Multicenter experience in revascularization of very elderly patients

Multicenter experience in revascularization of very elderly patients Eric D. Peterson, MD, MPH,a Karen P. Alexander, MD,a David J. Malenka, MD,b Edward L. Hannan, PhD,c Gerald T. O’Conner, PhD, DSc,b Ben D. McCallister, MD,d William S. Weintraub, MD,e and Fred L. Grover, MD,f for the American Heart Association Chronic CAD Working Group Durham, NC, Hanover, NH, Albany, NY, Kansas City, Mo, Atlanta, Ga, and Denver, Colo

Background Very elderly patients are increasingly referred for revascularization yet have been underrepresented in both prior percutaneous coronary intervention (PCI) and coronary bypass surgery (CABG) clinical trials. We pooled the largest PCI and CABG clinical registries in the United States to better understand revascularization procedure use, risks and outcomes in patients aged ⱖ75 years. Methods Six PCI registries (n ⫽ 48,439) and 8 CABG registries (n ⫽ 180,709) voluntarily contributed all procedural data in patients aged ⱖ75 years from 1990 through 1999. Patient characteristics, procedural process, and inhospital mortality and morbidity outcomes were evaluated. Risk factors for mortality in elderly patients were identified and compared across registries using standardized multivariable logistic regression. Results Between the years 1991 and 1999, the proportion of patients aged ⱖ75 years undergoing revascularization was on the rise (10% increase). Pooled estimates of inhospital mortality following PCI during this decade was 3.0% (range 1.5%–5.2% among databases), and following CABG was 5.9% (range 4.9%– 8.4% among databases). Mortality rates declined significantly in older patients for both PCI and CABG over this decade. While process measures varied across registries, the most significant predictors of inhospital death (procedural urgency, left ventricular dysfunction, prior CABG) seemed consistent across all sites. Conclusion

Over the last decade, the use of coronary revascularization in elderly patients increased and outcomes improved. While age remains a determinant of procedural risk, this risk varies markedly among elderly patients, emphasizing the need for individualized risk assessments. (Am Heart J 2004;148:486 –92.)

The elderly represent the fastest growing segment of the United States population. Currently, those aged ⱖ75 years account for approximately 5% of the US population, but by the middle of this century this number will triple.1,2 Since the prevalence of cardiovascular disease also increases with age, this demographic change is leading to a dramatic increase in the number of very elderly patients being considered for coronary revascularization. To date, however, there has been a paucity of clinical information available to From the aThe Outcomes Research and Assessment Group, The Duke Clinical Research Institute, Durham, NC, bNorthern New England Cardiovascular Disease Study Group, Hanover, NH, cUniversity of Albany, Albany, NY, dMid America Heart Institute, Kansas City, Mo, eEmory University, Atlanta, Ga and fUniversity of Colorado HSC, Denver, Colo. Dr. Peterson is a Paul Beeson Faculty Scholar, and Dr. Alexander is the recipient of a Doris Duke Clinical Scientist Development Award. Submitted September 2, 2003; accepted March 26, 2004. Reprint requests: Eric D. Peterson MD, MPH, FACC, Associate Professor of Medicine, Duke Clinical Research Institute, Box 17969, Durham, NC 27715. E-mail: [email protected] 0002-8703/$ - see front matter © 2004, Elsevier Inc. All rights reserved. doi:10.1016/j.ahj.2004.03.039

inform this clinical decision. Clinical trials of revascularization have routinely under-enrolled elderly subjects compared with their prevalence in the treated population. Similarly, observational case series of percutaneous coronary intervention (PCI) and coronary bypass surgery (CABG) in very elderly patients have generally been limited to single-institution experiences with small numbers of patients and are often not reflective of contemporary procedural techniques.3,4 To further our understanding of cardiovascular care in very elderly patients, the American Heart Association supported an effort to combine databases to answer questions in elderly patients with coronary artery disease. This included the voluntarily sharing of outcomes from 6 of the nation’s largest PCI databases and 8 of the nation’s largest CABG databases.5 Specific aims for this collaboration were 1) to create pooled estimates for inhospital morbidity and mortality following revascularization in elderly patients (aged ⱖ75 years); 2) to describe and compare the clinical characteristics and processes of care for the elderly across centers; and 3) to explore the feasibility of pooling

