Simple linear regression analysis was used. A, Age. B, Body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared). C, Diabetes mellitus. D, Advanced congestive heart failure (CHF) (New York Heart Association class III or IV heart failure). E, Left main coronary artery disease. F, Independent functional performance status. G, Advanced angina class (Canadian Cardiovascular Society class III or IV). H, Prior myocardial infarction [MI]. I, Ejection fraction of 34% or less. J, Current smoker. K, Urgent surgery. L, Emergent surgery. P < .05 was considered significant for the trends.
Simple linear regression analysis was used. A, Increased prevalence of prior percutaneous intervention. B, Decreased annual CABG volume.
Simple linear regression analysis was used.
Cornwell LD, Omer S, Rosengart T, Holman WL, Bakaeen FG. Changes Over Time in Risk Profiles of Patients Who Undergo Coronary Artery Bypass Graft SurgeryThe Veterans Affairs Surgical Quality Improvement Program (VASQIP). JAMA Surg. 2015;150(4):308-315. doi:10.1001/jamasurg.2014.1700
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Today’s coronary artery bypass grafting (CABG) population appears to comprise sicker patients than in the past; however, little is known about the change in the risk profile.
To evaluate the change with time in the risk profile of patients who undergo CABG.
Design, Setting, and Participants
Retrospective review of records from the Veterans Affairs (VA) Surgical Quality Improvement Program (VASQIP); 65 097 patients who underwent isolated primary CABG from October 1, 1997, to April 30, 2011, were evaluated.
Main Outcomes and Measures
Trends in risk profiles, surgical volume, and modern outcomes in the VA system. We determined the significance of changes in age and major comorbidities across time with simple linear regression analysis and evaluated the rates of perioperative mortality (30-day or in-hospital) and VASQIP predicted risk of mortality trends over time.
From 1997 to 2011, there were increases in mean (SD) patient age (63.1 [9.4] vs 64.3 [7. 8] years; R2 = 0.34; P = .02) and body mass index (28.3 [5.1] vs 30.1 [5.8]; R2 = 0.95). There were also increases in the prevalence of diabetes mellitus (32.8% vs 41.3%; R2 = 0.82), preoperative New York Heart Association (NYHA) class III or IV heart failure status (14.3% vs 34.2%; R2 = 0.74), and left main coronary artery disease (26.0% vs 32.8%; R2 = 0.82) (all P < .001). There was a decrease in the prevalence of advanced angina severity (Canadian Cardiovascular Society class III or IV) (R2 = 0.95), previous myocardial infarction (R2 = 0.82), and low ejection fraction (≤34%) (R2 = 0.88) (all P < .05). There was no significant change in the prevalence of cerebrovascular and peripheral vascular disease, chronic obstructive pulmonary disease, or 3-vessel coronary artery disease. Perioperative mortality rates and the VASQIP predicted risk of mortality, respectively, decreased with time (3.2% and 3.1% vs 1.7% and 1.6%). From 2004 to 2011, there was a significant increase in the prevalence of previous percutaneous coronary intervention (18.6% to 29.2%; R2 = 0.82; P = .002). Overall CABG volume decreased (5551 in 1998 vs 3857 in 2012; R2 = 0.95; P < .001).
Conclusions and Relevance
From 1997 to 2011, there was a progressive increase in the prevalence of obesity, diabetes, left main coronary artery disease, and advanced NYHA heart failure class among VA patients undergoing CABG. The prevalence of previous myocardial infarction, low ejection fraction, and advanced angina decreased, perhaps because of earlier surgical referral, improvement in medical management, or a shift in patient selection for CABG. Operative mortality also decreased with time. These trends confirm the general perception of significant, ongoing improvement in the care of patients who undergo CABG in the VA, despite an older, sicker population.
