Optimal Timing From Myocardial Infarction to Coronary Artery Bypass Grafting on Hospital Mortality

Transcription

1 ADULT CARDIAC J. MAXWELL CHAMBERLAIN MEMORIAL PAPER Optimal Timing From Myocardial Infarction to Coronary Artery Bypass Grafting on Hospital Mortality Elizabeth L. Nichols, MS, Jock N. McCullough, MD, Cathy S. Ross, MS, Robert S. Kramer, MD, Benjamin M. Westbrook, MD, John D. Klemperer, MD, Bruce J. Leavitt, MD, Jeremiah R. Brown, PhD, MS, Elaine Olmstead, BA, Felix Hernandez, MD, Gerald L. Sardella, MD, Carmine Frumiento, MD, David Malenka, MD, and Anthony DiScipio, MD, for the Northern New England Cardiovascular Disease Study Group The Dartmouth Institute for Health Policy and Clinical Practice, Geisel School of Medicine at Dartmouth, Lebanon, New Hampshire; Departments of Surgery and Medicine, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire; Department of Surgery, Maine Medical Center, Portland, Maine; Department of Surgery, Catholic Medical Center, Manchester, New Hampshire; Department of Surgery, Eastern Maine Medical Center, Bangor, Maine; Department of Surgery, University of Vermont Medical Center, Burlington, Vermont; Department of Surgery, Concord Hospital, Concord, New Hampshire; and Department of Surgery, Central Maine Medical Center, Lewiston, Maine Background. Whether delaying coronary artery bypass grafting (CABG) after myocardial infarction (MI) is associated with better outcomes or is an unnecessary use of health care resources is unclear. This study investigated the relationship between MI-to-CABG timing on in-hospital death. Methods. From the Northern New England Cardiovascular Disease Study Group (NNE) Cardiac Surgery Registry we identified 3,060 isolated CABG patients with prior MI from 2008 to We compared in-hospital death by MI-to-CABG timing of less than 1 day, 1 to 2 days, 3 to 7 days, and 8 to 21 days. We adjusted for patient characteristics using logistic regression. Results. Among patients with prior MI, CABG was performed within 1 day for 99 (3.2%), 1 to 2 days for 369 (12.1%), 3 to 7 days for 1,966 (64.3%), and 8 to 21 days for 626 (20.5%) patients. NNE-predicted mortality was similar for patients operated on within 1 day (1.8%), 1 to 2 days (1.8%), and 3 to 7 days (1.9%), but was higher for 8 to 21 days (2.4%) of MI. Crude in-hospital mortality was higher for those with MI-to-CABG time of less than 1 day (5.1%) compared with 1 to 2 days (1.6%), 3 to 7 days (1.6%), and 8 to 21 days (2.7%, p [ 0.044). Adjusted in-hospital mortality remained high for less than 1 day (5.4%; 95% CI, 1.5% to 9.4%), and similar for 1 to 2 days (1.8%; 95% CI, 0.4% to 3.1%), 3 to 7 days (1.7%; 95% CI, 1.1% to 2.3%), and 8 to 21 days (2.3%; 95% CI, 1.2% to 3.3%) between MI and CABG. Conclusions. Patients operated on 1 to 2 days and 3 to 7 days after MI had a similar mortality rate, suggesting it may be possible to reduce the MI-to-CABG interval for some patients without sacrificing outcomes. Patients operated on within 1 day after MI had a higher mortality rate. (Ann Thorac Surg 2017;103:162 71) Ó 2017 by The Society of Thoracic Surgeons Coronary artery bypass grafting (CABG) is performed in approximately 20% of patients with myocardial infarction (MI) [1]. The timing of CABG after MI may affect outcomes and resource utilization. Older studies suggest that the mortality rate is lower if CABG is delayed 3 to 5 days after an MI as opposed to operating earlier [2, 3]. More recent studies report no differences in outcomes if CABG is performed early vs late after MI [4 6]. Accepted for publication May 27, Presented at the Fifty-second Annual Meeting of the Society of Thoracic Surgeons, Phoenix, AZ, Jan 23 27, Winner of the J. Maxwell Chamberlain Memorial Award for Adult Cardiac Surgery. Address correspondence to Ms Nichols, HB 7505, Dartmouth-Hitchcock Medical Center, One Medical Center Dr, Lebanon, NH 03756; elizabeth.l.nichols.gr@dartmouth.edu. Further confusing the interpretation of the data is the variety and span of time points, inclusion of shock, and mixed clinical cohorts used in analyses [7]. The optimal timing of CABG after MI is that time frame in which death is minimized without unnecessarily increasing length of stay and resource utilization. Non ST-segment elevation MI (NSTEMI) guidelines primarily address timing relative to medication use [8]. Clopidogrel and ticagrelor should be discontinued in elective CABG patients 5 days before the operation and prasugrel 7 days before the operation [8]. In urgent CABG, clopidogrel and ticagrelor should be discontinued for at least 24 hours [8]. For ST-segment elevation MI (STEMI) patients, guidelines recommend delaying CABG 3 to 7 days or until patients are stabilized but acknowledge CABG can be performed within 48 hours for Ó 2017 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier

2 Ann Thorac Surg CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL 2017;103: OPTIMAL TIMING FROM MI TO CABG patients with multivessel coronary artery disease with recurrent angina [9, 10]. This study aims to inform the ongoing debate on optimal CABG timing after MI. We used the Northern New England Cardiovascular Disease Study Group (NNECDSG) CABG registry to investigate the effect of CABG timing on in-hospital death in MI patients. Patients and Methods Setting and Data Sources Data were abstracted from the NNECDSG Open Heart Surgery Registry to identify isolated CABG patients ( The NNECDSG is a voluntary, regional consortium of 8 medical centers in Vermont, New Hampshire, and Maine that have banded together to study the process and outcomes of cardiovascular care in the region and to use primary data collection to drive quality improvement. The Open Heart Surgery Registry consists of prospectively collected data on consecutive open heart operations performed at member institutions. Data are collected for demographics, comorbidities, cardiac history, indication, priority, coronary anatomy and function, procedural process, and in-hospital outcomes. The data collection is validated every 3 years by comparison with administrative records to ensure information on every operation has been collected and that in-hospital deaths are correct. The NNECDSG data collection is designated as a Quality Improvement Registry, and therefore, unique patient consent for this study was not required. Patient Population We identified 10,765 patients who underwent isolated CABG from 2008 to 2014 in the NNECDSG Surgery Registry. Of these, 3,526 patients had sustained a MI within 21 days. We excluded an additional 466 patients with emergency or salvage CABG, shock, or CABG within 6 hours of MI because guidelines recommend performing CABG as soon as possible for emergency and shock patients, and those with a CABG within 6 hours of MI most likely had an emergency reason for early operation timing [10]. After exclusions, our analytic sample size included 3,060 patients. Measures We assigned patients to four groups according to the time from MI to CABG: more than 6 hours to less than 1 day, 1 to 2 days, 3 to 7 days, and 8 to 21 days. Timing from MI to CABG is a data field collected prospectively in our registry as the defined time cutoffs we used. MI was defined as biomarker evidence of myocardial necrosis with ischemic symptoms, the development of Q waves, or electrocardiogram changes indicative of ischemia or pathologic findings. STEMI was identified using The Society of Thoracic Surgeons definition or the NNE definition. This was a new ST-segment elevation or left bundle branch block and cardiac biomarker elevation in 184 patients and ST-segment elevation or new Q waves in 345. All other MIs were identified as NSTEMI. The primary outcome was in-hospital death. Secondary outcomes included return to the operating room for bleeding, blood transfusion, cerebrovascular accident, or new-onset postoperative atrial fibrillation. Statistical Analysis We compared patient characteristics by the time from MI to CABG using the c 2 test for categoric variables and analysis of variance for continuous variables. We compared outcomes using direct standardization and multivariable logistic regression to adjust for differences in patient case mix across categories of time from MI to CABG. We used variables previously associated with death in the NNECDSG prediction model, including age, ejection fraction of less than 0.40, 3-vessel disease, left main disease, white blood cell count of 12,000 /ml or higher, prior CABG, vascular disease, diabetes; preoperative renal failure, hemodialysis-dependent renal failure, or preoperative creatinine 2 mg/dl; chronic obstructive pulmonary disease, recent STEMI, or percutaneous coronary intervention within this admission [11]. For sensitivity analyses, we investigated outcomes across groups within the STEMI and NSTEMI subgroups. We also analyzed outcomes across the 1 to 2 days and 3 to 7 days between MI and CABG among a cohort of patients admitted Monday or Tuesday, when admission day of the week would have minimal effect on CABG timing. We adjusted the Monday and Tuesday cohort using age, sex, ejection fraction of less than 0.40, 3-vessel disease, vascular disease, diabetes, chronic obstructive pulmonary disease, and STEMI. We used Stata 13.0 software (StataCorp LP, College Station, TX) for all statistical analyses. Results 163 Of the 3,060 patients with prior MI, CABG was performed within 1 day for 99 patients (3.2%), in 1 to 2 days for 369 (12.1%), in 3 to 7 days for 1,966 (64.3%), and in 8 to 21 days for 626 (20.5%). Preoperative characteristics are reported in Table 1 and operative characteristics are reported in Table 2. Patients with an MI-to-CABG time of less than 1 day were more likely to have a STEMI, a percutaneous coronary intervention this admission, a lower ejection fraction, and severe left main disease, and to receive Plavix (Bristol-Myers Squibb, Princeton, NJ) within 24 hours preoperatively and be supported with an intraaortic balloon pump than the other groups. Patients with an MI-to-CABG time of 8 to 21 days were more likely to have a prior CABG, vascular disease, diabetes, chronic obstructive pulmonary disease, and congestive heart failure, be on dialysis or have a creatinine 2 mg/dl, and be elective at time of operation. The characteristics of the groups waiting 1 to 2 days and 3 to 7 days were very similar. Predicted mortality using the NNECDSG risk model was similar for patients who had CABG within 1 day (1.8%), 1 to 2 days (1.8%), and 3 to 7 days (1.9%) but was higher for those undergoing CABG post-mi at 8 to 21 days (2.4%). ADULT CARDIAC