American Heart Journal Volume 148, Number 3

Peterson et al 487

Table I. Participating databases with patients aged ⱖ 75 years undergoing PCI or CABG Database (abbreviation) Society of Thoracic Surgery (STS) National Cardiovascular Network (NCN) Northern New England (NNE) New York State (NYS) Mid-America Heart Institute (MAH) Emory University (EMU) Cooperative Cardiovascular Project (CCP) Veterans Administration (VA) Total

Centers

Years

PCI Pts Age > 75

CABG Pts Age > 75

438 20 7 33 1 1 20 42

1995–98 1994–97 1994–99 1993–98 1993–98 1994–98 1992–93 1990–99 1990–99

N/A 16,611 4357 23,228 1715 1292 1236 N/A 48,439

132,301 14,233 3707 23,072 709 711 1127 4849 180,709

N/A, ⫽ Not Available

such estimates by comparing outcomes predictors across centers. Finally, temporal changes in outcomes were tracked.

Figure 1

Methods Participating databases The 6 PCI and 8 CABG databases participating in this effort ranged in size from 709 to 132,301 patients (Table 1). More complete information can be found on each of the databases via the following references. The Veterans Administration and Society of Thoracic Surgery are long-standing, cardiothoracic surgery databases for their respective organizations.6,7 Two databases were from single-institution registries at MidAmerica Heart Institute8 and Emory University.9,10 Northern New England11,12 and New York State13,14 represent PCI and CABG regional databases. The Cooperative Cardiovascular Project Database is a multicenter CABG and PCI registry sponsored by Medicare.15 The National Cardiovascular Network16,17 is a voluntary multicenter PCI and CABG database representing large academic and private medical centers across the United States.

Data inclusion While the years of data available varied slightly across these clinical registries, all procedures were performed between 1991 and 1999, with a focus on 1994 –1998. For CABG, our analysis was limited to those receiving isolated CABG, and excluded those receiving combined procedures with valve repair or replacement. While multiple age cutpoints can be used to define “very elderly,” our analysis concentrated on those aged ⱖ75 years (Figure 1). Each of the databases was analyzed separately for the outcomes of interest. Variable definitions of inhospital events and procedural characteristics were representative of the definitions used historically within each registry. Participants were provided with standardized variable definitions for baseline characteristics, process measures, and outcomes which match national standard definitions. Variables within the data sets were mapped as closely as possible to standard definitions. The time horizon for this analysis was limited to inhospital events. For both PCI and CABG we assessed inhospital mortality, stroke (defined as loss of neurologic function that

Increasing use of PCI and CABG in the elderly (aged ⱖ75) over the last decade.

persisted to hospital discharge), renal failure (dialysis-dependent), and myocardial infarction (MI) (defined as new Q-waves along with other supporting evidence, either markers or wall-motion abnormality). For PCI, use of emergency CABG within 12 hours of PCI was determined. For CABG, prolonged ventilatory support (defined as time to extubation ⬎24 hours), and reoperation during hospital stay was determined.

Statistical analysis Estimates. Descriptive statistics summarizing baseline characteristics and procedural variables are presented as percentages for discrete variables, and medians and interquartile ranges for continuous variables. Baseline characteristics are shown as the average and range in values from high to low across databases. Likewise, inhospital outcomes are presented as averages and ranges across databases. We used meta-analysis software (FastPro version 1.81, Academic Press, San Diego, Calif) to combine estimates of inhospital mortality, stroke (serious permanent neurologic event), renal failure (requiring dialysis), periprocedural Q-wave infarction, and emergent CABG. We used random effects (empirical Bayes) modeling methods for the outcome estimates. This model

American Heart Journal September 2004

488 Peterson et al

Table II. PCI and CABG patient baseline demographics PCI Average Mean Age (y) Male* (%) Nonwhite (%) Diabetes (%) HTN (%) COPD (%) Prior CHF (%) Prior CVA/TIA (%) Prior MI (%) Prior PCI (%) Prior CABG (%) Mean LVEF (%) No. diseased vessels 1 (%) 2 (%) 3 (%) Procedure priority (%) Urgent/emergent

CABG Range

Average

Range

79 53 6 22 62 11 14 10 41 36 23 47

79–80 48–60 4–7 21–25 55–70 6–12 11–18 6–14 24–50 25–52 16–31 40–53

78 60 5 26 69 16 20 14 48 N/A 9 49

77–79 58–61 1–7 22–29 58–81 10–17 11–27 9–27 34–60 N/A 6–16 45–52

41 30 29 33

22–56 27–33 17–48 8–60

6 24 69 36

3–9 20–35 56–77 11–71

Range of values from low to high and average of the values across databases. CABG, coronary artery bypass graft; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; HTN, hypertension; LVEF, left ventricular ejection fraction; MI, myocardial infarction; N/A, not available; PCI, percutaneous coronary intervention; TIA, transient ischemic attack. *Excludes the VA, which was 98% male.

