Effect of Cardiac Surgery Patient Characteristics on Patient Outcomes From 1981 Through 1995
Background Previous studies have demonstrated increased risk in patients undergoing coronary artery bypass surgery (CABG), but the effect of this increasing risk on outcomes has not been well documented. The purposes of this study were (1) to evaluate patients who had CABG from 1993 to 1995 (group III) and compare them with patients from 1981 through 1987 (group I) and 1988 through 1992 (group II) to determine if the trend toward higher-risk patients continued and (2) to evaluate what effect risk had on in-hospital outcomes.
Methods and Results Data were collected prospectively on patients undergoing CABG. Patients from the three time periods were compared by use of univariate and multivariate statistics. Risk models for mortality were developed by use of logistic regression. Significant changes were noted in the three time periods, with risk increasing over time. Increased risk was associated with increased mortality in group II, but mortality declined in group III despite the continued increase in patient risk. Group II had an increase in complications, with little change in group III. The actual mortality rate was lower than predicted in group III.
Conclusions Patients undergoing CABG are increasingly high risk. In-hospital mortality rates declined during the period from 1993 through 1995 and were lower than predicted despite the increase in risk. This decreased mortality rate may reflect greater experience in providing care to high-risk patients and improved myocardial protection and surgical and anesthetic techniques. Although continued analysis of patient risk and benefit is needed, researchers must be cognizant of the rapid changes in technology and knowledge and should correlate changes in the process of care with outcomes.
Coronary artery bypass graft surgery is an important and effective intervention for revascularization in patients with coronary artery disease. The number of CABG procedures performed yearly has increased steadily since the early 1980s,1 and the benefits of surgery have been extended to patients previously judged as “too sick” or “too old.” Studies have documented an increase in the mean age of patients undergoing CABG and in the percentage of octogenarians having cardiac surgery.2 3 4 5 It has also been noted that improvements in medical therapy and technological advances in coronary angioplasty have changed the referral patterns for surgical revascularization so that more of the referrals are high-risk patients.6
Previously published data from Emory Healthcare indicated that CABG patients through 1991 were getting older and that correlates of in-hospital death and complications (such as increased age) were becoming more common.7 8 This trend, if it continues, has implications for health policy, costs, and the outcomes of care. The purposes of this study were (1) to evaluate the characteristics of patients undergoing CABG during a later time period (1993 through 1995) and compare them with patients from earlier time periods (1981 through 1987 and 1988 through 1992) to determine if the trend toward older and sicker patients was continuing and (2) to evaluate what effect the trend toward higher-risk patients has had on in-hospital patient outcomes.
The study population comprised 23 512 patients who had CABG from 1981 through 1995. Group I (1981 through 1987) consisted of 12 088 patients, group II (1988 through 1992) comprised 7298 patients, and group III (1993 through 1995) comprised 4126 patients. Patients undergoing concomitant valve repair or other types of cardiac surgery were excluded from this analysis.
Angina class was defined by the Canadian Cardiovascular Society classification9 and congestive heart failure by the New York Heart Association classification system.10 Information such as history of diabetes, heart failure, hypertension, and previous MI were obtained from each patient’s history and physical examination. Surgery was classified prospectively as elective, urgent, or emergent by the operator. Complications were also classified by the operator and included sternal or leg-wound infections, pneumonia, postoperative MI, adult respiratory distress syndrome, major arrhythmias (atrial fibrillation, ventricular dysrhythmias, and bradyarrhythmias requiring pacemaker intervention), postoperative angina, and neurological events (stroke, transient ischemic attack, and prolonged depressed mental state).
Data were collected prospectively on patients undergoing cardiac surgery and were entered into a computerized database. Patients in the three groups were compared by use of Pearson’s χ2 statistic for categorical variables and ANOVA procedures for continuous variables. The α-level of significance was established as .05 for the overall three-group comparisons and as .025 for comparisons between groups I and II and groups II and III. Multivariate analysis of mortality and development of risk-adjusted models for mortality were done with the use of logistic regression. All statistical analyses were performed with BMDP statistical software.
Significant changes in group characteristics were noted in the three time periods (Table 1⇓), with patient risk significantly increasing over time. The greatest change occurred between groups I and II, but there was continuing change toward a higher-risk profile between groups II and III on several measures. Mean age and the percentage of patients older than 65 years were significantly higher in group II than group I (P<.0001) but showed very little change from group II to group III. However, the oldest patients (those ≥80 years) increased significantly from group I to group II and again from group II to group III (both P<.0001).