3 ADULT CARDIAC 164 CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL Ann Thorac Surg OPTIMAL TIMING FROM MI TO CABG 2017;103: Table 1. Preoperative Patient Characteristics Variable a <1 Day (n ¼ 99) 1 2 Days (n ¼ 369) 3 7 Days (n ¼ 1,966) 8 21 Days (n ¼ 626) p Value b Age in years < Female sex Body surface area, m < Preoperative WBC >12,000/mL <0.001 Prior CABG Prior PCI PCI this admission <0.001 STEMI, this admission <0.001 Comorbid disease Vascular disease <0.001 Diabetes <0.001 COPD <0.001 CHF <0.001 Dialysis or creatinine >2 mg/dl <0.001 NYHA class IV Ejection fraction < Coronary artery disease Left main stenosis 50% 89% <0.001 >90% vessel disease vessel disease Proximal LAD Priority at operation Elective <0.001 Urgent Plavix c 24 hours preoperatively <0.001 Aspirin within 5 to 7 days Preoperative IABP <0.001 Predicted mortality, % 1.8 (2.3) 1.8 (2.4) 1.9 (2.6) 2.4 (2.7) a Categoric data are presented as the percentage and continuous data as the mean (SD). b p values were determined by c 2 test for categoric variables and by analysis of variance for continuous variables. c Bristol-Myers Squibb, Princeton, NJ. CABG ¼ coronary artery bypass grafting; CHF ¼ congestive heart failure; COPD ¼ chronic obstructive pulmonary disease; IABP ¼ intraaortic balloon pump; LAD ¼ left anterior descending; MI ¼ myocardial infarction; NYHA ¼ New York Heart Association; PCI ¼ percutaneous coronary intervention; STEMI ¼ ST-segment elevation myocardial infarction; WBC ¼ white blood cell. Mortality The overall in-hospital mortality rate was 1.96% (60 patients). Crude in-hospital mortality was higher for those with an MI-to-CABG time of within 1 day (5.1%) compared with 1 to 2 days (1.6%), 3 to 7 days (1.6%), and 8 to 21 days (2.7%, p ¼ 0.044; Table 2). After direct standardization, in-hospital mortality remained similar and was 5.4% at less than 1 day, 1.8% at 1 to 2 days, 1.7% at 3 to 7 days, and 2.3% at 8 to 21 days (Fig 1 and Table 3). Logistic regression showed the adjusted odds ratio (OR) for in-hospital death compared with those operated on in 3 to 7 days after MI was significantly

4 Ann Thorac Surg CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL 2017;103: OPTIMAL TIMING FROM MI TO CABG Table 2. Intraoperative Characteristics Variable a <1 Day (n ¼ 99) 1 2 Days (n ¼ 369) 3 7 Days (n ¼ 1,966) Days (n ¼ 626) p Value ADULT CARDIAC Surgeon experience b, y 16.3 (5.5) 17.7 (6.8) 16.9 (7.5) 16.6 (7.9) Off pump Cross-clamp time, min 59.3 (27.1) 68.8 (38.1) 62.8 (33.7) 65.2 (31.7) Pump time, min 95.0 (33.8) 99.7 (32.2) 99.5 (34.8) (37.3) IMA use Left Right Bilateral a Continuous data are presented the mean (SD) and categoric data as the percentage. b Surgeon experience reflects the mean years of practice surgeons had in the Northern New England Cardiovascular Disease Study Group as of IMA ¼ internal mammary artery. higher in less than 1 day (OR, 3.7; 95% confidence interval [CI], 1.3 to 10.5) patients. The adjusted OR for in-hospital death did not differ significantly in those operated on 1 to 2 days (OR, 1.1; 95% CI, 0.4 to 2.6) or 8 to 21 days (OR, 1.3; 95% CI, 0.7 to 2.5) relative to 3 to 7 days after MI (Fig 2 and Table 3). Patients operated on within 1 day had significantly higher mortality, and patients operated on in 8 to 21 days had more comorbidities and increased, although not statistically significant, mortality. Meanwhile, crude and adjusted mortality were similar for those operated on in 1 to 2 or in 3 to 7 days. STEMI and NSTEMI Crude or adjusted in-hospital mortality did not differ significantly across timing groups among 529 STEMI patients (Table 3). Similarly, among the 2,531 NSTEMI patients, there was no significant difference in crude or adjusted in-hospital mortality by timing (Table 3). Day of the Week Subanalysis The day of the week on which a patient was admitted had some influence on the time from MI to CABG (p < 0.001, Fig 3). Patients admitted early in the week (Monday through Wednesday) were more likely to have an operation within 1 to 2 days of MI than patients admitted on Thursday through Sunday. We presumed that some of this was a consequence of scheduling unrelated to clinical status and that only emergency operations are performed on weekends. As a consequence, some portion of the patients operated on at 3 to 7 days was in that group not because of physician choice but because of scheduling. We repeated our analysis limiting the cohort to the 823 patients whose MI occurred on a Monday or Fig 1. Crude (blue) and direct standardized adjusted (red) in-hospital mortality.

