump was inserted in patients who had moderate operator or diagnostic test with a "gold standard" and to
or severe hemodynamic compromise. Patients who re- explore the trade-offs between sensitivity and specificity
ceived less than 4 pg/kg of dopamine to increase renal for a test. a°' 11 Tests that discriminate well will crowd the
perfusion were not considered to have low cardiac output curve toward the upper left corner. The overall accuracy
syndrome. Patients who received vasoconstricting medica- of a test can be described as the area under the curve;
tion because of a high cardiac output (->2.5 L/rain per increasing the area under the curve corresponds to a
square meter) and low peripheral resistance were not better test because it optimizes the sensitivity and speci-
considered to have low cardiac output syndrome. ficity.x2, 13 When calculated in the BMDP LR program
Operative mortality and MI. Operative mortality was (BMDP Statistical Software Inc., Los Angeles, Calif.), n
defined as any death that occurred within 30 days of the ROC curve is independent of both the cut-point
operation or during the same hospital admission. A criteria (predicted probabilities) and the prevalence of the
perioperative MI was documented when a new Q wave outcomes. This independence allows comparison of the
was found on the postoperative electrocardiogram. An MI ROC area of different study populations where sensitivity
was also diagnosed if the postoperative electrocardiogram and specificity would be distorted by differences in the
had a new left bundle branch block, loss of R wave prevalence of the outcomes of interest. Therefore we used
progression, or new ST and T wave changes if accompa- these curves to provide an additional "diagnostic" tool to
nied by a rise in the level of the MB isoenzyme of creatine determine the optimum model for our binary outcome
kinase (CK-MB) greater than 50 U/L and if the CK- variables.
MB/CK ratio was greater than 5%. An antibody inhibition The internal validity of the model was assessed by use of
technique was used to measure the CK-MB level. The a bootstrap method in which the regression coefficients for
highest postoperative CK-MB value was recorded and the entire data set were correlated with the regression
expressed as a fraction of total CK if a perioperative coefficients of a test data set. The test data set was derived
ischemic event was suspected clinically. This definition of as a subpopulation of the entire data set.
perioperative MI requires electrocardiographic changes
and therefore may underestimate this outcome. Results
Statistical analysis. Statistical analysis was done with
the SAS program (SAS Institute, Cary, N.C.). Categorical T h e p r e o p e r a t i v e characteristics o f the e n t i r e
data were analyzed by a )~2 or Fisher's exact test where p a t i e n t p o p u l a t i o n a r e listed in T a b l e I. T h e overall
appropriate. Continuous data were analyzed by two-tailed m o r t a l i t y r a t e was 2.4% (n = 109). L o w c a r d i a c
4. The Journal of Thoracic and
Cardiovascular Surgery Rao et aL 41
Volume 112, Number 1
output syndrome developed in 412 patients (9.1%). Table II. Operative data
Table II compares the operative data for patients in P
whom low cardiac output syndrome developed with No LOS LOS Value
data for those patients in whom it did not. Patients No. of patients 4146 (91%) 412 (9%)
in whom low cardiac output syndrome developed Age (yr) 61.3 ± 9.6 63.6 ± 9.6 <0.001
were older (64 versus 61 years, p < 0.001) and had Diseased vessels
more extensive coronary artery disease (p < 0.001). 1 286 (7%) 22 (5%) 0.0012
2 1011 (24%) 72 (18%)
There were no significant differences in the number 3 2844 (69%) 315 (77%)
of bypass grafts constructed between the two groups. Grafts
Complete revascularization was accomplished in 1 72 (2%) 10 (2%) 0.660
4193 (92%) patients. The operative mortality rate in 2 482 (12%) 51 (12%)
patients in whom complete revascularization was 3 1440 (35%) 138 (33%)
4 1778 (43%) 177 (43%)
achieved was 2.3% compared with 3.9% in patients 5 374 (9%) 36 (9%)
in whom revascularization was not complete (p = Pump time (min) 85 ± 24 110 ± 39 <0.001
0.068). The prevalence of low cardiac output syn- Crossclamp time 60 ± 18 67 ± 27 <0.001
drome was 8.1% in patients in whom revasculariza- (rain)
tion was complete versus 14.6% in patients in whom Days of ventilator 1.1 _+ 2.5 2.9 ± 5.0 <0.001
support
it was not (p < 0.001). Days in ICU 2.2 -- 3.3 5.0 ± 6.3 <0.001
The left anterior descending artery (LAD) terri- Days in hospital 9.9 _+ 9.2 13.8 +- 12.4 <0.001
tory was revascularized in 99.6% of the patients with CK (units) 897 _+ 769 1429 -+ 1309 <0.001
LAD disease. Similarly, the territories of the cir- CK-MB (units) 41 + 37 83 + 102 <0.001
cumflex and right coronary arteries were revascular- Percent CK-MB 5.9 ± 9.9 6.7 ± 6.4 <0.001
Periop. MI 74 (1.8%) 59 (14.3%) <0.001
ized in 96.2% and 94.7% of the respective patients Stroke 47 (1.1%) 14 (3.4%) <0.001
with disease in those territories. Mortality 39 (0.9%) 70 (16.9%) <0.001
Patients in whom low cardiac output syndrome
LOS, Low cardiac output syndrome; ICU,, intensive care unit.
developed had longer cardiopulmonary bypass
times, longer aortic crossclamp times, a longer post-
operative intensive care unit stay, more days of (10%, OR 1.3). Table IV presents the regression
ventilatory support, a longer hospital stay, and a coefficients, their standard errors derived from the
higher postoperative CKMB level. Patients in whom logistic regression analysis, the ORs, the 95% con-
low cardiac output syndrome developed had a fidence intervals (95% CIs) for the ORs, the im-
higher mortality rate (17%) than patients in whom it provement X2 p value, and the goodness-of-fit p
did not develop (1%, p < 0.001). Patients in whom value for the nine independent predictors. The
low cardiac output syndrome developed were more predictive probability of the development of low
likely to have a perioperative MI (14.3% versus cardiac output syndrome can be calculated by the
1.8%,p < 0.001). formula P = e~/(1 - e*), where x is the sum of the
Predictors of low cardiac output syndrome. Figs. regression coefficients (see Appendix B). Fig. 3
1 and 2 and Table III illustrate the univariate results depicts the predicted probability of low cardiac
for the multivariable predictors of low cardiac out- output syndrome (abscissa) versus the logit score
put syndrome. Stepwise logistic regression analyses (ordinate) for several combinations of covariate
identified nine independent predictors of postoper- patterns for low cardiac output syndrome. This
ative low cardiac output syndrome (percentage in figure can be used to determine the probability of
whom low cardiac output syndrome developed in low cardiac output syndrome for an individual pa-
parentheses) and the factor-adjusted odds ratio tient.