The percentage of patients with diabetes, hypertension, hyperlipidemia, nonelective surgery, and class III and IV angina all increased significantly from group I to group II (all P<.0001). Significant increases also occurred in the percentage of patients with diabetes, hypertension, and hyperlipidemia between groups II and III (all P<.001). The percentage of patients with heart failure increased from 12% to 14% between groups II and III (P<.002).
Mean EF decreased over the three time periods between groups I and II and between groups II and III (P<.0001). The percentage of patients with an EF ≤35% increased significantly from group I to group II (P<.0001) and from group II to group III (P<.0004). The percentage of CABG patients who were women increased significantly from group I to group II (P<.0001) with no further increase in group III. Cigarette smoking declined significantly from group I to group II (P<.0001), but the decline reversed in group III (P<.0001).
Along with the change in patient characteristics among the three time periods, there were accompanying changes in patient outcomes (Table 2⇓). The most significant change was the difference in mortality rates found among the three groups. The increased risk profile in group II was associated with an increase in the mortality rate over that of group I (P<.0001). However, despite a continuing increase in patient risk, mortality declined slightly although not significantly in group III.
Other perioperative complications showed some change among the three groups. As can be seen in Table 2⇑, there was a decrease in postoperative MI and wound infections from group I to group II ( both P≤.0001), but other complications showed increases. Most notably, rates of pneumonia, sternal wound infections, postoperative angina, arrhythmias, and neurological events increased from group I to group II (all P<.0002). Use of an IABP also increased from group I to group II.
In contrast, many complications showed either no change or decreased from group II to group III. Group III did show an increase in pneumonia and use of an IABP, but these increases were not statistically significant. The incidence of arrhythmias (atrial fibrillation or flutter, ventricular dysrhythmias, and bradyarrhythmias requiring pacing) changed only minimally over time. The incidence of postoperative atrial fibrillation alone was ≈14% at all time periods.
When female and male mortality rates were analyzed separately, significant change was found across the three time periods (Table 3⇓). Similar to the overall mortality rate, death rates for women and men increased in group II and declined in group III, although the latter difference was not significant. Female mortality rates were also significantly higher than male mortality rates for all three time periods (P≤.0005). Given that the percentage of patients aged ≥80 years increased in groups II and III, mortality was analyzed separately for these patients within each group. After a >3-fold increase in mortality in group II compared with group I, mortality for the oldest patients declined in group III.
The incidence of any complication in the oldest patients (≥80 years) was significantly higher than for younger patients at all three time periods (P≤.0005) (Table 4⇓). Patients aged ≥80 years had total complication rates of 29%, 39%, and 38% for the three time periods, respectively, compared with 18%, 24%, and 23% for the younger patients. Neurological events were higher in those aged ≥80 years in group I (P<.0001), and arrhythmias occurred more frequently in the oldest patients in group III (P<.0006). Both neurological events and pneumonia increased in frequency over the three time periods for the oldest patients and those aged <80 years, although neurological events did not increase significantly in patients aged ≥80 years. Although the use of an IABP increased in both age groups, the biggest trend was seen in the group III patients who were ≥80 years old. There was no significant difference in the overall incidence of complications other than death between women and men for any of the time periods.
Given the changes in outcomes, we analyzed which characteristics in each time period were significantly associated with higher mortality. The univariate odds ratios for dying were significantly higher for all three time periods for those patients who were older than 65 years, were female, had nonelective surgery, and had hypertension. Heart failure and diabetes had higher odds ratios for mortality in groups I and II, and patients in groups II and III aged ≥80 years had an increased odds ratio for dying. An EF of ≤35% had an odds ratio for mortality of >2 in groups I and III. The combined odds ratios and 95% CIs are given in Table 5⇓.
Multivariate analysis using logistic regression was used to determine the preoperative predictors of mortality for each group. For group I, nonelective surgery, age >65 years, EF ≤35%, female sex, hypertension, heart failure, and diabetes mellitus were associated with mortality. Group II variables changed slightly from group I: EF was no longer a predictor and, as might be predicted from the univariate analysis, reoperation and age ≥80 years were significant predictors. Group III included reoperation, nonelective status, age >65 years, female sex, EF ≤35%, and hypertension as the significant variables. Odds ratios and 95% CIs for the multivariate analyses are found in Table 6⇓.
The multivariate analysis was also run using the entire data set. The total mortality rate was 2.8%. A variable for the time period was included as a preoperative characteristic. The following variables were found to be associated with higher mortality rates: nonelective status, age >65 years, reoperation, heart failure, diabetes, hypertension, EF ≤35%, female sex, membership in group II, and age ≥80 years.