5 ADULT CARDIAC 166 CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL Ann Thorac Surg OPTIMAL TIMING FROM MI TO CABG 2017;103: Table 3. Crude and Adjusted In-Hospital Mortality and Adjusted Odds of In-Hospital Death Among All Patients and Those With and Without ST-Segment Elevation Myocardial Infarction Variable <1 Day (n ¼ 99) 1 2 Days (n ¼ 369) 3 7 Days (n ¼ 1,966) 8 21 Days (n ¼ 626) All patients Mortality, % Crude Adjusted a Adjusted a OR (95% CI) 3.7 ( ) 1.1 ( ) 1 (Ref) 1.3 ( ) STEMI Mortality, % Crude Adjusted a Adjusted a OR (95% CI) 6.7 ( ) 0.7 ( ) 1 (Ref) 3.8 ( ) NSTEMI Mortality, % Crude Adjusted a Adjusted a OR (95% CI) 3.0 ( ) 1.1 ( ) 1 (Ref) 1.2 ( ) a Adjusted using direct standardization or logistic regression for age, ejection fraction <0.40, 3-vessel disease, left main disease, white blood cell count >12,000 ml, prior coronary artery bypass grafting, vascular disease, diabetes, renal failure before the operation, renal failure or preoperative creatinine 2 mg/dl, chronic obstructive pulmonary disease, percutaneous coronary intervention within this admission. All patient models also adjusted for STEMI. CI¼ confidence interval; NSTEMI ¼ non ST-segment elevation myocardial infarction; OR ¼ odds ratio; Ref ¼ reference group; STEMI ¼ ST-segment elevation myocardial infarction. Tuesday and were operated on in 1 to 2 or in 3 to 7 days, reasoning that physician choice for this group would be more closely tied to surgical timing. Among patients admitted on Monday or Tuesday, there was no difference in crude hospital mortality between the 1- to 2-day and 3-to 7-day groups (1.5% vs 1.7%, p ¼ 0.876). After adjustment, patients operated on 1 to 2 days post-mi still had similar in-hospital mortality (OR, 1.2; 95% CI, 0.3 to 4.7) relative to patients operated on in 3 to 7 days post-mi. Secondary Outcomes The difference in the rate of return to the operating room for bleeding or postoperative red blood cells transfused Fig 2. Odds ratio (circles) and 95% confidence intervals (horizontal lines) of inhospital death among timing groups relative to 3 to 7 days between myocardial infarction (MI) and coronary artery bypass grafting (CABG). The vertical dashed line represents no difference in odds ratios relative to 3 to 7 days.

