Clinical Characteristics and In-Hospital Outcomes of Patients With Cardiogenic Shock Undergoing Coronary Artery Bypass Surgery : Insights From the Society of Thoracic Surgeons National Cardiac Database Rajendra H. Mehta, Joshua D. Grab, Sean M. O'Brien, Donald D. Glower, Constance K. Haan, James S. Gammie and Eric D. Peterson Circulation. 2008;117:876-885; originally published online February 4, 2008; doi: 10.1161/CIRCULATIONAHA.107.728147 Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2008 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/117/7/876
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
Cardiovascular Surgery Clinical Characteristics and In-Hospital Outcomes of Patients With Cardiogenic Shock Undergoing Coronary Artery Bypass Surgery Insights From the Society of Thoracic Surgeons National Cardiac Database Rajendra H. Mehta, MD, MS; Joshua D. Grab, MS; Sean M. O’Brien, PhD; Donald D. Glower, MD; Constance K. Haan, MD, MS; James S. Gammie, MD; Eric D. Peterson, MD, MPH; on Behalf of the Society of Thoracic Surgeons National Cardiac Database Investigators Background—There exist few studies that characterize contemporary clinical features and outcomes or risk factors for operative mortality in cardiogenic shock (CS) patients undergoing coronary artery bypass grafting (CABG). Methods and Results—We evaluated data of 708 593 patients with and without CS undergoing CABG enrolled in the Society of Thoracic Surgeons National Cardiac Database (2002–2005). Clinical, angiographic, and operative features and in-hospital outcomes were evaluated in patients with and without CS. Logistic regression was used to identify predictors of operative mortality and to estimate weights for an additive risk score. Patients with preoperative CS constituted 14 956 (2.1%) of patients undergoing CABG yet accounted for 14% of all CABG deaths. Operative mortality in CS patients was high and surgery specific, rising from 20% for isolated CABG to 33% for CABG plus valve surgery and 58% for CABG plus ventricular septal repair. Although mortality for CABG surgery overall declined significantly over time (P for trend ⬍0.0001), mortality for CS patients undergoing CABG did not change significantly during the 4-year study period (P⫽0.07). Factors associated with higher death risk for CS patients undergoing CABG were identified by multivariable analysis and summarized into a simple bedside risk score (c statistic⫽0.74) that accurately stratified those with low (⬍10%) to very high (⬎60%) mortality risk. Conclusions—Patients with CS represent a minority of those undergoing CABG yet have persistently high operative risks, accounting for 14% of deaths in CABG patients. Estimation of patient-specific risk of mortality is feasible with the simplified additive risk tool developed in our study with the use of routinely available preprocedural data. (Circulation. 2008;117:876-885.) Key Words: coronary artery bypass grafting 䡲 mortality 䡲 risk factors 䡲 shock, cardiogenic 䡲 surgery
C
ardiogenic shock remains an important complication of acute myocardial infarction, with in-hospital mortality approaching 70% to 80% among those managed medically.1– 4 The Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock (SHOCK) Trial showed significant improvement in survival from immediate coronary revascularization in patients with cardiogenic shock, with similar beneficial effects observed with both percutaneous and surgical approaches.5 These results are reflected in the American College of Cardiology/American Heart Association guidelines that recommend immediate revascularization with either strategy as a class I indication in patients with cardiogenic shock.6
Clinical Perspective p 885 Many patients with cardiogenic shock are referred for consideration of coronary artery bypass grafting (CABG) as a
result of unsuitable coronary anatomy for percutaneous coronary intervention (PCI) and/or because of mechanical complications such as ventricular septal or papillary muscle rupture. Despite this, limited information exists on the clinical and angiographic features and outcomes in large number of patients with cardiogenic shock undergoing cardiac surgery in the community. To date, most studies of cardiac surgery in the setting of cardiogenic shock are based on limited case series from major academic or high-volume centers and may not be representative of those seen in wider community practice.7–9 The Society of Thoracic Surgeons (STS) National Cardiac Database (NCD) provides an ideal opportunity to study a national sample of patients undergoing cardiac surgery among a diverse cross section of US hospitals (in this case,
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz. Received July 21, 2007; accepted December 7, 2007. From the Duke Clinical Research Institute (R.H.M., J.D.G., S.M.O., E.D.P.) and Duke University Medical Center (R.H.M., D.D.G., E.D.P.), Durham, NC; University of Florida (C.K.H.), Jacksonville; and University of Maryland (J.S.G.), Baltimore. Correspondence to Rajendra H. Mehta, MD, Box 17969, Duke Clinical Research Institute, Durham, NC 27715. E-mail
[email protected] © 2008 American Heart Association, Inc. Circulation is available at http://circ.ahajournals.org
DOI: 10.1161/CIRCULATIONAHA.107.728147
876 Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
Mehta et al those with cardiogenic shock).10,11 Using these data, we sought to (1) characterize the clinical and angiographic features and in-hospital outcomes of patients undergoing CABG with cardiogenic shock; (2) track trends in outcomes over time; (3) identify preoperative clinical variables associated with operative mortality; and (4) create a bedside tool to estimate the operative mortality in patients with cardiogenic shock undergoing cardiac surgery.