(OR) associated with each predictor: (1) left ven- There were 133 patients (2.9%) who had a peri-
tricular ejection fraction less than 20% (27%, OR operative MI. The operative mortality rate was
5.7); (2) repeat operation (25%, OR 4.4); (3) emer- 21.8% in this group of patients compared with 1.8%
gency operation (27%, OR 3.7); (4) female gender in patients who did not have an infarct. Patients with
(16%, OR 2.5); (5) diabetes (13%, OR 1.6); (6) age low cardiac output syndrome had a significantly
older than 70 years (13%, OR 1.5); (7) left main higher prevalence of perioperative MI (14.3% ver-
coronary artery disease (12%, OR 1.4); (8) recent sus 1.8%, p < 0.001). Conversely, patients who had
MI (16%, OR 1.4); and (9) triple-vessel disease a perioperative MI had a 44% prevalence of low
5. The Journal of Thoracic and
42 R a o et al. Cardiovascular Surgery
July 1996
35 [ . P < 0.001
• LOS [] OM
[
30 I *
27 27
25
z 20
15 15
10 8
0
1 2 3 4 N Y 2 3
LV G R A D E REDO TIMING
20
*p< 0.001
• LOS [] OM +p< 0.05
16
15
+
13 13
12
10
7 +
5
0
M F N Y N Y N Y
GENDER DIABETES AGE>70 LEFT MAIN
Fig. 1. Univariate results of multivariable predictors of low cardiac output syndrome (LOS) and operative
mortality (OM). Left ventricular grade (LV GRADE) designated by 1, ejection fraction 60%; 2, ejection
fraction 40% to 59%; 3, ejection fraction 21% to 39%; or 4, ejection fraction 20%. Repeat operation
(REDO), that is, previous aorta-coronary bypass, noted as yes (Y) or no (N). Timing of operation
designated by 1, elective operation; 2, operation during same hospitalization as cardiac catheterization or
cardiac event (semielective); or 3, urgent operation within 72 hours of cardiac event. Gender noted as male
(M) or female (F). Diabetes; age younger than 70 years; and left main coronary artery disease (LEFT
MAIN), that is, significant (greater than 50%) stenosis of left main coronary artery, noted as yes or no.
6. The Journal of Thoracic and
Cardiovascular Surgery Rao et aL 43
Volume 112, Number 1
'° I , • LOS [ ] OM I .po.oo,
+p< 0.01
15 8 16
i 10 +
10 l0 I
8
5 + +
N Y N Y N Y
RECENT MI TRIPLE VESSEL DISEASE HYPERTENSION
Fig. 2. Univariate results of multivariable predictors of low cardiac output syndrome (LOS) and operative
mortality (OM). Recent MI, that is, MI within 30 days before operation; triple-vessel disease; and
hypertension noted as yes (Y) or no (N).
cardiac output syndrome compared with 8% in p < 0.001); in patients with diabetes (3.8% versus
patients who did not have an infarct. 2.0%, p = 0.001); in female patients (3.8% versus
Operative mortality. There were 109 operative 2.0%,p = 0.002); in patients with left main coronary
deaths in this population. Among patients in whom artery disease (3.6% versus 2.1%, p = 0.014); in
low cardiac output syndrome developed (n = 412) patients with hypertension (3.0% versus 1.8%, p =
there were 70 deaths (17%) compared with 39 0.006); in patients who had an MI less than 30 days
deaths (0.9%) in those in whom low cardiac output before operation (4.0% versus 2.1%,p = 0.002); in
syndrome did not develop (n = 4146). The mean patients with chronic obstructive pulmonary disease
postoperative length of stay for patients who died (3.4% versus 2.2%, p = 0.036); and in patients with
after the development of low cardiac output syn- New York Heart Association class IV disease (3.6%
drome was 7.4 _+ 11.3 days (median 3.5 days, range versus 1.3% in New York Heart Association class I,
0 to 53 days). The mean postoperative length of stay p = 0.002).