6 Ann Thorac Surg CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL 2017;103: OPTIMAL TIMING FROM MI TO CABG 167 Fig 3. Myocardial infarction-to-coronary artery bypass grafting timing differed by day of admission (p < 0.001). ADULT CARDIAC by timing group was not significant (Table 4). Patients with an MI-to-CABG time of 8 to 21 days had higher rates of red blood cell transfusion (Table 4). The difference in rates of intraoperative or postoperative cerebrovascular accidents or postoperative atrial fibrillation was not significant (Table 4). Comment We found that patients undergoing CABG in 1 to 2 days after an MI had characteristics very similar to patients operated on 3 to 7 days after an MI and had the same inhospital mortality. By comparison, patients undergoing CABG within less than 1 day of MI had different characteristics, with higher rates of STEMI, preoperative percutaneous coronary intervention, and preoperative placement of an intraaortic balloon pump and had significantly higher in-hospital mortality than those that received CABG 3 to 7 days after MI. Ongoing ischemia, unstable refractory angina, and the burden of coronary disease all may have influenced both the timing of the operation and the higher in-hospital mortality rate. Patients operated on 8 to 21 days post-mi also had different characteristics than patients operated on at 1 to 2 or 3 to 7 days. This delayed revascularization group had more comorbidities and were more likely to have had a prior CABG, which may explain why they had a higher predicted mortality. Although their in-hospital mortality rate was not statistically higher than these other groups, there was a trend in that direction. With our data we cannot determine whether these patients could safely be operated on earlier in their hospital stay. We can only speculate that given their increased burden of comorbid conditions, it was thought that additional time was Table 4. Postoperative Outcomes Variable <1 Day (n ¼ 99) (%) 1 2 Days (n ¼ 369) (%) 3 7 Days (n ¼ 1,966) (%) 8 21 Days (n ¼ 626) (%) p Value a In-hospital mortality Red blood cell transfusion Intraoperative <0.001 Postoperative Any <0.001 Bleeding, return to OR CVA, intra-op or post-op Atrial fibrillation, post-op a p values determined by c 2 test. CVA ¼ cerebrovascular accident; OR ¼ operating room; Post-op ¼ postoperative.

7 ADULT CARDIAC 168 CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL Ann Thorac Surg OPTIMAL TIMING FROM MI TO CABG 2017;103: needed to optimize their readiness for the operation. The trends that we found in in-hospital death for the overall cohort did not change when the analysis was stratified by STEMI and NSTEMI patients. CABG timing may have also been influenced by the day of the week patients were admitted. Patients admitted early in the week were more likely to be operated on in 1 to 2 days of MI until Thursday, when patients are more likely to be held through the weekend. However, for patients admitted Monday or Tuesday, who had equal opportunity to be operated on in 1 to 2 days or 3 to 7 days, we still saw no differences in the mortality rate between 1 to 2 days and 3 to 7 days from MI to CABG. These results suggest that when the effect of scheduling is eliminated and physician choice and clinical characteristics drive the timing, mortality is similar between 1 to 2 and 3 to 7 days from CABG to MI. We investigated whether other adverse outcomes were associated with CABG timing. We saw no differences among timing groups in the rate of returning to the operating room for bleeding. Patients operated on in 8 to 21 days were more likely to have received a blood transfusion, but transfusion rates were similar across the remaining timing groups. We did not find any increased bleeding risk among earlier operating timing groups. We also found no difference in intraoperative or postoperative cerebrovascular accident or postoperative atrial fibrillation among timing groups. The optimal timing of CABG after MI has not been investigated using a controlled randomized trial; however, multiple observational analyses have been conducted. The largest contemporary study, by Weiss and colleagues [3], similarly noted a U-shaped relationship between timing of CABG after MI and death. The study also showed death was highest when patients underwent operations on the initial day of admission, a similar finding to the high mortality of the patients who were less than 1 day from MI in our analysis [3]. However, when divided into early (0 to 2 days after admission) and late (3 days after admission) timing of CABG, significantly more deaths occurred in the early timing group [3]. These differences in results are most likely connected with the combination of timing groups used, because Weiss and colleagues showed a similar trend to our results when analyzed by day [3]. Our results were also similar to large study of NSTEMI patients, the analysis of the ACTION Registry-GWTG (Acute Coronary Treatment and Intervention Outcomes Network Registry Get With The Guidelines) by Parikh and colleagues [4]. Their analysis showed no difference in mortality rates between patients undergoing CABG early (<48 hours) or late (>48 hours). Although these results are a combination of our timing groups, we found no difference between our 1 to 2 day and 3 to 7 day groups. Recent smaller studies have demonstrated differences in death by time from MI to CABG [5, 6]. A single-center study by Davierwala and colleagues [5] found no difference in in-hospital mortality rates or long-term survival between NSTEMI patients operated on in less than 24 hours, 24 to 72 hours, or after 72 hours to 21 days [5]. An investigation by Khan and colleagues [6] also found no difference in deaths in patients operated on within or after 24 hours of STEMI. Limitations Our study has several limitations. First, we present data from an observational study, and unmeasured confounding could have influenced our results. Unmeasured