Given the changes in patient risk and mortality over time, a risk-adjusted model for mortality was developed using the multivariate predictors in group I. Because there were very few reoperations in group I, the group I risk model was used to predict the mortality rate in patients undergoing first CABG in groups II and III. In group II, the predicted mortality rate was lower than the actual mortality rate, but in group III, it was higher than the actual rate. Reoperation had significant univariate and multivariate odds ratios for death in both groups II and III, and therefore a second risk-adjusted model was developed that was based on the multivariate predictors in group II. This model was used to predict mortality in group III, and again, the predicted mortality rate was higher than the actual mortality rate. These data are presented in Table 7⇓.
The trend toward extending the benefits of CABG to patients who are older and at higher risk appears to be continuing. Patient risk increased most sharply from group I to group II, but group III showed continuing increases in the percentage of patients with lower EFs, hypertension, and diabetes and those undergoing reoperation. Although initially an increase in mortality rate accompanied the increase in high-risk patients, data from later years (1993 through 1995) indicate that mortality rates tended to decline. This is an important finding because it demonstrates that the benefits of surgery can be extended to patients who may be most in need of revascularization. The decrease in mortality in group III is gratifying and may be reflective of greater experience in providing care to high-risk patients and improved myocardial protection and surgical and anesthetic techniques, coupled with better postoperative medical and nursing care.
The continuing high-risk profile of patients from 1993 through 1995 also demonstrates that changes in healthcare economics (eg, decreased reimbursement for CABG through global package pricing and other arrangements and increasing surveillance of outcomes of care) have not necessarily increased the incentive to select lower-risk patients. Although it is true that complications and the complexity of care required for high-risk patients increase the length of stay and the cost associated with CABG,11 12 13 economic pressures have not contributed to the selection of low-risk or “ideal” patients at this academic medical center.
Changes in complication rates among the three time periods are interesting. Higher rates of some complications (notably neurological events, pneumonia, sternal wound infection, postoperative angina, and use of an IABP) occurred in group II, and a few complications decreased (leg-wound infection and postoperative MI). There were no significant increases in complications found between groups II and III. Some investigators contend that an increased rate of complications may be unavoidable to some extent with higher-risk patients and that the mortality rate remains the best indication of quality of care.14 A previous study found that there was a poor correlation between hospital rankings based on complications and hospital rankings based on mortality. Mortality rate and a measure defined as “failure to rescue” were appropriately associated with other quality-of-care indicators, but complication rates were not.14 Others have argued that adverse events after cardiac surgery are a more sensitive measure of quality of care than mortality rates.15
The incidence of any complication was higher in the patients aged ≥80 years for all three time periods, a not- unexpected finding. Other studies4 5 16 have documented higher complication rates and higher costs associated with CABG in the elderly. Weintraub et al7 8 found that increasing age was a multivariate correlate of in-hospital stroke and increased incidence of wound infection, although another study17 did not find age to be associated with increased risk of stroke. Neurological events were higher in the oldest patients for all three time periods, although this was statistically significant only in group I. Rates of pneumonia, arrhythmias, and use of an IABP were also higher in the oldest patients. Leg-wound infection rates were slightly higher in the patients aged ≥80 years, but the differences were not significant.
Age and Sex
Patients aged ≥80 years made up almost 5% of the patients in group III, a substantial increase from group I. In a previous report, Weintraub et al8 reported on the trend toward older and increasingly ill patients from 1981 to 1989. In that earlier analysis, it was found that only 1% of the patients in the sample were >80 years of age. Two other studies from the 1980s to the early 1990s documented that patients ≥80 years old made up 1% of cardiac surgery patients and 2.2% of CABG patients.4 5 The present study found that there was a >3-fold increase in the number of patients aged ≥80 years from group I (1.3%) to group III (4.6%). Although the mortality rate was significantly higher in the patients aged ≥80 years for groups II and III, which is consistent with previous studies,4 5 8 the mortality rate for the oldest patients followed the same trend and also declined in group III. Although long-term outcomes were not measured in the present study, other studies have reported that selected patients aged ≥80 years do benefit from cardiac surgery despite higher mortality and morbidity rates. The majority of these patients have been reported to experience significant symptom relief, improved performance status and activity levels, and an acceptable quality of life with follow-up to 3 years after surgery.4 5 18
Mortality for women remained significantly higher than for men for all three time periods. The female-male difference in mortality rate is consistent with previously published reports7 and remained significant even when controlling for other variables in this analysis. Although age and comorbidities such as heart failure were controlled in the multivariate analysis, our analysis did not control for body surface area, which has been shown to negate the gender effect in some studies.6
Mortality in group II for patients undergoing first CABG was higher than predicted by a risk-adjustment model that used multivariate predictors from group I. Inclusion in group II also appeared as a multivariate predictor of mortality when logistic regression was performed for the entire data set. Although it is impossible to determine the reason from these data, it may represent a learning curve associated with the change toward higher-risk patients. Alternatively, other important risk factors may not have been included in the regression analysis that could have decreased the predicted mortality rate. The modest trend toward an overall lower mortality rate in group III becomes even more salient in light of the difference between the predicted and actual mortality rates in groups II and III. Although the absolute decline in the mortality rate in group III is small, it is lower than the predicted mortality rate by both risk models. It is interesting to speculate that a learning-curve phenomenon and improved intraoperative and postoperative therapies have reversed a trend toward higher mortality rates associated with higher patient risk. Other researchers have correlated improvements in myocardial protection techniques with both a decrease in mortality rate and improved regional contractility in patients undergoing emergency surgical revascularization after failed angioplasty.19 The improved outcomes in that study occurred despite an increasingly high-risk profile of the patients.