accounts for variability in the size of the data set in determining the composite estimate of events, which allows a stronger emphasis to be placed on the larger samples. Trends The proportion of patients in each revascularization database who were elderly (age ⱖ75) is reported by year (1991–1999). Inhospital mortality among elderly patients is also reported by year (1991–1998) across each individual database, and summary mortality for PCI and CABG is shown across time. Modeling Agreement in the major predictors of inhospital mortality for PCI and CABG in very elderly patients should be consistent across databases. To test this hypothesis, we determined univariate predictive power of baseline risk factors for PCI and CABG mortality within each database. Finding a high degree of similarity across databases in the predictive variables, we then created a standardized model using prespecified selected variables (11 for PCI, 12 for CABG). These variables included shock, emergency procedure, MI within 24 hours, ejection fraction, dialysis-dependent renal failure, prior PCI or CABG, age, diabetes, multivessel disease, history of congestive heart failure, peripheral vascular disease, and sex. The four largest PCI and CABG databases were used to apply and compare the contribution of these variables as predictors of inhospital mortality in elderly patients. The models were run in each database and total ␹2 and C-indices were determined. The contribution of each variable to the overall predictive power of the model in that database was determined (ie, the percentage of total ␹2 that each variable represented in that database). For comparison, the variables contribution to the overall prediction in each database is shown.

Results Table 1 describes the 8 revascularization databases and details the years of data included and the total numbers of patients aged ⱖ75 years undergoing revascularization from each database. Our overall population included 48,439 PCI and 180,709 CABG patients who had their procedures performed between 1990 and 1999. Figure 1 illustrates the gradual increase in the percentage of total PCI and CABG cases performed in those aged ⱖ75 years. During the last decade, there was a 7% rise in the proportion of patients undergoing CABG who are aged ⬎75, and a 10% rise in elderly patients undergoing PCI. Currently, elderly patients (aged ⱖ75 years) comprise over 20% of revascularization populations. Table 2 compares the baseline characteristics of PCI and CABG patients. Across databases, there was a general agreement in most baseline characteristics by procedure type. Baseline characteristics of PCI patients show them to be slightly older and more often female than CABG patients. PCI patients also had fewer cardiac risk factors, less comorbid illness, and more single-vessel disease than CABG patients. In contrast, CABG patients were less likely to have had a prior CABG than those undergoing PCI (9% vs 23%). Although CABG patients had more 3-vessel disease than PCI patients overall, there was variability across registries. Procedure priority definitions also varied across

American Heart Journal Volume 148, Number 3

Figure 2

Peterson et al 489

Table III. PCI outcomes in patients aged ⱖ 75 years across registries

Inhospital mortality Stroke Q-wave MI Renal failure/dialysis Emergency CABG

Inhospital mortality following PCI or CABG as a function of patient age

registries. For example, urgent procedures were defined in one registry as “prior to patient discharge” and another as “not elective, not emergent,” allowing a degree of interpretation in this middle zone. One registry did not have an “emergent” category and another only recently began collecting procedure priority. We show this variability to be instructive in the importance of understanding variable coding when combining across registries. There was also variability in process measures among elderly patients for PCI and CABG. For PCI, the percentage of “primary” PCI cases for acute MI (within 24 hours of presentation for acute MI) ranged from 5% to 28%. Similarly, the percentage of PCI cases performed on multiple vessels ranged from 7% to 33%, while the use of stents ranged from 35% to 58% among the databases. For CABG, use of preoperative intra-aortic balloon pump ranged from a low of 1% to 8% and use of internal mammary artery grafts ranged from 52% to 80%.