differences in patient characteristics, such as myocardium at risk, cardiac troponin levels, or patient frailty, could influence the timing of the operation and in-hospital death. Timing of the operation may have also been influenced by administration of antiplatelet medications. However, our registry only collects data on antiplatelet use within 24 hours preoperatively. Timing may have also been affected by myocardial damage, and although our registry does not include complete troponin data to study this, we investigated proxies such as STEMI and left ventricular function. We also recognize the possibility that myocardial viability may influence timing because the operation may be delayed to assess the viability of a vascular distribution. However, we have no information on how often such an assessment is done and the effect it has on the outcomes. Second, our data represent a regional experience and are generalizable only to the extent that our patients and practices are similar to those in other regions of the country. It is worth noting that several of our clinical prediction rules have performed well when tested in other patient populations, suggesting our experience is generalizable [12]. Third, we have no information on patients who might have died while waiting for their operation. To the extent that this happens, the point estimates of death for patients waiting 3 to 7 days or 8 to 21 days for their operation might actually be falsely low. Implications We found in-hospital mortality rates and patient characteristics were similar for patients operated on in 1 to 2 days and in 3 to 7 days after acute MI. Outcomes within this interval may be based on unquantified clinical judgment, or the interval from MI to CABG may not affect death in this time range. Therefore, these results suggest that reducing the MI-to-CABG interval for some patients may be possible without sacrificing optimal outcomes. Patients with an MI-to-CABG timing of less than 1 day had worse outcomes, and some differences in patient characteristics that suggest patients were operated on early and had a high mortality rate due to ongoing ischemia and unstable angina. Patients operated on in 8 to 21 days had more comorbidities, which likely influenced timing. Because the prediction of death was higher in patients in this group, it is most likely appropriate to delay the operation until high-risk patients are stabilized. The similar number of deaths in patients operated on in 1 to 2 days and in 3 to 7 days after MI in our analysis suggests that waiting 5 days to operate may not be necessary for some low-risk patients, unless the clinical scenario requires delaying the operation. Choosing the

8 Ann Thorac Surg CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL 2017;103: OPTIMAL TIMING FROM MI TO CABG timing of CABG after MI should balance the risk of ongoing ischemia and high-risk comorbidities while reducing unnecessary length of stay and resources for uncomplicated cases. This study was funded by the Northern New England Cardiovascular Disease Study Group. References 1. Culler SD, Kugelmass AD, Brown PP, Reynolds MR, Simon AW. Trends in coronary revascularization procedures among medicare beneficiaries between 2008 and Circulation 2015;131: Lee DC, Oz MC, Weinberg AD, Ting W. Appropriate timing of surgical intervention after transmural acute myocardial infarction. J Thorac Cardiovasc Surg 2003;125: Weiss ES, Chang DD, Joyce DL, Nwakanma LU, Yuh DD. Optimal timing of coronary artery bypass after acute myocardial infarction: a review of California discharge data. J Thorac Cardiovasc Surg 2008;135: Parikh SV, de Lemos JA, Jessen ME, et al. Timing of inhospital coronary artery bypass graft surgery for non-stsegment elevation myocardial infarction patients results from the National Cardiovascular Data Registry ACTION Registry-GWTG (Acute Coronary Treatment and Intervention Outcomes Network Registry-Get With The Guidelines). JACC Cardiovasc Interv 2010;3: Davierwala PM, Verevkin A, Leontyev S, Misfeld M, Borger MA, Mohr FW. Does timing of coronary artery bypass surgery affect early and long-term outcomes in patients with 169 non-st-segment-elevation myocardial infarction? Circulation 2015;132: Khan AN, Sabbagh S, Ittaman S, et al. Outcome of early revascularization surgery in patients with ST-elevation myocardial infarction. J Interv Cardiol 2015;28: Anderson JL, Doty JR. Bypass surgery after non-st-segment elevation myocardial infarction better early than late? JACC Cardiovasc Interv 2010;3: Amsterdam EA, Wenger NK, Brindis RG, et al AHA/ ACC guideline for the management of patients with non STelevation acute coronary syndromes: a report of the American College of Cardiology/American Heart Association task force on practice guidelines. Circulation 2014;130:e Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1999 Guidelines for the Management of Patients With Acute Myocardial Infarction). Can J Cardiol 2004;20: Hillis LD, Smith PK, Anderson JL, et al ACCF/AHA guideline for coronary artery bypass graft surgery: a report of the American College of Cardiology Foundation/American Heart Association task force on practice guidelines. Circulation 2011;124:e O Connor GT, Plume SK, Olmstead EM, et al. Multivariate prediction of in-hospital mortality associated with coronary artery bypass graft surgery. Northern New England Cardiovascular Disease Study Group. Circulation 1992;85: Jin R, Grunkemeier GL, Starr A. Validation and refinement of mortality risk models for heart valve surgery. Ann Thorac Surg 2005;80: ADULT CARDIAC DISCUSSION DR ROBERT A. GUYTON (Atlanta, GA): I appreciate the opportunity to discuss this excellent and beautifully presented study, and I thank the authors for providing the manuscript to me. This study makes an important contribution to the growing body of evidence that in most patients after an acute infarction who have an indication for coronary bypass, it seems safe to operate at 24 to 72 