Methods Study Sample and Data Collection The Society of Thoracic Surgeons National Cardiac Database The formation, rationale, and methodology of the STS NCD have been published previously.10,11 Briefly, this registry was initiated in 1986 with a goal to provide participants with their risk-adjusted outcomes compared with the national experience. This information is used by member institutions as part of their continuous quality improvement efforts to help design strategies to improve their outcomes. This data set contains detailed clinical information on ⬎2.5 million patients undergoing cardiac surgery from ⬎600 academic, private, military, and Veterans Affairs hospitals from 50 US states and 5 Canadian provinces.
Patient Population We analyzed data on patients with and without cardiogenic shock undergoing CABG either alone or in conjunction with aortic or mitral valve surgery or ventricular septal rupture repair during 2002–2005 at participating sites. Patients undergoing isolated valve, congenital heart, or aortic surgery were excluded from the analysis.
Data Definitions The STS NCD collects information on preoperative cardiogenic shock, defined as whether the patient was, at the time of procedure, in a clinical state of hypoperfusion according to either (1) systolic blood pressure ⬍80 mm Hg and/or cardiac index ⬍1.8 L/min per square meter despite maximal treatment or (2) intravenous inotropes and/or intra-aortic balloon pump necessary to maintain systolic blood pressure ⬎80 mm Hg and/or cardiac index ⬎1.8 L/min per square meter. Operative mortality was defined as (1) all deaths occurring during the hospital period in which the operation was performed and (2) those deaths occurring after hospital discharge but within 30 days of the procedure. Major morbidity was defined by the presence of any 1 of the following: need for reoperation (for bleeding, cardiac tamponade, graft occlusion, valvular dysfunction, or other cardiac or noncardiac problem), sternal wound infection, permanent stroke, renal failure, and/or prolonged ventilation. Information on all other adverse events was ascertained as in-hospital events. Remaining definitions of risk factors were as provided on the STS Web site (http://www.sts.org).
Statistical Analysis Summary statistics are presented as frequency and percentage or as median and interquartile range. We used the 2 test for comparing categorical variables and the Wilcoxon rank sum test for continuous variables. Few records had missing data. Records with missing values of age, gender, and race (⬍1%) were excluded from all data presented here; records with missing elective operative status (⬍1%) were excluded from multivariable analysis of the cardiogenic shock subpopulation. Missing values of body mass index (⬍1%) were imputed to gender-specific median values. Missing values of information of valvular stenosis or insufficiency (⬍3%) and the remaining risk factors (⬍1%) were defaulted to their most common value (not present).
Predictive Modeling Logistic regression modeling was used to estimate risk of operative mortality among patients with cardiogenic shock. The model devel-
CABG in Patients With Cardiogenic Shock
877
opment was restricted to patients with cardiogenic shock undergoing salvage, emergent, or urgent cardiac surgery. The full model included a comprehensive list of 33 predictor variables including interaction terms between serum creatinine and body mass index and gender. Continuous variables were either assumed to have a linear relationship (age [in 5-year increments], body mass index) or categorized (serum creatinine, left ventricular ejection fraction). The full model was used as a gauge by which to measure the accuracy of a subsequent simpler model approximation and bedside risk scoring system.