PCI Procedural Outcomes Procedural mortality rates rose in a consistent curvilinear fashion as a function of patient age across databases for both PCI and CABG (Figure 2). In aggregate, patients aged ⱖ75 years undergoing PCI had inhospital mortality rates ranging from 1.5% to 5.2% across databases, with an average of 3.0% (Table 3). The highest mortality was found in the Cooperative Cardiovascular Project registry, which was reflective of early experience (1992–1993) and a high percentage of primary PCI (23%). To clarify the impact of case mix on mortality estimates and the potential role of acute PCI, we excluded primary PCI cases from the databases and reanalyzed mortality outcomes. While risks are certainly higher in acute PCI, the composite estimates of

Databases contributing

Range across registries

Composite estimate (random effects)

6 5 5 3 5

1.5%–5.2% 0.1%–0.9% 0.5%–2.3% 0.2%–2.6% 0.8%–1.5%

3.0% (2.2–4.1) 0.4% (0.3–0.6) 0.9% (0.5–1.5) 0.8% (0.2–2.0) 1.2% (0.9–1.4)

Figure 3

Inhospital mortality following PCI or CABG in the elderly (aged ⱖ75) over the last decade. Data for figure includes National Cardiovascular Network, New York State, Society of Thoracic Surgery, and Emory. Trends significant for PCI and CABG (P ⬍ .01).

risk varied little. For example, the aggregate mortality in the National Cardiovascular Network PCI population declined only slightly from 2.96% to 2.43% after excluding primary PCI. PCI mortality has been declining steadily over the last decade, from 2.7% in 1991 to 1.15% in 1997 (P ⬍ .01) (Figure 3). PCI complications were infrequent in this very elderly patient population across databases and were uniformly below 2% for stroke, Q-wave infarction, renal failure, and emergency CABG (Table 3). The most commonly reported nonfatal serious post-PCI complication was the need for emergent bypass surgery, which occurred in about 1.2% of cases. We were not able to determine the incidence of bleeding or local arterial complications in this data set.

CABG procedural outcomes In aggregate, patients aged ⱖ75 years undergoing CABG had inhospital mortality rates ranging from 4.9% to 8.4% across databases, with an average estimate of all databases across all years of 5.9% (Table 4). CABG mortality also declined over the last decade in very elderly patients, from 5.9% in 1991 to 4.98% in 1998

American Heart Journal September 2004

490 Peterson et al

Table IV. CABG outcomes in patients aged ⱖ 75 years across registries Range Databases across contributing registries Inhospital mortality Stroke Renal failure/dialysis Q-wave MI Reoperation Prolonged vent support*

8 8 6 6 7 6

Composite estimate (random effects)

4.9%–8.4% 5.9% (5.3–6.6) 1.7%–5.6% 3.5% (2.7–4.5) 1.1%–6.7% 3.2% (1.6–5.4) 1.1%–8.0% 1.5% (1.0–2.2) 0.2%–6.8% 2.0% (0.7–3.9) 8.6%–21.6% 11.2% (8.3–14.7)

*Ventilatory support more than 24 hours following surgery.

(P ⬍ .001) (Figure 3). CABG complications were below 5% for stroke, renal failure, reoperation, and Qwave MI, and were highest for prolonged ventilatory support longer than 24 hours (11%).

Predictors of inhospital PCI and CABG mortality The 11 PCI variables could be grouped in 3 tiers regarding their general contribution to the model’s ␹2 across databases. While the specific rank varied slightly (1 versus 2), the top 3, second 3, and remaining variables were consistent across databases. For example, PCI variables describing procedural acuity (shock, emergent, and PCI ⬍24 hours from MI) were always ranked in the top 3 variables across databases. These were followed by a second tier of variables related to myocardial dysfunction (history of congestive heart failure and ejection fraction) or renal failure. The remaining variables, including age and other comorbid illnesses, were less powerful predictors of acute mortality risks among this elderly population. The discriminatory ability of analyses based on this limited set of core PCI predictors was high among databases (C-indices, or area under the receiver operator curve, ranging from 0.78 to 0.87). The 12 core CABG variables also tended to be grouped in several tiers based on their general contribution to the model’s ␹2 across databases. The top 3 variables were prior CABG (number 1 across the databases), emergency surgery, and renal failure. These were followed closely by a second tier of variables including shock, congestive heart failure, and age. The remaining variables were less powerful predictors of acute mortality risks among this elderly population. The discrimination ability of the CABG mortality models tended to be lower than that seen for PCI, ranging from a C-index of 0.69 to 0.77.