for patients who died without having the develop- The multivariable predictors of operative mortal-
ment of low cardiac output syndrome was 23.1 ___ ity were (1) left ventricular ejection fraction less
26.6 days (median 12.5 days, range 0 to 84 days). than 20% (OR 8.1); (2) repeat operation (OR 4.9);
The operative mortality rate was significantly (3) peripheral vascular disease (OR 2.8); (4) age
higher by univariate analysis in patients with a left older than 70 (OR 2.8); (5) emergency operation
ventricular ejection fraction less than 20% (10.9% (OR 2.7); (6) diabetes (OR 1.7); (7) female gender
versus 1.2% with left ventricular ejection fraction (OR 1.7); (8) left main coronary artery stenosis (OR
greater than 60%, p < 0.001); in patients undergo- 1.5); and (9) hypertension (OR 1.4). Table V pre-
ing repeat operation (7.5% versus 2.0%,p < 0.001); sents the regression coefficients, their standard er-
in patients with peripheral vascular disease (6.6% rors derived from the logistic regression analysis, the
versus 1.8%, p < 0.001); in patients older than age ORs, the 95% CIs for the ORs, the improvement X2
70 (5.0% versus 1.1% in patients younger than age p value, and the goodness-of-fit p value for the nine
50, p < 0.001); in patients undergoing emergency independent predictors. The 95% CIs for left main
operation (6.6% versus 1.7% in elective operation, coronary artery stenosis and hypertension both in-
7. The Journal of Thoracic and
44 Rao et al. Cardiovascular Surgery
July 1996
Table III. Univariate analysis for low cardiac output syndrome and operative mortality
LOS No LOS Mortality
P Value p Value
N % N % (LOS) N % (034)
No. of patients 412 9 4146 91
LVEF
>60% 87 6 1443 94 <0.001 19 1 <0.001
40%-60% 146 8 1780 92 44 2
20%-40% 139 15 817 85 30 3
<20% 40 27 106 73 16 11
Repeat operation 82 25 253 75 <0.001 25 8 <0.001
No Repeat 330 8 3893 92 84 2
Timing <0.001 <0.001
Elective 169 6 2451 94 45 2
Semielective 112 10 1018 90 35 3
Urgent 90 14 567 86 19 3
Emergency 41 27 110 73 10 7
Sex
Male 271 7 3402 93 <0.001 75 2 0.002
Female 141 16 744 84 34 4
Diabetes 137 13 949 87 <0.001 41 4 0.001
No diabetes 275 8 3197 92 68 2
Age
<70 283 8 3261 92 <0.001 58 2 <0.001
->70 129 13 885 87 51 5
Left main CAD
Y 94 12 714 88 0.005 29 4 0.014
N 318 9 3432 91 80 2
Recent MI
Y 112 16 580 84 <0.001 28 4 0.002
N 300 8 3566 92 81 2
Triple-vessel CAD
Y 315 10 2759 90 0.002 80 3 0.651
N 94 7 1391 93 29 2
Hypertension 230 10 2107 90 0.053 70 3 0.006
No hypertension 182 8 2039 92 39 2
LOS, Low cardiac output syndrome;OM, operative mortality; LVEF, left ventricular ejection fraction; Y, yes; N, no; recent MI, MI within 30 days before
operation; CAD, coronary artery disease.
I
Table IV. Multivariable analysis for low cardiac output syndrome
Regression Improvement GOF
Variable coefficient SE OR 95% CI X2 p value p value
Constant -3,866 0.166
LVEF
40%-60% 0.1777 0.146 1.19 0.89-1.59
20%-40% 0.8614 0.153 2.37 1.75-3.19
<20% 1.7410 0.232 5.70 3.62-8.99 <0.001 0.031
Repeat operation 1.4770 0.151 4.38 3.26-5.89 <0.001 0.277
Timing of operation
Urgent 0.2451 0.128 1,28 0.99-1.64
Emergency 1.3120 0.237 3.72 2.34-5.91 <0.001 0.966
Female gender 0.9287 0.122 2.53 1.99-3.22 <0.001 0.825
Diabetes 0,4644 0.119 1.59 1.26-2.01 <0.001 0.983
AGE ->70 0.3691 0.122 1.45 1.14-1.84 0.001 0.990
Left main CAD 0,3500 0.133 1.42 1.09-1.84 0.011 0.993
Recent MI 0.3241 0.148 1.38 1.04-1.85 0.023 0.994
Triple-vessel disease 0.2841 0.130 1.33 1.03-1.71 0.026 0.996
SE, Standard error; GOF, goodness of fit; LVEF, left ventricular ejection fraction; CAD, coronary artery disease; Recent MI; MI within 30 days before
operation.
8. The Journal of Thoracic and
Cardiovascular Surgery Rao et al. 45
Volume 112, Number 1
%
90 [] 95% Confidence Interval
80
70
60
50-
40-
30-
20-
10-
0
-4
Logit Score
LVGRADE 1 2 2 12 1 1 2 33 1 3 32 332 3 3 42 4323 3 3 34 4 4 4
ACBREDO 0 0 0 00 0 0 0 00 0 0 01 001 1 1 110110 10 1101 0
SEX MM MMMM FMMM FM FMMFMM MMM F F M F M FMM FM F
TIMING E E S S E S S S SES S EESSS E S EUESUU SU USUU U
DIABETES 0 0 0 0 0 1 0 0 1 100111001 0 001011 1 1 110 0 1
A G E > 70 0 0 0 1 0 0 0 0 0 11 0001110 1 000100 1 1 01 I 0 1
L MAIN DISEASE 0 0 0 0 0 0 0 1 0 001001110 0 011000 1 0 0010 0
RECENTMI 0 0 0 01 0 0 ! 0 01 1 000000 1 000011 1 1 111 1 1
TVD 0 0 0 01 1 0 1 0 001 011011 0 011011 1 1 I101 1
Fig. 3. Predictive probability of low cardiac output syndrome after coronary artery bypass grafting. Left
ventricular grade (LV GRADE) scored from i to 4. Repeat aorta-coronary bypass (ACB REDO), diabetes,
age older than 70 years, left main coronary artery disease (L MAIN DISEASE), recent MI, and triple-vessel
disease (TVD) scored 0 for no, 1 for yes. M, Male; F, female; E, elective; S, semielective; U, urgent.