Simplified Model Approximation To reduce the number of variables, a parsimonious subset of predictor variables was identified that together explained ⬎95% of the variation in the predicted log-odds of death as estimated by the aforementioned full model. Our simplified model is the one that explained the most variation of all models with only 11 variables. Regression coefficients were estimated by treating the predicted log-odds from the full model of the outcome variable in a leastsquares regression model.12 The c index was used to determine the ability of the simplified model to discriminate between patients with and without in-hospital death.13 Both the full model and the process for determining the simplified model were internally validated with the use of the bootstrap resampling method. Calibration of the model was assessed by the Hosmer-Lemeshow test statistic.
Bedside Tool Regression coefficients from the simplified model approximation were then converted to whole integers by multiplying the coefficients by a factor of 10 and rounding to the nearest integer to create a bedside risk prediction tool.14 For each possible risk score, the risk of operative death was estimated by averaging the predicted values from the original full model among all patients having the same risk score. The heterogeneity of mortality risk among patients having the same risk score was assessed by plotting the distribution of predicted risk within risk score subgroups. Finally, we assessed whether our risk score for mortality could also be used to predict the combined end point of mortality and/or major morbidity. Discrimination was quantified by the c index, which was adjusted for optimism by the use of bootstrapping. The risk of mortality and/or major morbidity was estimated within each risk score subgroup in 2 ways: (1) by calculating the average observed rate of mortality and major morbidity in each risk score subgroup and (2) by averaging the predicted values from a logistic regression model within the same risk score subgroup. The variables used in this logistic model were identical to those in the full mortality model. Because information on left ventricular ejection fraction was missing in 11% of patients, we explored 2 strategies to derive the risk scores: record exclusion versus imputing ejection fraction to median values within subgroups defined by gender, congestive heart failure, and operative status. These 2 sets of risk scores (with and without the imputation of ejection fraction) were highly correlated (correlation coefficient⫽0.98). Thus, we elected to exclude patients with missing ejection fraction from our multivariable modeling. All analyses were performed with the use of S-Plus 6 (Insightful Corporation, Seattle, Wash) and SAS 8.1 (SAS Institute, Cary, NC) software. The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Results Baseline Clinical Characteristics of Patients With Cardiogenic Shock Undergoing Cardiac Surgery Preoperative shock was present in 14 956 (2.1%) of the total of 708 593 patients undergoing CABG enrolled in the STS NCD during 2002–2005. The proportion of patients with CABG having cardiogenic shock remained constant over this time period, constituting 2.1%, 2.1%, 2.1%, and 2.2% of
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
878
Circulation Table 1.