Discussion This collaborative effort represents the largest aggregation of clinical information on process and outcomes

of coronary revascularization in very aged patients. Studying nearly 230,000 patients aged ⱖ75 years undergoing PCI and CABG, this project provides unique insights into cardiovascular care for very elderly patients across a decade of data. There are several important findings. First, patients aged ⱖ75 years constitute a substantial and increasing percentage of patients undergoing coronary revascularization in this country (currently ⬎20%). Second, while procedural complications increase with advancing age, current estimates of inhospital mortality rates following PCI and CABG in those aged ⱖ75 years are modest (ⱕ5%). In addition, procedural risks in an elderly patient can vary markedly from this “average” depending on other factors such as procedural acuity, left ventricular function, and renal function. Finally, there are promising trends in declining revascularization risks for both PCI and CABG compared with prior decades, and seen even across the years reported in the current study.17,18 Previous studies of outcomes of PCI and CABG in very elderly patients have generally been small in sample size and often from single institutions.19 –22 Thus the generalizability of these findings for routine community-based practice remains unclear. Our study establishes more stable estimates of contemporary risk for PCI and CABG due to its sample size and breadth of US centers involved. Although we found some expected variation in variable definitions and processes across databases, we were able to find consistent results among centers. The risk of inhospital mortality associated with revascularization increases by ⬃1% per decade of life for PCI and ⬃2% per decade of life for CABG (trends seen in Figure 2). Between the ages of 75 and 85 years, this gradual increase in risks is not prohibitive for considering these procedures as viable treatment options. This rate then exponentially increases beyond the age of 85, although even in this large data set, numbers are limited beyond age 85 (Figure 2). Consistent with prior studies, the relative mortality risk of PCI remains 50% lower than with CABG (3% vs 5.9%). Similarly, the risks of nonfatal complications are lower following PCI relative to CABG (stroke risk 0.4% for PCI vs 3.5% for CABG). These comparisons are not at all surprising as the populations and procedures themselves significantly differ, yet they establish a frame of reference for the important choices between revascularization strategies. In addition, our study provides confirmation of the major risk predictors for short-term outcomes following PCI and CABG in very elderly patients. For PCI, the most powerful predictors of inhospital mortality were disease acuity variables (emergent case, shock, or recent MI), which are not pertinent in elective decisions for treatment. For CABG, prior surgery remained the best single predictor of inhospital mortality. Acuity

American Heart Journal Volume 148, Number 3

variables and left ventricular dysfunction were close behind in predictive power. Advancing age was ranked 6th for CABG mortality and 7th for PCI mortality in this already elderly population. Combined risk estimates using all 12 patient factors allowed for accurate discrimination of those with low and high likelihood for death from either procedure (with area under the receiver operator curve of 0.71 to 0.86). A prior analysis from the National Cardiovascular Network database revealed that octogenarians undergoing CABG surgery without any of the clinical predictors of worse outcomes (similar to our 12 CABG variables) had half the inhospital mortality as the general group of octogenarians (mortality in octogenarians without comorbidity ⫽ 4.2% vs 8.1% in octogenarians overall).17 Thus, the clinical stability and function of the patient are as or more important than age, even in an elderly population, which emphasizes the need to personalize risk estimates for each patient. There were notable differences in processes across centers that may impact outcomes. For example, the use of stenting in elderly patients varied from 35% to 58% among leading centers, despite their proven benefits in young and old patients.23–25 This may have been a consequence of the rate at which institutions adopted this practice during the transition years of 1994 –1996. Similarly, there was a high variability in the use of intra-aortic balloon pumps prior to CABG in elderly patients, which may reflect controversies in the practice standards for this intervention.26 Finally, despite prior studies demonstrating long-term survival benefits gained from using internal mammary artery grafts in elderly patients,27 use varied greatly from a low of 50% to a high of 80%. These findings suggest that outcomes in very elderly patients could continue to improve through the identification and standardization of optimum care practices for revascularization. This is underscored by the improvements in outcomes already seen throughout the 1990s for both PCI and CABG in elderly patients. Short-term risks of PCI and CABG in very elderly patients must be placed in the context of longer term benefits offered by these procedures. While randomized studies of PCI and CABG have generally not included very elderly patients,4 observational studies have found that both PCI and CABG appear to provide for longer adjusted survival rates relative to medical therapy in those with multivessel coronary disease.28 Even after adjusting for likelihood of receiving revascularization and other clinical variables, a greater absolute risk reduction at 4 years is seen following revascularization compared with medical therapy in elderly patients (age ⱖ80 years) than in younger patients.28 In addition, other studies have demonstrated the effectiveness of both PCI and CABG in improving quality of life in elderly patients with better relief of angina and

Peterson et al 491

improved functional status.29 –31 Therefore, long-term benefits from revascularization in many elderly patients are realized and must be included in treatment decisions.