Recent reductions in mortality due to coronary artery disease (CAD) in the United States have been attributed to epidemiologic factors and better overall medical care, including advanced CAD revascularization strategies with percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) surgery.1 The trend toward improved survival has been accompanied by a decline in the number of CABG operations performed annually in the United States over the past decade.2,3 This decrease in surgical referrals may represent a decreased need for CABG because of improved results of noninvasive treatment as well as a shift in patient selection for surgery, but the availability of less-invasive alternatives with advanced PCI stenting techniques is thought to play an important role. A general perception persists among cardiac surgeons that current patients referred for coronary surgery are more likely to have undergone PCI before undergoing definitive CABG revascularization and are generally older and sicker than were patients who underwent CABG in the past.
However, to our knowledge, few studies have documented recent trends in age and risk profiles for patients in the United States who receive CABG. One report, based on the Society of Thoracic Surgeons Adult Cardiac Surgery Database,4 confirmed an increase in the proportion of patients who had PCI before being referred for CABG in 2000 to 2009 and also showed an increase in some risk factors, such as diabetes mellitus and chronic obstructive pulmonary disease (COPD), but not much change in age. They also confirmed a recent ongoing trend toward reduced CABG morbidity and mortality rates, with an associated decrease in the predicted risk of mortality (PROM) on the Society of Thoracic Surgeons’ risk-adjustment model. However, a reduced PROM is not necessarily a true indication of a reduction in comorbidities because the models are designed to evolve to represent current outcomes. Ongoing improvement in outcomes is more likely the result of improved surgical care than preoperative factors alone. The effect of scrutinized outcome reporting may also play a large role in patient selection for CABG, with surgeons carefully choosing their patients in an attempt to maintain the best possible outcomes; this practice has unknown effects on the risk-factor profiles of current patients undergoing CABG.5- 7
The veteran patient population is unique and has an increased prevalence of various risk factors when it is compared with the general population of patients referred for CABG. The Veterans Affairs (VA) system, a single-payer system, also has a different approach to patient management than does the US health care system at large; in the VA system, it is possible that more collaboration between physicians can occur without the strain of billing and productivity concerns. This led us to ask: Are the veteran patients who have recently undergone CABG older and sicker than similar patients in the past? Alternatively, have risk factors decreased, possibly because of changes in patient selection?
To answer these questions, we looked at a large national database of US veteran patients8 who underwent CABG surgery and examined the trend in risk profiles. In addition, we evaluated trends in surgical volume and modern outcomes in the VA system. We hypothesized that the trend would be toward older and sicker patients over time.
This study was approved by the Baylor College of Medicine and Michael E. DeBakey Veterans Affairs Medical Center institutional review board. The requirement for informed consent was waived. The VA Surgical Quality Improvement Project (VASQIP) is part of the VA Surgical Quality Improvement Program9,10 and prospectively collects risk and outcomes data regarding all patients who undergo cardiac operations at 42 VA cardiac surgery centers that participate in this nonvoluntary database.11 We received approval for the study from the VA Surgical Quality Use Data Group. We identified all patients who underwent CABG at the participating VA hospitals between October 1, 1997, and April 30, 2011. The VASQIP’s data fields and definitions were used. Patients who underwent reoperations and combined procedures were excluded. The VASQIP database includes complete vital status data, and time to death from the date of the operation was computed for patients who died before May 17, 2011.
We evaluated the trend over time for the preoperative factors recorded in the database, and we focused on factors considered to increase surgical risk, including advanced age, female sex, diabetes mellitus, increased body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared), New York Heart Association class III or IV heart failure, cerebrovascular disease, peripheral vascular disease, COPD, current smoking, and dependent functional status. We also looked at the trend for operative volume, prevalence of previous PCI, previous myocardial infarction, advanced angina severity, low ejection fraction (≤34%), and left main and 3-vessel CAD. Outcomes were analyzed by using the definitions recorded in the VASQIP database, with estimated mortality calculated with the use of the VASQIP PROM and operative mortality defined as death within the index hospitalization or within 30 days postoperatively. The trends were graphed over time, and a simple linear regression analysis was done to test for significance of the slope.