hours after infarct rather than waiting the 3 to 5 days that others have recommended. This is a generalization based on an analysis of patient groups. To apply this to specific patients, we need to dig a little deeper into the sorting of patients into the group. In particular, we need to know more about the details of the clinical work flow and the Northern New England database. First, what led to operation in the 97 patients who had an operation between 6 and 24 hours? Presumably this was related to the opportunity to salvage threatened myocardium or to compelling anatomy or to clinical urgency related to hemodynamic instability or continuing ischemia, but not urgency that rose to the level of an emergent procedure. The problem is that 466 patients, or 13% of the overall study group, were excluded. They were excluded from analysis because of an emergent or salvage status, shock, or operation in less than 6 hours. These 466 patients, if mostly in the less than 1 day cohort, overwhelmed the 97 patients included in the analysis. If the excluded patients were after 24 hours, or a substantial number were after 24 hours, then that is a problem, because it could be then that these patients who had a delayed operation would make a difference in the outcomes, as these patients were excluded from the analysis having waited for surgery and then become emergent. So we need to know more about the sorting of the 466 patients between cohorts. Were a substantial number in the patients that were after 24 hours? Secondly, the manuscript raises some concerns about the generalizability to different areas of this study. Can this be generalized from Bangor, Maine, to Atlanta, Georgia? This is related potentially to field data definitions that are different between the Northern New England database and The Society of Thoracic Surgeons (STS) database, which is now used by 90% of our cardiac surgeons in this country. For example, 50% of the patients in this study were identified as having New York Heart Association class IV status, but only 15% had congestive failure. The STS database applies the New York Heart Association only to congestive failure. Obviously, the New England database applies it to not just congestive failure but other parameters. The second question is, are there important data field definitions that differ between the New England database and the STS database that are relevant to this study? My third question, and finally, is related to the fact that I am struck by the high number of preoperative balloon pumps used in this study. Obviously if there is a difference in clinical work flow between Northern New England and the rest of the nation, then this would be an issue relative to the generalizability of this study. Fifty percent of the patients operated on in the first day had a balloon pump inserted, and these were the nonemergent patients; 27% of the patients operated on in the day 1 to 2 cohort had a balloon pump inserted. You wonder how many of the emergent patients had balloon pumps inserted? So my third question is, have you looked at the use of balloon pumps and the outcomes of patients who had balloon pumps inserted? Thank you very much for this opportunity. It is a great study.

9 ADULT CARDIAC 170 CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL Ann Thorac Surg OPTIMAL TIMING FROM MI TO CABG 2017;103: MS NICHOLS: First, I would like to thank Dr Guyton for his insightful comments and questions. Considering the first question about less than 1 day patients, the less than 1 day patients were patients that were operated on in less than 1 day but had medically refractory ongoing ischemia and were not in shock, deemed salvage, or in need of emergency surgery within hours. In the Northern New England (NNE) database, emergent surgery is defined loosely by the clinician s judgment that the patient needs the operation within hours. So because of that, there are patients in that less than 1 day group who do have ongoing ischemia, and these signs of a high-risk patient but are not considered emergent based on the NNE database definition. Within those that were excluded from the analysis, the majority of those did come from the less than 1 day group. So out of patients who if you kind of took the cohorts before the exclusion, so patients between 2008 and 2014 with a prior myocardial infarction (MI) and timing information, and exclude also the less than 6 hours, we find that about 50% of the less than 1 day patients are excluded due to shock or emergency, 25% from the 1 to 2 days due to shock or emergency, then 5% from 3 to 7 days, and 3% from 8 to 21 days. So we do see this persistent decrease in the number who are excluded due to that shock or emergency status. As they wait, there are fewer that are becoming emergent, but it does signify the need for clinicians to keep in mind the clinical scenario and what that means for whether the patient needs to go to surgery immediately as an emergent patient or whether they should be waiting. The reason we excluded those groups was that we recognized that although there is value in quantifying the patients that have become emergent while waiting for surgery, we felt that excluding patients where the timing was predetermined by that clinical scenario would produce the most meaningful results. That way these results would apply to patients whose surgery could be timed based on clinical judgment. Because guidelines have specified that emergent and shock patients should be operated on as soon as possible, their timing is predetermined by the clinical scenario, and we did not feel that those patients should be compared to those who had timing that could be feasibly altered. So that is why we excluded them and the proportion that got excluded. There are some differences in the NNE and STS definitions. Congestive heart failure (CHF) is collected here as a dichotomous variable and can be collected as current or prior CHF, as it is collected