Table V. Multivariable analysis for operative mortality
Regression Improvement GOF
Variable coefficient SE OR 95% CI Xe p value p value
Constant -5.760 0.316
PVD 1.0330 0.215 2.81 1.84-4.28 <0.001 1.000
LVEF
40%-60% 0.4847 0.284 1.62 0.93-2.83
20%-40% 0.7036 0.306 2.02 1.11-3.68
<20% 2.0920 0.371 8.10 3.91-16.80 <0.001 0.448
Repeat operation 1.5810 0.251 4.86 2.97-7.95 <0.001 0.357
Age >-70 1.0250 0.208 2.79 1.85-4.19 <0.001 0.439
Female gender 0.5178 0.227 1.68 1.08-2.62 0.004 0.214
Diabetes 0.5560 0.212 1.74 1.15-2.64 0.009 0.637
Timing of operation
Urgent 0.1317 0.217 1.14 0.75-1.75
Emergency 0.9907 0.390 2.69 1.25-5.79 0.039 0.230
Left main CAD 0.4341 0.232 1.54 0.98-2.43 0.059 0.660
Hypertension 0.3542 0.212 1.43 0.94-2.16 0.091 0.682
SE, Standard error; GOF, goodness of fit; PVD, peripheral vascular disease; LVEF, left ventricular ejection fraction; CAD, coronary artery disease.
clude unity, which indicates that they are weak The area under the ROC curve for low cardiac
predictors of operative mortality. output syndrome is similar to that for operative
Fig. 4 illustrates the predicted operative mortality mortality (74% versus 76%), which indicates that
for all combinations of the nine independent pre- both models are similar in terms of their sensitiv-
dictors. Fig. 5 depicts the R O C curves for both low ity and specificity to detect their respective out-
cardiac output syndrome and operative mortality. comes.
9. The Journal of Thoracic and
46 Rao et aL Cardiovascular Surgery
July 1996
%
95% Confidence Interval
70
60
50
40
30
20
10
0
-6 -4 -2 0
Logit Score
PVD 00 0 0 000 10100000001100011101011 00111 1 1 1 1
LVGRADE 11 1 1 13 1 11113221322 1324412424'33 34344 2 3 3 4
ACB REDO 00 0 0 000 00000001000001001010001 10100 1 1 0 1
AGE > 70 00 0 0 000 00011110111 110000100010 1111 0 1 1 1 0
SEX MMM FMMMMMMMMMMMMMMFMMMMMFMFFMMMFMMFMF FM
DIABETES 00 0 0 101 01000001100010100000110 10010 1 0 1 0
TIMING ES E E SSE SEESEESEESE SSSSSESEUEES SSESU S S U U
L MAIN DISEASE 00 0 0 001 01110000010011001001111 11000 0 0 1 0
HYPERTENSION 00 1 0 001 010001100100011011011 11 01101 1 1 1 0
Fig. 4. Predictive probability of operative mortality after coronary artery bypass grafting. Peripheral
vascular disease (PVD), repeat aorto-coronary bypass (ACB REDO), age older than 70 years, diabetes, left
main coronary artery disease (L MAIN DISEASE), and hypertension scored 0 for no, 1 for yes. Left
ventricular grade (LV GRADE) scored from 1 to 4. M, Male; F, female; E, elective; S, semielective; U,
urgent.
Discussion sures impractical in modern studies of myocardial
An increasing number of patients with high-risk protection.
conditions are undergoing coronary artery bypass For example, to detect a 20% reduction in the
grafting. 1 The extension of cardiac operation to operative mortality rate from 2% to 1.6%, one
patients at higher risk is a result of improved would need to study 8504 patients to achieve a 5%
operative techniques and perioperative myocardial level of significance with a power of 80%. To detect
protection. Cardiologists and surgeons have ex- a 20% reduction in the prevalence of perioperative
tended the benefits of coronary artery bypass to MI from 3% to 2.4%, one would require 5618
patients at higher risk as the risks of operation have patients. A similar 20% decrease in the prevalence
decreased) 4 To continue to reduce perioperative of low cardiac output syndrome from 10% to 8%
morbidity and mortality, cardiac surgeons must de- would require 1577 patients to achieve a 5% level of
vise strategies to improve myocardial protection in significance. Patients in whom low cardiac output
patients at high risk. syndrome develop have a significantly higher prev-
Traditionally, the results of coronary artery by- alence of perioperative MI and a higher operative
pass have been evaluated by operative mortality and mortality. Thus the development of low cardiac
perioperative MI. 14-16Low cardiac output syndrome output syndrome represents either inadequate re-
is another clinical outcome that can be used to vascularization or inadequate myocardial protection
assess the efficacy of perioperative myocardial pro- and may act as a marker of intraoperative myocar-
tection. Two large randomized clinical trials failed dial injury. Our results indicate that the territory
to show any difference in perioperative mortality or supplied by the LAD was revascularized in more
MI between the treatment groups. 2' 3 The low rates than 99% of patients with disease in that distribu-
of operative mortality and the problems inherent in tion. Revascularization was incomplete in the terri-
uniformly defining and identifying perioperative in- tory of the right coronary artery in 5% of the
farction have made the use of these outcome mea- patients with right coronary artery disease. These
10. The Journal of Thoracic and
Cardiovascular Surgery Rao et al. 47
Volume 112, Number 1
Low Output Syndrome Operative Mortality
Sensitivity (%) Sensitivity (%)
100 100-
90 90
80 80
70 70
60 60
50 50
40 40
30 30
20 20
10 10
0 OT
0 20 40 60 80 100 0 20 40 60 80 100
100 - Specificity (%) I00 - Specificity (%)
Fig. 5. ROC curves of predictive models for low cardiac output syndrome (left) and operative mortality
(right).
patients likely represent a subpopulation with a Poor ventricular function. Poor left ventricular
previous inferior MI and an occluded coronary function continues to be the most important predic-
artery. Despite having preserved left ventricular tor of postoperative morbidity and mortality. Pa-
function, these patients were still at risk for the tients with poor ventricular function have a limited
development of postoperative low cardiac output margin for myocardial protection. 16 However, the
syndrome. dysfunctional myocardium may not be irreversibly
Diagnosis. A diagnosis of low cardiac output damaged and may be "stunned" or "hibernating."
syndrome required the active intervention of the Revascularization of the reversibly injured heart
cardiac surgeon, and at our institution this interven- may result in improved left ventricular performance.