February 19, 2008
Patient Characteristics
Characteristics
Overall
Sample size, n (%)
708 593
No Preoperative Cardiogenic Shock 693 637 (97.9)
Preoperative Cardiogenic Shock
P
14 956 (2.1)
Demographics Age, y, median (IQR)
67 (58–75)
67 (58–74)
67 (57–75)
0.0232
Age ⱖ75 y, %
25.0
25.0
25.9
0.0114
Female, %
28.9
28.8
33.3
⬍0.0001
White, %
86.9
86.9
84.7
⬍0.0001
BMI, kg/m2, median (IQR)
28.3 (25.2–32.1)
28.3 (25.2–32.1)
27.5 (24.5–31.2)
⬍0.0001
Surgery type 86.8
86.9
81.1
⬍0.0001
8.2
8.3
5.2
⬍0.0001
CABG⫹mitral valve, %
4.9
4.8
11.7
⬍0.0001
CABG⫹ventricular septal rupture repair, %
0.1
0.1
2.0
⬍0.0001
Hypertension, %
77.7
77.8
70.8
⬍0.0001
Diabetes mellitus, %
35.5
35.5
34.7
0.0793
Current smoker, %
22.2
22.0
29.7
⬍0.0001 ⬍0.0001
Isolated CABG, % CABG⫹aortic valve, %
Medical history
Hypercholesterolemia, %
71.8
72.0
60.0
Chronic lung disease, %
20.6
20.5
25.1
⬍0.0001
Preoperative renal failure, %
5.9
5.7
12.4
⬍0.0001
Preoperative dialysis, %
1.7
1.6
2.8
⬍0.0001
Stroke, %
7.7
7.7
8.4
0.0007
Immunosuppressive treatment, %
2.1
2.1
3.1
⬍0.0001
Peripheral vascular disease, %
16.3
16.3
17.9
⬍0.0001
Cerebrovascular disease, %
15.1
15.1
13.5
⬍0.0001
Prior CABG, % Prior PCI, % Prior valve surgery, %
5.8
5.8
4.8
⬍0.0001
11.8
11.7
16.6
⬍0.0001
0.7
0.7
0.9
0.0163
Preoperative cardiac status and angiographic findings 43.5
42.6
86.6
⬍0.0001
Preoperative MI (ⱕ24 h), %
3.7
3.0
38.8
⬍0.0001
Preoperative IABP, %
7.1
5.8
70.3
⬍0.0001
Preoperative MI within 7 days, %
Preoperative serum creatinine, g/dL, median (IQR) Glomerular filtration rate, mL/min per 1.73 m Mitral regurgitation (moderate to severe), % Severe aortic stenosis, % Left main disease ⱖ50%, % Coronary artery disease ⱖ2 vessels, % LVEF, %, median (IQR)
2
1.0 (0.9–1.2)
1.0 (0.9–1.2)
1.1 (0.9–1.4)
71.6 (57.4–87.0)
71.9 (57.5–87.0)
64.6 (47.9–81.1)
7.0
6.8
17.7
⬍0.0001 ⬍0.0001 ⬍0.0001
8.2
8.3
5.1
⬍0.0001
25.8
25.5
37.6
⬍0.0001
92.1 50 (40–60)
92.1
91.9
50 (40–60)
35 (25–50)
0.0906 ⬍0.0001
Operative details Cardiopulmonary bypass, %
82.1
Perfusion time, min, median (IQR) Cross-clamp time, min, median (IQR) Insertion of ventricular assist device, %
81.9
88.0
⬍0.0001
101 (77–133)
100 (77–132)
116 (85–158)
⬍0.0001
70 (51–96)
70 (51–96)
73 (51–104)
⬍0.0001
2.53
⬍0.0001
0.24
0.20
Internal mammary artery graft, %
84.4
85.0
58.1
⬍0.0001
No. of SVG grafts ⱖ2, %
91.1
91.0
95.5
⬍0.0001
IQR indicates interquartile range; BMI, body mass index; MI, myocardial infarction; IABP, intra-aortic balloon pump; LVEF, left ventricular ejection fraction; and SVG, saphenous vein graft.
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
Mehta et al Table 2.
CABG in Patients With Cardiogenic Shock
879
Unadjusted In-Hospital Outcomes
Complications
Overall
No Preoperative Cardiogenic Shock
Sample size, n (%)
708 593
693 637 (97.9)
Operative mortality, %
3.4
Time to death, d, median (IQR) Proportion of death occurring same day of surgery, %
8 (2–19) 11.5
Preoperative Cardiogenic Shock
P
14 956 (2.1)
3.0
22.1
⬍0.0001
8 (3–19)
4 (1–14)
⬍0.0001
10.6
17.3
⬍0.0001
8.6
7.6
14.9
⬍0.0001
17.4
16.5
60.9
⬍0.0001
Permanent stroke, %
1.7
1.7
4.3
⬍0.0001
Coma, %
0.5
0.5
3.4
⬍0.0001
Proportion of death occurring next day of surgery, % Major morbidity, %
1.1
1.1
1.5
⬍0.0001
10.3
9.5
47.5
⬍0.0001
Pneumonia, %
3.6
3.4
13.3
⬍0.0001
Cardiac tamponade, %
0.4
0.4
1.1
⬍0.0001
Perioperative myocardial infarction, % Prolonged ventilation, %
Reoperation for bleeding/tamponade, %
3.0
2.9
6.2
⬍0.0001
Deep sternal wound infection, %
0.5
0.5
1.2
⬍0.0001
Length of ICU stay, h, median (IQR) Postoperative length of stay, d, median (IQR)
38 (23–70)
36 (23–69)
96 (48–216)
⬍0.0001
6 (4–8)
6 (4–8)
8 (5–15)
⬍0.0001
IQR indicates interquartile range; ICU, intensive care unit.