The limitations of this retrospective study include the fact that outcomes beyond the initial hospitalization were not evaluated, and despite consistent definitions, there is the potential for change in how specific variables were valued over the time frames used. Data completion on potentially important variables such as cerebrovascular disease, preoperative cardiogenic shock, and renal disease were insufficient during the earlier time frames to allow for meaningful comparisons. The potential contribution of these variables to mortality could not be analyzed. It is possible that the predicted mortality rate was therefore underestimated, but this would enhance the finding that the actual mortality rate was lower than predicted in group III. These changes in outcomes have not been correlated with specific changes in myocardial protection, operative technique, or postoperative care. The data presented here reflect the experience of one academic healthcare system and thus may not be applicable to other settings.
Patients undergoing CABG are increasingly high risk by accepted risk factors such as older age, lower EF, reoperation, nonelective status, and preoperative comorbidities. This trend continues despite economic pressure in the healthcare system and increasing scrutiny of procedural outcomes. It is noteworthy that in-hospital mortality rates declined during the period from 1993 through 1995 and were lower than predicted despite the increase in high-risk patients undergoing CABG. More information is needed on the long-term outcomes of high-risk patients, including functional status and quality of life. Continued study and consideration are needed to monitor the increasing risk profile of CABG patients and to maintain a balance between risk and benefit. However, any analysis must be cognizant of the rapid changes in technology and knowledge occurring in cardiovascular surgery and should attempt to correlate changes in the process of care with patient outcomes.
Selected Abbreviations and Acronyms
|CABG||=||coronary artery bypass graft surgery|
|IABP||=||intra-aortic balloon pump|
- Received February 19, 1997.
- Revision received March 31, 1997.
- Accepted April 2, 1997.
- Copyright © 1997 by American Heart Association
Deiwick M, Mollhoff T, Budde T, Scheld HH. Cardiac surgery in patients aged 80 years and above: does outcome justify significant postoperative morbidity? Cardiology in the Elderly. 1995;3:381-386.
Weintraub WS, Wenger NK, Jones EL, Craver JM, Guyton RA. Changing clinical characteristics of coronary surgery patients: differences between men and women. Circulation. 1993;88(pt 2):79-86.
Weintraub WS, Craver JM, Cohen CL, Jones EL, Guyton RA. Influence of age on results of coronary artery surgery. Circulation. 1991;84(suppl III):III-226-III-235.
The Criteria Committee of the New York Heart Association. Diseases of the Heart and Blood Vessels: Nomenclature and Criteria for Diagnosis. 6th ed. New York, NY: New York Heart Association/Little Brown & Co; 1964.
Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation. 1989;79(suppl I):I-3-I-12.
Lazar HL, Fitzgerald C, Gross S, Heeren T, Aldea GS, Shemin RJ. Determinants of length of stay after coronary artery bypass graft surgery. Circulation. 1995;92(suppl II):II-20-II-24.
Weintraub WS, Jones EL, Craver J, Guyton R, Cohen C. Determinants of prolonged length of hospital stay after coronary bypass surgery. Circulation. 1989;80:276-284.
Brandrup-Wognsen G, Berggren H, Hartford M, Karlson BW, Karlsson T, Herlitz J. Mortality and morbidity during a period of 2 years after coronary artery bypass grafting in relation to age. Cardiology in the Elderly. 1995;3:289-295.
Vogt AR, Funk M, Remetz M. Comparison of symptoms, functional ability, and health perception of elderly patients with coronary artery disease managed with three different treatment modalities. Cardiovasc Nursing. 1994;30:33-38.