We also analyzed the data in another way by dividing the cohort into 2 periods: 1997-2003 (the early cohort) and 2004-2011 (the later cohort). We compared the variables between these 2 cohorts as another way to report the general trends over time. We used the 2-tailed, unpaired t test for continuous variables and the χ2 test for categorical variables. Significance was defined as P < .05. All statistical analyses were conducted with SAS, version 9.1 (SAS Institute Inc).
From October 1, 1997, to April 30, 2011, isolated primary CABG was performed on 65 097 patients in the VA system. Figure 1 shows the change over time, with graphs of the factors that demonstrated significant trends. During the study period, there was a slight increase in the mean (SD) patient age (63.1 [9.4] years in 1997 vs 64.3 [7.8] years in 2011; P = .02) and a more substantial increase in the BMI (28.3 [5.1] vs 30.1 [5.8]) and prevalence of diabetes (32.8% vs 41.3%), New York Heart Association class III or IV heart failure status (14.3% vs 34.2%), and left main CAD (26.0% vs 32.8%) (P < .001 for all). However, there was a decrease in the prevalence of advanced angina severity (78.6% Canadian Cardiovascular Society class III or IV vs 54.8% in 2011), previous myocardial infarction (55.4% vs 48.7%), and low (≤34%) ejection fraction (12.0% vs 6.6%) (P < .001 for all). There was a significant (P = .04) trend toward a decrease in the number of current smokers (31.1% vs 27.3%). An increased number of patients with independent functional status was documented (89.1% vs 93.1%; P < .001). The number of cases with urgent priority increased (10.6% vs 16.4 %), but the number of emergency cases decreased (6.4% vs 2.8%) (both P < .001).
There was no significant change in the prevalence of cerebrovascular and peripheral vascular disease, COPD, or 3-vessel CAD. The percentage of women was low (1%) throughout the study period, without a significant change over time. Female sex was not associated with an increased risk for operative death or complications on multivariate analysis.
We also evaluated geographic differences by comparing the Northeast, Southeast, West, Central, and Middle South regions and racial and ethnic variation. Compared with the other 4 regions combined, the Northeast had a lower prevalence of diabetes (15.4% vs 16.7%; P = .009), the West had a lower prevalence of smoking (26.8% vs 30.1%; P < .001), and the Central Region had a higher mean BMI (29.5 [5.5] vs 29.2 [5.4]; P < .001). Race and ethnicity data were available only after 2004. The Southeast and Middle South combined had a higher number of African American patients (12.8%) than did the other regions (6.8%) (P < .001), and the Southeast and West combined had a higher number of Hispanic patients (10.5%) than did the other regions (5.1%) (P < .001). In multivariate models used to evaluate predictors for adverse outcomes, race and ethnicity were not significant predictors, but geographic area was a significant predictor.
Detailed data concerning the prevalence of previous PCIs were not available in the VA database until 2004. From 2004 to 2011, there was a significant increase in the number of patients who had undergone previous PCI (P = .002) (Figure 2). During the same period and over the course of the entire study, there was a trend toward a reduced volume of CABG operations in the VA system, from 5551 operations in 1998 to 3857 procedures in 2010. Perioperative mortality rates and VASQIP PROM decreased with time, being 3.2% and 3.1%, respectively, in 1997, compared with 1.7% and 1.6%, respectively, in 2011 (Figure 3). Overall morbidity rates and rates of specific complications also decreased.
The early cohort included 33 155 patients who underwent isolated CABG from October 1, 1997, to December 31, 2003. The later cohort included 31 942 patients who underwent isolated CABG from January 1, 2004, to April 30, 2011. Table 1 compares the 2 cohorts. By using this method to evaluate the database, we confirmed the trends reported in Figure 1, except that a significant decrease in peripheral vascular disease was seen; moreover, the change in age and the number of current smokers was not significant with this method. Table 2 reports the outcomes for the 2 cohorts. The mortality rate decreased from 3.0% in the early cohort to 1.9% in the later cohort.