overall CABG patients in 2002, 2003, 2004, and 2005, respectively. Table 1 demonstrates the baseline, angiographic, and operative characteristics of the study patients. Despite the similar age of those with and without cardiogenic shock, those with cardiogenic shock generally had more comorbid conditions. In particular, a much higher percentage of patients with cardiogenic shock suffered an acute myocardial infarction within 24 hours before CABG versus those without shock (45% versus 7%; P⬍0.0001). Additionally, significant left main stenosis was 1.5 times more common and median left ventricular ejection fraction was 15% lower in patients with cardiogenic shock than in those without it. Preoperative intra-aortic balloon pump was used in ⬎70% of patients with cardiogenic shock as opposed to in 5.8% patients without cardiogenic shock. Operative procedure in patients with cardiogenic shock was more likely to include mitral valve surgery (usually for mitral regurgitation) or ventricular septal rupture repair and less likely to be an isolated CABG or CABG with aortic valve procedure. Use of an internal thoracic artery graft was only 58.1% in patients with preoperative cardiogenic shock compared with 85.0% in those without it. Median pump and cross-clamp times were modestly longer for shock patients. Although the use of a ventricular assist device was 13-fold higher in the cardiogenic shock patients than in those without it, ⬍3% of cardiogenic shock patients received these devices.
In-Hospital Outcomes Patients with cardiogenic shock undergoing CABG had worse in-hospital cardiac outcomes and longer lengths of intensive care unit and hospital stays than those without it (Table 2). Patients with cardiogenic shock had a shorter time to death, with a third of these deaths occurring by the next day of CABG. Major morbidity occurred in 61%. Operative
mortality was ⬎7-fold higher in patients with cardiogenic shock than in those without it undergoing CABG and accounted for 1 of ⬇7 deaths in CABG patients (14%). Mortality varied accordingly to the type of surgical procedure performed among patients with cardiogenic shock and was lowest in patients undergoing isolated CABG and highest in those undergoing CABG plus ventricular septal rupture repair (Figure 1). Operative mortality was higher for patients with cardiogenic shock operated within 24 hours of their myocardial infarction (26%) compared with those operated from 1 to 7 days (20%), 8 to 21 days (20%), and ⬎21 days (18%) after their myocardial infarction. Whereas the operative mortality in overall patients undergoing CABG declined over time (3.0%, 3.1%, 2.9%, and 2.8% for years 2002, 2003, 2004, and 2005, respectively; P for trend ⬍0.0001), this did not change significantly in patients with cardiogenic shock undergoing urgent, emergent, or salvage procedures during the 4-year study period (22.9%, 22.5%, 21.9%, 21.2% for years 2002, 2003, 2004, and 2005, respectively; P for trend⫽0.0675). Increasing age was also associated with higher operative mortality. In patients ⱖ75 years of age, mortality increased 1.7-fold compared with those ⬍75 years of age (31% versus 18%).
Multivariable Analysis Multivariable analysis was performed on 12 339 of the 14 956 cardiogenic shock patients who met the inclusion criteria and did not have the aforementioned exclusions. The initial full model identified 18 factors independently associated with increased risk of death (model not shown) (c statistic⫽0.75). Figure 2 displays a comparison between observed and predicted rates of mortality based on this full model. Although the Hosmer-Lemeshow probability value for lack of fit was 0.04 (related to a very large sample size), from a clinical perspective good concordance existed be-
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
880
Circulation
February 19, 2008
58 %
Cardiogenic Shock
STS Population (excl. Cardiogenic Shock)
33 % 31 %
22 %
22 %
Figure 1. In-hospital death in patients without cardiogenic shock and those with cardiogenic shock undergoing CABG, CABG and aortic valve surgery (AV), CABG and mitral valve surgery (MV), and CABG and ventricular septal rupture (VSR) repair.