Limitations While variability in data definitions across sites is a potential concern, baseline characteristics were remarkably similar, and concordance in outcomes across databases was likewise reassuring. Procedural priority and PCI for AMI are the more challenging variables to justify across databases. The variability in coding of procedural priority was accounted for with other similar variables such as acute MI and shock across centers. Our outcomes were limited to inhospital events, so longer term survival and functioning for our population are not available. Finally, as coronary revascularization technology continues to evolve rapidly, the use of adjuvant pharmacologic therapies including intravenous and oral antiplatelet agents and improved anticoagulants are reducing acute post-PCI risks. Unfortunately, we do not have the use of 2b3a inhibitors as adjunctive therapy in PCI treatment in our analysis, but this became standard therapy toward the later years of our study. Additionally, early studies have demonstrated that off-pump surgery may reduce risks of both fatal and nonfatal events following CABG among older patients.32,33 To continue from this analysis, further data will be needed to track the improvements in the risk of revascularization associated with advanced age in the future with the advent of better technology.

Conclusions PCI and CABG in very elderly patients (aged ⱖ75 years) are well tolerated with an average inhospital mortality of 3% and 5.9%, respectively. While age is still a major risk factor for adverse outcomes, the increase in risk is gradual until after the age of 85. Importantly, as more elderly patients are undergoing procedures, the risk of death during the index admission is declining over time. Continued monitoring of outcomes across large databases will help track further improvements with advancing technology and longer follow-up. More than age, process and acuity variables are powerful predictors of risk. These and other risk factors should be incorporated into personalized risk estimates for very elderly patients. Ultimately, treatment decisions must incorporate the individual’s assessment of their risks and benefits. With special thanks for the statistical support of Michael M. Coen, MA, MBA, Phillip G. Jones, MS, Laura P. Coombs, PhD, Kevin J. Anstrom, PhD, Emir Veledar, PhD, Winthrop D. Piper, MS, Elaine M. Olm-

492 Peterson et al

stead, and Laurie A. Shroyer, PhD, MSHA, and for the editorial assistance of Cindy M. Olson.

References 1. US Bureau of the Census. Sixty-five plus in America, P23-190: population projections of the United States, by age, sex, race and Hispanic origin 1993–2050. Washington, DC: US Government Printing Office; 1996. p. 25–1104. 2. US Bureau of the Census. Life tables of the United States. Washington, DC: US Government Printing Office; 1986. 3. Gurwitz JH, Col NF, Avorn J. The exclusion of the elderly and women from clinical trials in acute myocardial infarction. JAMA 1992;268:1417–22. 4. Lee PY, Alexander KP, Hammill BG, et al. The elderly and women: are they adequately represented in randomized trials of acute coronary syndromes? JAMA 2001;286:708 –13. 5. Cheitlin MD, Gerstenblith G, Hazzard WR, et al. AHA Conference Proceedings. Database Conference: January 27-30, 2000, Washington, DC: Do existing databases answer clinical questions about geriatric cardiovascular disease and stroke? Am J Geriatr Cardiol 2001;10:207–23. 6. Grover FL, Shroyer AL, Hammermeister KE. Calculating risk and outcome: the Veterans Affairs database. Ann Thorac Surg 1996; 62:S6 –11. 7. Ferguson TB Jr, Hammill BG, Peterson ED, et al. A decade of change—risk profiles and outcomes for isolated coronary artery bypass grafting procedures, 1990 –1999: a report from the STS National Database Committee and the Duke Clinical Research Institute. Ann Thorac Surg 2002;73:480 –9. 8. O’Keefe JH Jr, Kreamer TR, Jones PG, et al. Isolated left anterior descending coronary artery disease: percutaneous transluminal coronary angioplasty versus stenting versus left internal mammary artery bypass grafting. Circulation 1999;100:II114 – 8. 9. Taddei CF, Weintraub WS, Douglas JS Jr, et al. Influence of age on outcome after percutaneous transluminal coronary angioplasty. Am J Cardiol 1999;84:245–51. 10. Klein LW, Block P, Brindis RG, et al. Percutaneous coronary interventions in octogenarians in the American College of Cardiology– National Cardiovascular Data Registry: development of a nomogram predictive of in-hospital mortality. J Am Coll Cardiol 2002; 40:394 – 402. 11. O’Connor GT, Malenka DJ, Quinton H, et al. Multivariate prediction of in-hospital mortality after percutaneous coronary interventions in 1994 –1996: Northern New England Cardiovascular Disease Study Group. J Am Coll Cardiol 1999;34:681–91. 12. Wennberg DE, Malenka DJ, Sengupta A, et al. Percutaneous transluminal coronary angioplasty in the elderly: epidemiology, clinical risk factors, and in-hospital outcomes. Am Heart J 1999; 137:639 – 45. 13. Hannan EL, Racz MJ, Arani DT, et al. Short- and long-term mortality for patients undergoing primary angioplasty for acute myocardial infarction. J Am Coll Cardiol 2000;36:1194 –1201. 14. Hannan EL, Racz MJ, Arani DT, et al. A comparison of short- and long-term outcomes for balloon angioplasty and coronary stent placement. J Am Coll Cardiol 2000;36:395– 403. 15. Peterson ED, DeLong ER, Muhlbaier LH, et al. Challenges in comparing risk-adjusted bypass surgery mortality results: results from the Cooperative Cardiovascular Project. J Am Coll Cardiol 2000; 36:2174 – 84.