Our study revealed trends toward increased age and prevalence of comorbidities in VA patients who underwent CABG over nearly 14 years, from October 1, 1997, to April 30, 2011. The trend toward older age was similarly shown in a recent study of the Society of Thoracic Surgeons’ database, in which age marginally increased among US patients who underwent CABG from 2000 to 2009.4
Our study confirmed an increase in the number of patients who underwent CABG after previous PCI, as well as a decline in operative volume over the same time. Although the exact cause and effect is unclear, an increased prevalence of previous PCI and advanced congestive heart failure (CHF) may indicate that patients are referred for surgical evaluation later in the course of their CAD, more often after failed PCI. However, improved medical management before surgery may be suggested by the data showing a reduction in the prevalence of advanced angina, previous myocardial infarction, and low ejection fraction. The increased prevalence of advanced CHF is difficult to reconcile with the decreased prevalence of a low ejection fraction unless one considers that the CHF class is a functional consideration. Thus, worsening CHF may also be related to patients’ inability to tolerate their heart condition, perhaps because of increased age and noncardiac comorbidities.
The increased prevalence of diabetes noted in our study may reflect improved consideration of evidence-based guidelines for surgical referral as a result of recent trials12- 16 showing improved outcomes for CABG compared with PCI revascularization in patients with diabetes. Therefore, our data may represent improved understanding of the evolving indications for surgical revascularization and of the risk-benefit analysis for particular patient groups.
Although we largely confirmed the perception that the CABG patient population in the VA system has become older and sicker, we found that some comorbidities were reduced, such as current smoking. This factor probably reflects a general US trend toward decreased smoking, an epidemiologic change rather than a risk-factor change specific to the CABG population. Other factors, such as COPD and cerebrovascular disease, showed no significant change in our study. Nevertheless, we confirmed certain risk-factor profile characteristics that are specific to veterans. Comparison of our study population with those described in other reports, such as a recent evaluation of the Society of Thoracic Surgeons’ database,4 shows that our population had a high prevalence of COPD, current smoking, peripheral vascular disease, cerebrovascular disease, and male sex; all of these factors were more prevalent in our veteran population than in the general US CABG population.
Regardless of the risk-factor trends, the observed morbidity and mortality rates have continued to improve in the VA system as they have in the Society of Thoracic Surgeons’ database. Medical and surgical care for patients with CAD have evolved over time, resulting in better understanding of factors that affect surgical indications, patient selection, preoperative optimization, and perioperative care. The trends observed in this study may best be explained by further considering the process-related improvements that have occurred during the past decade.
First, let us consider how potential improvements in preoperative care, patient selection, and medical management of CAD may have affected the CABG patient population. Better understanding of the effects and appropriate use of medications, such as statins, β-blockers, angiotensin-converting enzyme inhibitors, and antiplatelet agents, may account for the reduction in myocardial infarctions and better control of angina before CABG is performed. Within the VA system, the opportunity for a true “heart-team” approach, with active collaboration between specialties, may help improve clinical judgments concerning the best procedure and its timing for a particular patient.
Preoperative optimization of patients’ conditions with improvement in clinical judgment likely explains why there has been a decrease in emergency CABG cases but an increase in urgent cases. Cardiologists and cardiac surgeons now know that emergency surgery is one of the most important risk factors for operative morbidity and mortality. Our results indicate that more patients are currently able to have their condition stabilized preoperatively and, thus, to undergo surgery urgently instead of emergently. The proportion of patients who need surgery emergently should further decrease as knowledge and experience increase.
Improved patient selection is also likely reflected in our study by the trend toward an increased prevalence of independent functional status. Recent studies17 have discussed the issue of frailty as a surgical risk factor, and surgeons are becoming more keenly aware of the effect of poor functional status on outcome. Perhaps patients with these characteristics receive better treatment with fewer invasive interventions.
