The age of patients treated for coronary heart disease continues to rise because of the aging of the population and improved survival after diagnosis, so more older patients are referred for coronary artery bypass graft (CABG) surgery and percutaneous coronary intervention (PCI) (1). Although the clinical outcomes of CABG and PCI for multivessel disease have been compared in a number of randomized trials, the comparative effectiveness of CABG and PCI in relation to patient age has been uncertain.

In a collaborative meta-analysis of 10 randomized trials that compared CABG with PCI for multivessel disease (2), we found no overall difference in mortality, but age seemed to modify the treatment effect. The purpose of this study was to assess further the effect of age on clinical outcomes after CABG compared with those after PCI.

We analyzed data pooled from 10 randomized clinical trials comparing CABG with PCI for patients with multivessel disease. Methods and main results have been published (2). In brief, each trial shared individual patient data for pre-specified variables (6) that were pooled in a central database for analysis.

The primary outcome was all-cause mortality. Secondary outcomes were the composite of death or myocardial infarction (MI), repeat coronary revascularization, and angina at 1 year of follow-up.

The primary analysis used age as a continuous variable, but for descriptive purposes the study population was divided into tertiles of age (≤56.6 years, 56.7 to 65.1 years, and ≥65.2 years). Clinical characteristics were compared across age tertiles using the Mantel-Haentzel chi-square test for trend. Hazard ratios (HR) and 95% confidence intervals for each clinical outcome were estimated using a Cox proportional hazards regression model, which was stratified by trial. The CABG-to-PCI HR was estimated by Cox models that included treatment assignment, age, and their interactions; age was specified in these models first as a continuous variable, then as a tertile of age. We tested for the significance of the risk modification after adjustment for other baseline characteristics: age, hypertension, diabetes, congestive heart failure, 2-vessel disease, 3-vessel disease, and prior MI; we performed additional adjustments for smoking, unstable angina, and peripheral vascular disease in analyses restricted to studies that provided data on these factors (6). To examine an age threshold at which CABG might be offered preferentially over PCI (or vice versa), we divided the population into 5-year age intervals and calculated the CABG-to-PCI HR and 95% confidence limits within each interval. All statistical analyses were conducted with SAS software version 9.1 (SAS Institute, Cary, North Carolina) and R software version 2.4.0.

The corresponding author had full access to all data and takes final responsibility for the decision to submit the manuscript for publication. The sponsor of the study had no role in the study design, data analysis and interpretation, or manuscript preparation.

Older patients were more likely (p < 0.001) to be female and to have diabetes, hypertension, peripheral vascular disease, congestive heart failure, and 3-vessel disease, but were less likely to be smokers or to have a history of MI (Table 1). The baseline characteristics were well balanced within each tertile of age between patients assigned to CABG and PCI (data not shown).

Mortality was lower among younger patients assigned to PCI than among those assigned to CABG, whereas among the older patients, this pattern was reversed (Figure 82_gr1). The comparative effectiveness of CABG and PCI on mortality was modified significantly by patient age, whether age was specified as a continuous variable (interaction p < 0.002) or as tertiles of age (interaction p < 0.01). The effect of age on the rates of death and of death or MI persisted after adjustment for baseline characteristics and for whether stents were available for use in PCI (Table 2). A similar variation in the effect of CABG and PCI on the composite of death or MI also was evident ((Figure 82_gr2),Table 2). The CABG-to-PCI HR for mortality declined progressively with increasing age when the study cohort was divided into 5-year age groups (Figure 82_gr3), with a CABG-to-PCI HR of <1.0 (more than 59 years of age).

The effect of CABG and PCI in reducing angina, however, did not differ by age. One year after randomization, angina was reported by roughly half as many patients assigned to CABG as assigned to PCI in each tertile of age: 15% versus 28% in the youngest tertile, 13% versus 26% in the middle tertile, and 13% versus 25% in the oldest tertile. In the multivariate analysis, CABG reduced angina overall (p < 0.0001), with no effect of age (p = 0.11) and no evidence of an interaction between treatment and age (p = 0.94).

Although the rate of repeat revascularization procedures was lower in older patients ((Table 2), Figure 82_gr4), the difference in revascularization rates between the CABG and PCI assigned patients was evident in each tertile of age. In a multivariate model, repeat procedures were reduced by CABG (p < 0.0001) and older age (p = 0.002), with no evidence of an interaction between age and treatment assignment (p = 0.24) (Table 2).

The rate of death in the 30 days after randomization was higher among older patients, but was not different between patients assigned to CABG and those assigned to PCI: 0.86% versus 0.53% in the youngest tertile, 1.07% versus 1.31% in the middle tertile, and 1.38% versus 1.38% in the oldest tertile. Data on stroke were available from only 6 trials. The composite outcome of death or stroke within 30 days increased with age and was higher after assignment to CABG: 0.91% versus 1.02% in the youngest group, 1.93% versus 1.53% in the middle tertile, and 2.64% versus 1.56% in the oldest tertile.

In models stratified by diabetes status, the same trends of improved mortality and cardiac events with CABG compared with PCI in older patients were evident (Table 3). The CABG-to-PCI HR for mortality was <1.0 (i.e., favored CABG) at age older than 63 years among patients without diabetes and at age older than 47 years among patients with diabetes (Figure 82_gr3).