in the NNE. With the STS, it is collected very similarly and then has the New York Heart Association as a follow-up variable, and in the NNE that is similar, but it does apply to a larger group is my understanding. The high number of intraaortic balloon pumps (IABP) that was used was one of the signs for high-risk patients that were operated on in the less than 1 day and some of the 1 to 2 day groups. We did consider how the IABP would act as a marker for highrisk patients, and when we included it in our final model of analysis, our final regression model, it did not actually change the results but did exaggerate them. So all the results remained the same. We did consider IABP as a marker, but, as expected, it did not really change the results as much as just sort of act as a marker for those high-risk patients. With all of those caveats in mind, the NNE is a regional database, so we do only collect the information on what has happened in the NNE, and so our results do need to be considered in how they can be applied to other groups. In the future, it is also important to consider how our practice variation compares to the practice around the country and how to apply our findings in different regions. DR JAMES SHOPTAW (Hazard, KY): I just wanted to know about the use of Plavix and whether or not that had anything to do with the timing of the surgery and is there a set number of days that you wait after on Plavix or not? MS NICHOLS: We did look at that, and the variable that we have in our database is Plavix used within 24 hours of the surgery. So, obviously, there were higher rates of that used within the less than 1 day group, but the rates of that variable were pretty similar between the remaining groups. In general, the physicians in our region do tend to try to wait after Plavix is given in order to minimize the bleeding risk, but it still falls into the timing groups with what we see. We cannot get any farther back, 5 days from Plavix or 8 days from Plavix. We just do not have that in the database. DR JOHN V. CONTE (Baltimore, MD): Congratulations on a nice presentation. I have concerns about the overall applicability of this paper and the data widespread. All institutions are looking at ways to cut costs, and certainly keeping patients in the intensive care unit or the stepdown units waiting for an operation is low hanging fruit for an administrator. Do you have the ability in the NNE database to become more granular with the data; ie, we all know that a small, nondominant right coronary infarct is not the same as a proximal left anterior descending infarct. Do you have the ability to quantify myocardial infarctions by areas, by any of the markers that we are able to look at, or even function on a gross level? Do you have any ability to allow surgeons to use your data to say, no, it is still not safe to operate the day that they have an infarct? MS NICHOLS: We do have some clinical variables of the degree, or mainly it is procedural variables of what vessels were operated on. In looking at our data, what we do find is in the 8 to 21 day group that clinical judgment that caused those patients to wait longer with 8 to 21 days was warranted. So, for us, we are not trying to disown the clinical judgment. We do think that that clinical judgment is a really important driving factor of when the timing should be decided. Within our database we do have some information on the myocardial infarction (MI) that can be considered, but hopefully we can incorporate that more into the final paper. DR CONTE: I would hope that that is a very important point that is made in the paper, that this should not be widely applied without more in-depth clinical variables. MS NICHOLS: Thank you. DR MAHMOOD MIRHOSEINI (Germantown, WI): The paper really coincides with the finding of reperfusion injury as shown in the literature. There are two groups of patients, asymptomatic infarct and symptomatic. The results are quite different between the two groups. MS NICHOLS: We did see signs of the less than 1 day group really being a very different patient population, as I mentioned before, in that they do have these signs of ongoing ischemia, and some of them do still have refractory angina. Does that answer your question? DR MIRHOSEINI: If the patient continues having pain, the injury is in progress and they have a higher mortality with intervention and surgery. However, if the patient is pain free, the process of infarct is finished. These patients are a different

10 Ann Thorac Surg CHAMBERLAIN MEMORIAL PAPER NICHOLS ET AL 2017;103: OPTIMAL TIMING FROM MI TO CABG category and reperfusion injury is less and they probably have less mortality. DR NICHOLS: Yes, I agree with you on that. Mainly why we tried to exclude those emergent patients was trying to get rid of that group that has an ongoing infarct and really should be operated on immediately. So that is why we excluded those emergent patients, and we did have some with ongoing ischemia in the less than one day group still. DR McCULLOUGH: Excellent presentation, Elizabeth. I think it is important to remember that this is an observational cohort study about how we practice cardiac surgery. We did not set out to prospectively randomize any variable to determine when a patient should be operated on. So clinical judgment reigned supreme for all patients at all times. The message of this study is you do not have to arbitrarily wait 72 hours to 5 days simply because a patient has met the definition for an infarct. In our 1 to 2 day group and our 3 to 7 day group, our observed and expected mortalities were equivalent, saying that barring any medically refractory angina driver to the operating room, it is okay to go ahead and operate in a day or two. And tempering the discussion with Dr Conte s questions and others, I think that it is necessary to keep in mind about what we were trying to say and why we were able to say it. MS NICHOLS: Thank you. 171 ADULT CARDIAC