Furthermore, let us consider how operative and postoperative care have improved considerably over time. Each of the 42 cardiac surgical programs currently in the VA system has been carefully vetted and undergoes continuous quality monitoring. Intraoperative techniques for CABG have been widely evaluated and scrutinized throughout the medical literature, resulting in multiple concomitant areas of improvement, such as the use of internal mammary artery and other arterial grafts, limited pump time, complete revascularization when feasible, blood conservation, use of hemostatic agents, improved sternal reconstruction methods that limit sternal wound breakdown, off-pump surgery when appropriate,18- 20 and epi-aortic ultrasonography to guide cannulation. Many factors combined have led to improved technical expertise, resulting in increased operative success as further experience has been gained.
Postoperative care has also evolved and improved. Cardiac surgery has been carefully scrutinized for decades, and outcomes reporting has been accompanied by process improvement. When a complication occurs, experienced cardiac surgeons are often able to manage the situation and prevent mortality; this success is reflected in mortality rates that have improved more than morbidity rates. During the past decade, improved critical care has resulted from better nursing care, understanding and use of vasoactive medications, glucose control,21,22 arrhythmia management, ventilator management, prevention of arterial line complications, and infection control. Thus, the effect of age and comorbidity is further mitigated by meticulous postoperative care.
One limitation of our study is that it was retrospective and therefore subject to bias. We cannot track the number of patients who were not selected for CABG. In addition, we did not track the number of veteran patients referred to hospitals outside the VA system for CABG, and this also may bias the results of our study. In the future, fee-basis claims data could be used to study outside referrals. Another limitation of the study is that because of the nature of the veteran population, approximately 99% of the patients were male, and our results are not necessarily generalizable to the overall CABG population in the United States. In addition, the available database does not capture all pertinent factors. However, our study is strengthened by the inclusion of a large number of patients whose data were obtained from a nonvoluntary, multi-institutional national database as well as by the ability to evaluate the unique veteran patient population.
Further investigations should include longer follow-up observation times and analysis of risk factors that have the strongest effect on patient survival as well as further investigation of trends and outcomes in the use of PCI and surgical revascularization. Future clinical care goals that might emerge from this and other outcome studies could be to enhance collaboration with cardiologists to improve patient selection for PCI vs CABG in accordance with the national appropriateness criteria guidelines for revascularization.23 In addition, for CABG to compete effectively with PCI as the optimal revascularization strategy for most patients with multivessel CAD, cardiac surgeons must continue to strive to decrease the rates of postoperative stroke and other complications.
Our study showed a general trend toward older and sicker patients undergoing CABG in the VA population along with a reduction in CABG volume and an increase in the proportion of patients who had a previous PCI. Operative outcomes continue to show improvement with time.
Accepted for Publication: June 19, 2014.
Corresponding Author: Lorraine D. Cornwell, MD, Division of Cardiothoracic Surgery, Michael E. DeBakey Veterans Affairs Medical Center, 2002 Holcombe Blvd, Operative Care Line 112, Houston, TX 77030 (firstname.lastname@example.org).
Published Online: February 11, 2015. doi:10.1001/jamasurg.2014.1700.
Author Contributions: Drs Cornwell and Bakaeen had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Cornwell, Rosengart, Holman, Bakaeen.
Acquisition, analysis, or interpretation of data: Cornwell, Omer, Holman, Bakaeen.
Drafting of the manuscript: Cornwell, Rosengart.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Cornwell, Omer.
Administrative, technical, or material support: Cornwell, Rosengart, Holman.
Study supervision: Cornwell, Rosengart, Bakaeen.
Conflict of Interest Disclosures: None reported.
Disclaimer: The opinions expressed herein are those of the authors and not necessarily those of the Department of Veterans Affairs or the US government.
Previous Presentation: This research was presented at the 38th Annual Surgical Symposium of the Association of Veterans Affairs Surgeons; April 6, 2014; New Haven, Connecticut.
Additional Contributions: Virginia C. Fairchild, BA, and Stephen N. Palmer, PhD, Texas Heart Institute Department of Scientific Publications, contributed to the editing of the manuscript. There was no financial compensation.