tion represented a failure of our usual perioperative Cold injury or inhomogeneous cardioplegic delivery
strategy. Thus the diagnosis of low cardiac output may exacerbate perioperative ischemic injury and
syndrome was a reproducible clinical outcome. The result in inadequate early postoperative ventricular
Warm Heart Investigators established an indepen- function. 17 Prolonged reperfusion with a terminal
dent committee to evaluate postoperative low car- "hot shot" of cardioplegic solution may restore
diac output syndrome. 2 Their criteria for diagnosis function in patients with poor ventricular function, is
were similar to ours and again support the concept Warm cardioplegia may improve postoperative left
that this syndrome can be used as an objective ventricular function in patients with high-risk con-
measurement of perioperative myocardial injury. ditions, inasmuch as we have previously shown that
This study identified nine independent preopera- warm cardioplegia improves ventricular function in
tive predictors of low cardiac output syndrome after patients at low risk undergoing elective opera-
coronary artery bypass grafting (Figs. 1 and 2). The tions. 17 Unfortunately, some patients will continue
potential causes for inadequate postoperative car- to have poor ventricular function after operation
diac performance are not well established. and the role of myocardial protection in these
11. The Journal of Thoracic and
48 Rao et aL Cardiovascular Surgery
July 1996
patients may be to limit the extent of perioperative warm (37° C) cardioplegia and improved left ven-
injury. tricular function compared with cold (10 ° C) blood
Reoperation. Patients undergoing repeat opera- cardioplegia. 22
tions represent a challenge for intraoperative man- Female gender. The Coronary Artery Surgery
agement. Patients undergoing reoperation have Study investigators reported female gender to be
more diffuse disease and are at risk of having a associated with higher perioperative morbidity and
shower of emboli from their previous bypass grafts. mortality. 15 The authors postulated that this in-
The management of patent grafts is difficult. The creased morbidity was a result of the smaller size of
study by Lytle and associates 19 from the Cleveland the coronary vessels and the higher risk of graft
Clinic revealed that the operative mortality rate in thrombosis. A recent review from our institution 22a
patients undergoing repeat coronary artery bypass revealed that female patients had a higher preva-
was 4.3%. They found that patients with stenoses in lence of operative mortality and postoperative low
vein grafts to the LAD region had decreased sur- cardiac output syndrome. The predictors of opera-
vival. However, there were no in-hospital deaths tive mortality and postoperative low cardiac output
among patients with totally occluded vein grafts or syndrome were similar for both men and women.
patent internal thoracic artery grafts to the LAD Small body size was an independent risk factor for
region. The authors speculated that the retrograde postoperative morbidity. In patients of similar body
introduction of cardioplegic solution significantly size, female gender was an independent risk factor
lowered the operative mortality rate of repeat oper- for low cardiac output syndrome and operative
ation as a direct consequence of reducing athero- mortality.
sclerotic emboli from previous vein grafts. 19 Diabetes mellitns. Patients with diabetes may
Urgent operation. Patients who require urgent have more diffuse atherosclerotic disease, which
operation because of unstable angina may benefit may limit complete revascularization. In addition,
most from improved strategies of myocardial pro- patients with diabetes may have silent ischemia and
tection. Prolonged preoperative ischemia may de- be seen for operation after extensive MI or may
plete metabolic reserves. Substrate enhancement have diffuse coronary artery disease. Cardioplegic
with Krebs cycle intermediates 2°'21 may benefit protection may pose a problem in patients with
these patients. Rousou and associates 21 showed that diffuse coronary disease. Proximal coronary lesions
cardioplegic enhancement with Krebs cycle interme- may impair antegrade delivery of cardioplegic solu-
diates such as glutamate, malate, succinate, and tion and venovenous and thebesian shunting may
fumarate preserved high-energy phosphates during reduce the retrograde delivery of cardioplegic solu-
ischemic arrest but that this preservation did not tion. Homogeneous distribution of cardioplegic
extend to the reperfusion period. To date the effec- solution may improve intraoperative myocardial
tiveness of substrate-enhanced cardioplegia remains protection in these patients. A recent study by
controversial. The use of warm rather than cold Hayashida and associates 23 showed that a combina-
cardioplegia may help to resuscitate the ischemic tion of antegrade and retrograde cardioplegic solu-
myocardium. Cold cardioplegia reduces myocardial tion delivery may provide the best protection by
oxygen consumption and lactate production, but improving the distribution of cardioplegic solution.
delays the recovery of oxidative metabolism and Intermittent antegrade infusions after each period
ventricular function. Cold cardioplegia has the ad- of continuous retrograde cardioplegic solution de-
vantage of improved protection to areas that are livery resulted in the washout of accumulated lac-
difficult to perfuse because of coronary obstructions. tate. This finding suggested that the two directions
Warm cardioplegia may resuscitate the ischemic of cardioplegic solution delivery perfuse different
myocardium if it can be delivered uniformly and myocardial territories. Combining the two tech-
continuously. Discontinuation of normothermic niques may result in a more homogeneous distribu-
blood cardioplegic solution delivery to permit visu- tion of cardioplegic solution2
alization of the distal anastomosis may result in Advanced age. Elderly patients continue to be at
ischemic anaerobic metabolism. Perhaps the ideal high risk for postoperative complications (Fig. 2). At
cardioplegic temperature lies between the two ex- our institution an increasing proportion of patients
tremes. We have recently reported the use of "tep- older than age 70 are presenting for surgical treat-
id" (29 ° C) cardioplegia. 22 Tepid cardioplegia re- ment, but the prevalences of perioperative infarc-
duced lactate and acid production compared with tion and operative mortality have decreased during
12. The Journal of Thoracic and
Cardiovascular Surgery R a o et al. 49
Volume 112, Number 1
the past decade. 24 The elderly are at increased risk sonography to detect atherosclerotic plaques to
not only because of the obvious effects of aging but aid aortic cannulation has been advocated by
also because of the increased prevalence of comor- Barzilai and colleagues. 26
bid conditions in this population. Misare, Kruken- Models. Fig. 3 illustrates the predictive probabil-
kamp, and Levitsky25 showed an age-dependent ity of the development of low cardiac output syn-
sensitivity to myocardial ischemia in an ovine model. drome on the basis of data from our patient
These authors termed this phenomenon the senes- population. Fig. 4 illustrates the predictive prob-
cent myocardium. Thus, independent of other co- ability of operative mortality after aorta-coronary
morbid conditions, elderly patients may be at in- bypass.