19 %
8.6 % 6.4 % 3.0 %
2.3 %
All
Isolated CABG
AV+CABG
MV+CABG
n= 14956 n= 693637
n= 12132 n= 602835
n= 772 n= 57327
n= 1755 n= 33146
tween predicted and observed outcomes across the risk continuum (Figure 2). Furthermore, bootstrap validation of the full model with the use of 200 replicates showed modest internal validation (c statistic⫽0.746; 95% confidence interval [CI], 0.745 to 0.747). The simplified model (Table 3) retained much of the predictive power of the full model (c statistic⫽0.74; for bootstrap of variable selection process, c statistic⫽0.741; 95% CI, 0.740 to 0.741) and showed reasonable fit when we compared deciles of observed versus predicted rates of death (Hosmer-Lemeshow P⫽0.09 for lack of fit). The simplified model coefficients were then converted to an additive risk score (Figure 3). The average absolute difference between estimates calculated from the full model compared with estimates based on the risk scores was 2.5%. This risk score demonstrated excellent ability to identify those with low, moderate, and high risk for operative mortality (Figure 4). Bootstrap validation of the risk score for death predicting mortality or morbidity showed a modest concordance between predicted outcome with the use of the mortality risk score and observed mortality plus morbidity (c statistic⫽0.726; 95% CI, 0.719 to 0.733; Hosmer-Lemeshow P⫽0.07 for lack of fit) (Figure 5). Finally, excellent correlation was present between the risk score based on the simpli-
VSR+CABG n= 297
n= 329
fied model and that based on the full model (Pearson correlation coefficient⫽0.9394 or R2⫽0.88).
Discussion CABG represents an important therapeutic option for revascularization in patients with cardiogenic shock.6,15 Our study suggests that, although patients with cardiogenic shock constitute a minority of patients undergoing CABG, they account for 16% of all CABG deaths. Even more striking was the high incidence of major morbidity that prolonged hospitalization and resource use. Despite improvements in surgical and anesthesia techniques and newer technologies (ventricular assist devices), the risks with CABG in cardiogenic shock have remained stable over time. Our study provides a bedside tool that makes prediction of the risks of mortality and the combined end point of mortality plus major morbidity possible on the basis of simple preoperative factors. CABG offers some unique potential benefits in the treatment of patients with cardiogenic shock. It provides immediate protection of myocardium with cardioplegia, the ability to cool the heart to reduce the energy requirement, unloading of ventricle during cardiopulmonary bypass, and the ability to achieve more complete revascularization than percutaneous revascularization. However, nationally, very few patients with cardiogenic shock and 3-vessel disease are referred for
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
50
Mehta et al
CABG in Patients With Cardiogenic Shock
881
Observed
30
Figure 2. Observed vs predicted mortality based on original full model (Hosmer-Lemeshow P for lack of fit⫽0.04).