Six of the trials in this analysis were conducted using balloon angioplasty during PCI, and 4 trials were conducted using bare-metal stents during PCI (6). Increasing age was associated with a lower CABG-to-PCI HR for mortality in both the balloon-era trials and the stent-era trials (Figure 82_gr5), with no difference in the strength of the interaction between age and treatment according to the era of PCI treatment.

This study demonstrates that older patients have a greater reduction in the relative risk of hard cardiac events (death and the composite of death or MI) after CABG than after PCI. Because older patients generally are at higher risk of cardiac events than younger patients, this finding implies that older patients have a particular reduction in absolute risk, corresponding to a lower number needed to treat to prevent a cardiac event over the subsequent 5 years. In contrast, we found that the rates of angina and repeat coronary revascularization procedures were reduced to a similar extent in all age groups by CABG compared with PCI.

The mechanism by which age alters the comparative effectiveness of CABG and PCI is unclear. It does not seem to be explained by differences in procedural morality, which was similar between CABG and PCI in all age groups. Nor can these results be explained by the higher prevalence of diabetes among older patients, because results were consistent in patients with and without diabetes (Table 3). An alternative hypothesis is that older patients may have more extensive coronary atherosclerosis, and that CABG may have a greater effect on mortality among patients with more extensive disease. We were able to adjust statistically for the presence of 3-vessel disease, and the interaction between age and treatment remained significant after this adjustment. However, we did not have more detailed measures of the extent and severity of coronary atherosclerosis such as the SYNTAX (Synergy Between Percutaneous Intervention With Taxus and Cardiac Surgery) score, and thus may not have adjusted completely for a greater extent of coronary disease among older patients.

The effect of age on the comparative effectiveness of CABG and PCI also was examined in several clinical registries. Although those studies were not based on randomized data, as this study was, registries tend to enroll a broader spectrum of patients and thus provide more representative patient samples. A large clinical registry in Alberta, Canada (3), found that the absolute risk difference between CABG and PCI increased from 1.2% in patients younger than 70 years, to 3.4% in those 70 to 79 years of age, and to 5.8% in those 80 years of age and older. However, this study did not test for an interaction of age with treatment. The Northern New England Registry (4) evaluated 1,693 octogenarians and found that 5-year survival was 68% among CABG patients compared with 61% among PCI patients. However, this study did not compare outcomes among octogenarians with the outcomes among younger patients. The Intermountain Heart Registry found that the overall CABG-to-PCI HR of 0.85 was similar in patients older and younger than 65 years of age (5). In the New York State Registry (6), survival was better among patients 80 years of age and older treated with CABG compared with PCI (HR: 0.74, p = 0.05). In the Kyoto Registry (7) the CABG-to-PCI HR favored CABG more strongly among patients 75 years of age and older (0.73) than among younger patients (0.92), but the test for interaction was not significant (p = 0.61).

The contrast between the effect of age on hard outcomes (death, MI) and on other outcomes (angina, repeat procedures) in this study is striking. At 1 year, CABG reduced angina by roughly half in all age groups, and repeat coronary revascularization was less common after CABG in all age groups (Figure 82_gr4). Interestingly, older patients were significantly less likely to undergo repeat revascularization, perhaps because of the reluctance of older patients or their doctors to choose a second procedure. We did not collect data on the reasons for performing repeat procedures, but the similar rates of angina in follow-up in older and younger patients support the hypothesis that there is a higher threshold for performing repeat procedures among older patients.

Our study is limited by several factors, especially by the relatively small number of patients older than 75 years. Consequently, we cannot evaluate the comparative outcomes of CABG and PCI in the oldest of older patients, particularly among patients 80 years of age and older. Also, we lack detailed information on the extent of coronary disease and on other key clinical factors, such as renal function, anemia, and frailty, that are correlated with patient age and that may affect clinical outcomes.

Our study did not include more recent trials that used drug-eluting stents during PCI, such as the SYNTAX and CARDIA (Coronary Artery Risk Development in Young Adults) studies. These trials were not completed when this collaborative pooling study was designed, and they still have not reported results of 5-year follow-up. Although inclusion of more contemporary trials would be desirable, we do not believe that treatment advances have eliminated the adverse effect of age on hard cardiac outcomes. In our data, the effect of age on the comparative effectiveness of CABG and PCI was not affected by the availability of coronary stents (Figure 82_gr5). Meta-analyses of trials comparing balloon angioplasty with bare-metal stents and comparing bare-metal stents with drug-eluting stents have shown no differences in mortality (8). Finally, recent data from the National Cardiovascular Data Registry in the drug-eluting stent era (2004 to 2007) show that older age still has a highly significant association with greater mortality after PCI (9), and data from the Society of Thoracic Surgeons show a strong adverse effect of age on subsequent mortality in a cohort of contemporary CABG patients (10). These data suggest that older age remains a powerful predictor of increased adverse cardiac outcomes after CABG and PCI. Nevertheless, additional studies of the effect of age on the comparative effectiveness of CABG and PCI are warranted to confirm our findings among more contemporary patients.

In this pooled analysis, older patients had lower rates of death and of death or MI than did younger patients after CABG compared with PCI. The effect of age on the choice of treatment has not yet been addressed in coronary revascularization guidelines. Long-term follow-up from more recent trials, including patients with drug-eluting stents, complex multivessel disease, and diabetes, will add to this analysis.