creased risk for perioperative myocardial injury Greenland 27 concluded that both stratified (pop-
because of their senescent myocardium. ulations stratified on the basis of risk) and modeling
Left main coronary artery disease. The presence (predictive models derived from multivariable anal-
of left main coronary artery disease is no longer as yses) techniques to analyze populations have limita-
important a predictor of operative mortality as it tions based on the control of variable selection and
was previously.1 Although left main stenosis was the size and quality of data sets. Neither approach
selected as an independent predictor of operative can compensate for fundamental methodologic er-
mortality, the confidence interval for the OR in- rors such as misclassifications, selection bias, or lack
cludes one, which indicates that left main disease of statistical power to address the questions of
may not be a major predictor of operative mortality. interest. Greenland 27 therefore concluded that
However, in this study, left main coronary artery more effort should be put into correctly interpreting
disease was still an independent risk factor for the and intelligibly presenting modeling results to re-
development of low cardiac output syndrome. Im- flect these underlying errors. An evaluation of the
provements in myocardial protection may have com- ability of a model to correctly represent the under-
pensated for the increased risk in left main coronary lying data set is mandatory when presenting regres-
artery disease. The use of retrograde or combination sion analyses. The Pearson or Hosmer-Lemeshow
cardioplegic techniques may further reduce the im- statistics yield a measure of the goodness of fit for a
portance of left main disease on the outcome of low particular model and can be used to compare mod-
cardiac output syndrome. els derived from the same underlying data. The
Recent MI and triple-vessel disease. Patients goodness-of-fit p values for our final models of low
with an MI within 30 days of operation were at cardiac output syndrome and operative mortality
slightly higher risk for the development of low were greater than 0.99 and 0.65, respectively, and
cardiac output syndrome. Similarly, patients with suggested an excellent agreement between observed
triple-vessel disease were at higher risk for the and predicted values. The internal validity of our
development of low cardiac output syndrome. Al- models was evaluated with use of a bootstrap
though both of these risk factors were selected as method. The regression coefficients between the
independent risk factors for the development of low whole data set and the test data set correlated for
cardiac output syndrome, the confidence interval of both operative mortality (ra = 0.981,p < 0.001) and
their ORs approached one. Thus, although statisti- low cardiac output syndrome (r2 = 0.987,p < 0.001).
cally significant, these risk factors are weak predic- These results lend further support to the validity of
tors of low cardiac output syndrome. these models.
Hypertension and peripheral vascular disease. The ROC curves are another tool to evaluate the
Hypertension and peripheral vascular disease ability of a model to accurately represent the under-
were independent predictors of mortality but not lying data. The advantage of the ROC curve is that
of low cardiac output syndrome. These risk factors it is independent of the prevalence of the outcome
predispose patients to stroke and mortality may of interest and thus can be used to compare logistic
have been a result of noncardiac causes. There- models derived from different data sets with varying
fore these variables did not predict postoperative levels of incidence and prevalence. The areas under
low cardiac output syndrome. Strategies to mini- the ROC curve for the low cardiac output syndrome
mize the impact of these variables include im- and operative mortality models were 74% and 76%,
proved management of perioperative blood pres- respectively (Fig. 5). These figures can be used to
sure and minimal manipulation of the aorta compare our logistic models with other models
during operation. The use of intraoperative ultra- derived from a different patient population. The
13. The Journal of Thoracic and
50 Rao et al. Cardiovascular Surgery
July 1996
larger the area under the ROC curve, the more balloon counterpulsation during early reperfusion after isch-
accurate the model is in predicting outcomes in the emic arrest of the heart. J Thorac Cardiovasc Surg 1987;93:
597-608.
underlying data set. With use of these objective 9. Hosmer DW, Lemeshow S. Applied logistic regression. New
criteria, one can determine whether differences in York: John Wiley, 1989.
predictive models are true differences in the under- 10. Fletcher RH, Fletcher SW, Wagner EH. Clinical epidemiol-
lying data or a result of inaccuracies in a poorly fit ogy: the essentials. Baltimore: Williams & Wilkins, 1988.
model. Similarly, these objective criteria can be used 11. Dixon WJ, ed. BMDP statistical software manual, part 2.
Berkeley: University of California Press, 1992.
to compare models derived from different data sets 12. Rosenkranz ER, Vinten-Johansen J, Buckberg GD, Oka-
and patient populations. Thus differences in predic- moto F, Edward H, Bugyi H. Benefits of normothermic
tive variables between patient populations can once induction of blood cardioplegia in energy-depleted hearts,
again be ascribed to either true differences between with maintenance of arrest by multidose cold blood car-
the populations or to inaccuracies in poorly fit dioplegic infusions. J Thorac Cardiovasc Surg 1982;84:667-
77.
models. The use of such objective criteria will
13. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology: a
alleviate the problem of different multivariable basic science for clinical medicine. Boston: Little Brown,
equations arising from the same data set. 2s' 29 1985.
This study presents the independent predictors of 14. Teoh KH, Christakis GT, Weisel RD, et al. Increased risk of
low cardiac output syndrome in 4558 consecutive urgent revascularization. J Thorac Cardiovasc Surg 1987;93:
291-9.
patients who underwent isolated coronary artery
15. Fisher LD, Kennedy JW, Davis KB, et al. Association of sex,
bypass grafting at The Toronto Hospital. Our defi- physical size, and operative mortality after coronary artery
nition of low cardiac output syndrome is objective bypass in the Coronary Artery Surgery Study (CASS). J
and reproducible and can be used as an alternative Thorac Cardiovasc Surg 1982;84:334-41.
clinical outcome of interest when the efficacy of new 16. Christakis GT, Weisel RD, Fremes SE, et al, Coronary artery
bypass surgery in patients with poor ventricular function. J
myocardial protective strategies is evaluated.