0
10
20
Mortality (%)
40
Predicted, Based on Full Model
1
2
3
4
5
6
7
8
9
10
Deciles of Predicted Risk
CABG, ranging from 3.2% to 8.8%.16 This reservation in referral for CABG may be a reflection of the logistical challenges in arranging timely emergency surgery, especially during off hours, when the alternative strategy of primary PCI can be undertaken expeditiously and in more hospitals than those offering CABG, as well as the fear of high operative mortality. Our study summarizes the national results of CABG performed in the setting of cardiogenic shock. Overall, we found that even when faced with an unarguably sick and critically ill population, operative mortality was not prohibitive in appropriately selected patients. Additionally, although age was strongly associated with operative mortality, a full 70% of patients even in the oldest age group (ⱖ75 years) survived this major surgery. Therefore, it is reasonable for clinicians to consider surgical options in appropriate functional elderly patients with cardiogenic shock. Although not designed to test various treatment options, our study suggests some potential opportunities to improve outcomes of patients with cardiogenic shock undergoing CABG. Intra-aortic balloon pump support was utilized in only two thirds of such patients. Although direct trials of intra-aortic balloon pump in this setting are not available, it appears that greater use of preoperative intra-aortic balloon pump support has the potential for improving operative outcomes. Shorter pump and cross-clamp times could also
decrease the impact of prolonged cardiopulmonary bypass on outcomes. The majority of procedures were done with the use of cardiopulmonary bypass, and the role of off-pump surgery merits evaluation in this cohort. In addition, very few patients in the cardiogenic shock group had a ventricular assist device placed during their hospital stay. This seems low, given the fact that many of these may have been candidates for this therapy as a bridge to cardiac transplant. Recent STS NCD data suggest that up to 60% of patients with persistent shock after CABG can be salvaged with the use of ventricular assist devices.17 Improving the availability of this technology at centers offering CABG may represent an opportunity to save more lives.18 Although emergent revascularization has been shown to improve outcomes in patients with cardiogenic shock,5 the optimal revascularization strategy in patients with cardiogenic shock needs to be determined. The 22% overall mortality (18% versus 31% in those aged ⬍75 and ⱖ75 years, respectively) observed for the cohort with cardiogenic shock undergoing CABG may appear to be low but is a stable estimate among a large number of such patients at a national level. For example, this rate is very similar to that observed for years 2002–2004 in the National Registry of Myocardial Infarction, which reported in-hospital mortality between 16.7% and 28.6% in a very small number of patients with
Downloaded from http://circ.ahajournals.org/ by guest on June 8, 2013
882
Circulation
February 19, 2008
Table 3. Simplified Reduced Model Showing Preoperative Factors Independently Associated With Operative Mortality Model 2
Odds Ratio
95% CI
P
CABG⫹aortic valve
31.2
1.79
1.46–2.19
⬍0.0001
CABG⫹mitral valve
71.7
1.75
1.54–1.99
⬍0.0001
117.2
4.55
3.46–5.99
⬍0.0001
182.3
1.17
1.15–1.20
⬍0.0001
Variable Surgery type (vs isolated CABG)
CABG⫹ventricular septal repair Age (increment of 5 y vs ⬍50 y) Preoperative creatinine (vs ⬍1.0 mg/dL) 1.0–⬍1.2 mg/dL
14
1.30
1.13–1.49
0.0002
1.2–⬍1.4 mg/dL
40.7
1.71
1.45–2.02
⬍0.0001
87.4
2.13
1.82–2.50
⬍0.0001
215.5
3.40
2.89–4.00
⬍0.0001
40–49%
0.1
1.03
0.89–1.20
0.7066
30–39%
10.7
1.26
1.10–1.45
0.0011
⬍30%
54.3
1.61
1.42–1.82
⬍0.0001
1.4–⬍1.9 mg/dL ⬎1.9 mg/dL Left ventricular ejection fraction (vs ⬎50%)
Operative status: salvage vs emergent or urgent
251.4
3.32
2.86–3.85
⬍0.0001
84.5
1.63
1.47–1.80
⬍0.0001
Female gender (vs male) Intra-aortic balloon pump use
46.4
1.57
1.38–1.79
⬍0.0001
Prior cardiac surgery
50.8
1.97
1.64–2.38
⬍0.0001
Preoperative resuscitation
53.0
1.53
1.36–1.71
⬍0.0001
Myocardial infarction within 1 week of CABG
16.2
1.27
1.13–1.43
⬍0.0001
Immunosuppressive therapy
17.6
1.66
1.31–2.11
⬍0.0001
cardiogenic shock undergoing CABG.16 Similarly, in the SHOCK trial registry of 884 patients with predominant left ventricular pump failure, in-hospital mortality occurred in 24% of patients treated with CABG.19
Finally, we also identified important clinical factors that were associated with increased risk for operative mortality (and mortality plus morbidity) in patients with cardiogenic shock undergoing CABG. The prognostic importance of Score
A ge Points