Thorac Cardiovasc Surg 1992;103:1083-92.
17. Yau TM, Ikonomidis JS, Weisel RD, et al. Ventricular
We recognize the continued support of the cardiovas- function after normothermic versus hypothermic cardiople-
cular surgeons at The Toronto Hospital: R. J. Baird, R. J. gia. J Thorac Cardiovasc Surg 1993;105:833-44.
Cusimano, T. E. David, C. M. Feindel, I. H. Lipton, L. L. 18. Teoh KH, Christakis GT, Weisel RD, et al. Accelerated
Mickleborough, C. M. Peniston, H. E. Scully, and R. D. myocardial metabolic recovery with terminal warm blood
Weisel. We also acknowledge the assistance of Ms. Susan cardioplegia (hot shot). J Thorac Cardiovasc Surg 1986;91:
Dougherty in the preparation of this manuscript. 888-95.
19. Lytle BW, Loop FD, Taylor PC, et al. The effect of coronary
REFERENCES reoperation on the survival of patients with stenoses in
1. Christakis GT, Birnbaum PL, Weisel RD, et al. The changing saphenous vein bypass grafts to coronary arteries. J Thorac
pattern of coronary bypass surgery. Circulation 1989;80(Suppl): Cardiovasc Surg 1993;105:605-14.
I151-61. 20. Lazar HL, Buckberg GD, Manganaro AJ, Becker H. Myo-
2. The Warm Heart Investigators. Normothermic versus hypo- cardial energy replenishment and reversal of ischemic dam-
thermic cardiac surgery: a randomized trial in 1732 coronary age by substrate enhancement of secondary blood cardio-
bypass patients. Lancet 1994;343:559-63. plegia with amino acids during reperfusion. J Thorac
3. Menasch6 P, Jamieson WRE, Flameng W, Davies MK. Cardiovasc Surg 1980;80:350-9.
Acadesine: a new drug that improves myocardial protection 21. Rousou JA, Engelman RM, Anisimowicz L, et al. Metabolic
in coronary artery bypass graft surgery: results of the first enhancement of myocardial preservation during cardioplegic
multicenter study. J Thorac Cardiovasc Surg 1995;110:1096- arrest. J Thorac Cardiovasc Surg 1986;91:270-6.
106. 22. Hayashida N, Ikonomidis JS, Weisel RD, et al. The optimal
4. Weisel RD, Hoy FBY, Baird RJ, et al. Improved myocardial cardioplegic temperature. Ann Thorac Surg 1994;58:961-71.
protection during a prolonged crossclamp period. Ann Tho- 22a. Christakis GT, Weisel RD, Buth KJ, Fremes SE, Rao V,
rac Surg 1983;36:664-74. Panagiotopoulos KP, et al. Is body size the cause for poor
5. Weisel RD, Berger RL, Hechtman HB. Measurement of outcomes of coronary artery bypass operations in women? J
cardiac output by thermodilution. N Engl J Med 1975;292: Thorac Cardiovasc Surg 1995;110:1344-58.
682-4. 23. Hayashida N, Ikonomidis JS, Weisel RD, et al. Antegrade
6. Weisel RD, Burns RJ, Baird RJ, et al. Optimal postoperative distribution of warm cardioplegic solution. J Thorac Cardio-
volume loading. J Thorac Cardiovasc Surg 1983;85:552-63. vasc Surg 1995;110:800-12.
7. Lazar HL, Buckberg GD, Foglia RP, Manganaro AJ, Ma- 24. Goldman BS, Scully HE, Tong CP, et al. Coronary artery
loney F. Detrimental effects of premature use of inotropic bypass graft surgery in the elderly. Geriatric Cardiovasc Med
drugs to discontinue cardiopulmonary bypass. J Thorac Car- 1988;1:201-7.
diovasc Surg 1981;82:18-25. 25. Misare BD, Krukenkamp IB, Levitsky S. Age-dependent
8. Mickleborough LL, Rebeyka I, Wilson GJ, Gray G, Desro- sensitivity to unprotected cardiac ischemia: the senescent
siers A. Comparison of left ventricular assist and intra-aortic myocardium. J Thorac Cardiovasc Surg 1992;103:60-5.
14. The Journal of Thoracic and
Cardiovascular Surgery Rao et al. 51
Volume 112, Number 1
26. Barzilai B, Marshall WG, Safiftz JE, Kouchoukos N. Avoid- 72 hours of an event), or emergency (within 12 hours of an
ance of embolic complication by ultrasonic characterization event). An event is either cardiac catheterization or an
of the ascending aorta. Circulation 1989;80(Suppl):I275-9. ischemic event after cardiac catheterization. Timing of
27. Greenland S. Modelling and variable selection in epidemio- operation is dependent on the symptomatic status of the
logic analysis. Am J Public Health 1989;79:40-9. patient.
28. Kirklin JW, Blackstone EH. Presenting multivariable analy- Recent MI. MI within 30 days before operation (Q wave
ses. J Thorac Cardiovasc Surg 1994;107:1544-55. or non-Q wave with an elevation of CK-MB level).
29. Naftel DC. Do different investigators sometimes produce Hypertension. Preoperative systemic hypertension ne-
different multivariable equations from the same data? J cessitating medical treatment.
Thorac Cardiovasc Surg 1994;107:1528-43. Smoking. A history of smoking or current smoking.
Left ventricular grade. Ejection fraction greater than
Appendixes 60% (grade 1), ejection fraction between 40% and 60%
Appendix A: Definitions of perioperative variables (grade 2), ejection fraction between 20% and 40% (grade
3), ejection fraction less than 20% (grade 4), as assessed
Elderly. Patients older than age 70 years.
by a single-plane contrast ventriculogram or by echocar-
Diabetes. A preoperative diagnosis of diabetes mellitus
diography or nuclear ventriculography if a contrast ven-
treated with insulin, oral hypoglycemic agents, or diet.
Triple-vessel disease. Critical lesions (greater than 50% triculogram was not done.
luminal narrowing) affecting the territories supplied by Appendix B. To calculate the predicted probability of
the right, LAD, and left circumflex coronary arteries. low cardiac output syndrome or operative mortality for a
Left main coronary artery disease. Greater than 50% given patient, start with the constant and then add the
stenosis of the left main coronary artery. regression coefficients that describe the patient's charac-
Peripheral vascular disease. Known carotid, aortoiliac, or teristics for a coefficient total (x). Then use the formulap
femoropopliteal disease or cases in which the patient had = e*/(1 + e~). For example, from Table IV the predicted
a previous carotid endartectomy or peripheral vascular probability of low cardiac output syndrome for a 72-year-
operation. old male patient with a left ventricular ejection fraction of
Transient ischemic attacks. A preoperative history of 35% and left main coronary artery disease but no diabe-
transient ischemic attacks, reversible ischemic neurologic tes, no previous coronary artery bypass grafting, and no
deficits, or stroke. recent MI undergoing urgent operation would be x =
Normothermia. Systemic temperature higher than 35 ° C -3.866 + 0.3691 + 0.8614 + 0.350 + 0 + 0 + 0 + 0 +
during cardiopulmonary bypass. 0.2451 - 2.0404; p = e2°4°4/(1 + e 2'°4°4) = 0.115 (or
Timing. Elective, same hospitalization, urgent (within 11.5%).
![The Journal of Thoracic and
Cardiovascular Surgery Rao et al. 39
Volume 112, Number 1
may represent a failure of myocardial protection or Table I. Patient population
difficulty with bypass grafting and is often associated N %
with a higher mortality and prolonged hospital stay. N 4558
A recent randomized clinical trial done at the Sex
University of Toronto by The W a r m H e a r t Investi- Male 3673 81
gators showed no difference in operative mortality Female 885 19
or perioperative myocardial infarction (MI) be- Age (yr) 61.5 _+ 9.6
Timing
tween patients who received w a r m blood or cold Elective 2620 57
blood cardioplegic solution. 2 However, the occurrence Semielective 1130 25
of low cardiac output syndrome was significantly lower Urgent 657 14
in the warm group. Similarly, a large clinical trial that Emergency 151 4
evaluated acadesine as a myocardial protective agent Recent MI 692 15
LVEF
failed to show any difference in the prevalence of >60% 1530 34
perioperative infarction except in a subgroup of pa- 40%-60% 1926 42
tients with high-risk conditions. 3 Because of the low 20%-40% 956 21
rates of operative mortality and perioperative MI, <20% 146 3
studies aimed at developing improved strategies of NYHA class
I 76 2
myocardial protection are unlikely to show a benefit II 654 14
unless they include a large number of patients or limit III 1861 41
their focus to subgroups of patients at high risk. Both IV 1967 43
operative mortality and a new perioperative MI occur CAD
infrequently and these outcome measures are difficult ] vessel 308 7
2 vessel 1083 24
to use in a trial of alternate myocardial protective 3 vessel 3159 69
strategies even in patients at high risk. Left main CAD 808 18
Low cardiac output syndrome is a more frequent Repeat operation 335 7
event and may be modified by improved methods of Diabetes 1086 24
myocardial protection in high-risk subgroups. How- Hypertension 2337 51
PVD 591 13
ever, the syndrome must be carefully defined and its
pathophysiologic components may be better under- LVEF, Left ventricular ejection fraction; NYHA, New York Heart Asso-
ciation; CAD, coronary artery disease; PVD, Peripheral vascular disease.
stood by assessment of the predisposing risk factors.
To make comparisons between studies, the defini-
tion of low cardiac output syndrome must become artery stenoses, the mean arterial pressure was main-
standardized. This study presents the independent tained higher than 70 mm Hg in an attempt to improve
cerebral perfusion. Blood cardioplegic solutions were
preoperative risk factors for the development of low
delivered after oxygenated blood from the bypass circuit
cardiac output syndrome in 4558 consecutive pa- was mixed with a crystalloid solution in a 4:1 ratio. Blood
tients undergoing isolated coronary artery bypass cardioplegic solution was administered either in an ante-
grafting at The Toronto Hospital. grade fashion via the aortic root or in a retrograde fashion
via the coronary sinus.
In all patients, the heart was arrested with an aortic root
Methods
infusion of high potassium (27 mEq/L) blood cardioplegic
Patient population. Preoperative, perioperative, and solution. Cardioplegia was maintained with a low potas-
postoperative data were collected prospectively on all sium formulation (15 mEq/L). Proximal and distal anas-
patients undergoing isolated coronary artery bypass graft- tomoses were constructed during a single prolonged cross-
ing between July 1, 1990, and December 31, 1993, at The clamp period. 4 A left internal thoracic artery graft was
Toronto Hospital. The preoperative characteristics of the used in 3789 (83%) patients. Cardiac output was mea-
4558 consecutive patients are shown in Table I. sured with a thermodilution catheter placed percutane-
Operative technique. After median sternotomy and ously via the internal jugular vein into the pulmonary
heparinization, cardiopulmonary bypass was established artery. 5
with a single two-stage right atrial cannula and an ascend- Patients in whom weaning from cardiopulmonary by-
ing aortic cannula. During bypass, the hematocrit was pass was ditficult or in whom inadequate cardiac perfor-
maintained between 20% and 25%, pump flow rates mance developed in the intensive care unit had an in-
between 2.0 and 2.5 L/min per square meter, and mean traaortic balloon pump (Datascope Corporation, Paramus,
arterial pressures between 50 and 60 mm Hg by use of N.J.) inserted percutaneously via the common femoral ar-
sodium nitroprusside or phenylephrine hydrochloride as tery. Patients with less severe hemodynamic compromise
required. In elderly patients or in patients with carotid received inotropic medication.](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)