American Heart Journal September 2004

16. Batchelor WB, Anstrom KJ, Muhlbaier LH, et al. Contemporary outcome trends in the elderly undergoing percutaneous coronary interventions: results in 7,472 octogenarians. J Am Coll Cardiol 2000;36:723–30. 17. Alexander KP, Anstrom KA, Muhlbaier LH, et al. Outcomes of cardiac surgery in the very elderly: results from the National Cardiovascular Network. J Am Coll Cardiol 1999;35:731– 8. 18. Alexander KP, Peterson ED. Coronary artery bypass grafting in the elderly. Am Heart J 1997;134:856 – 64. 19. Dynina O, Vakili BA, Slater JN, et al. In-hospital outcomes of contemporary percutaneous coronary interventions in the very elderly. Cathet Cardiovasc Interven 2003;58:351–7. 20. Minai K, Horie H, Takahashi M, et al. Long-term outcome of primary percutaneous transluminal coronary angioplasty for low-risk acute myocardial infarction in patients older than 80 years: a single-center, open, randomized trial. Am Heart J 2002;143:497– 505. 21. Williams DB, Carrillo RG, Traad EA, et al. Determinants of operative mortality in octogenarians undergoing coronary bypass. Ann Thorac Surg 1995;60:1038 – 43. 22. Hannan EL, Burke J. Effect of age on mortality in coronary artery bypass surgery in New York, 1991–1992. Am Heart J 1994;128: 1184 –91. 23. Thompson RC, Holmes DR, Grill DE, et al. Changing outcome of angioplasty in the elderly. J Am Coll Cardiol 1996;27:8 –24. 24. Tan KH, Sulke N, Taub N, et al. Percutaneous transluminal coronary angioplasty in patients 70 years of age or older: 12 years’ experience. Br Heart J 1995;74:310 –7. 25. Wong SC, Papdokos S, Rosenberg C, et al. In-hospital clinical outcomes post-angioplasty in the octogenarians: a report from the New York State angioplasty registry. Circulation 1999;100:I-455. 26. Holman WL, Li Q, Kiefe CI, et al. Prophylactic value of preincision intra-aortic balloon pump: analysis of a statewide experience. J Thorac Cardiovasc Surg 2000;120:1112–9. 27. Ferguson TB Jr, Coombs LP, Peterson ED. Internal thoracic artery grafting in the elderly patient undergoing coronary artery bypass grafting: room for process improvement? J Thorac Cardiovasc Surg 2002;123:869 – 80. 28. Graham MM, Ghali WA, Faris PD, et al. Survival after coronary revascularization in the elderly (APPROACH). Circulation 2002; 105:2378 – 84. 29. The TIME Investigators. Trial of invasive versus medical therapy in elderly patients with chronic symptomatic coronary-artery disease (TIME): a randomized trial. Lancet 2001;358:951–7. 30. Conaway DG, House J, Bandt K, et al. Elderly patients derive similar health status benefits from coronary artery bypass surgery (CABG) as compared to younger patients despite a slower rate of physical recovery [abstract]. Circulation 2002;106:II-604. 31. Seto TB, Taira DA, Berezin R, et al. percutaneous coronary revascularization in elderly patients: impact on functional status and quality of life. Ann Intern Med 2000;132:955– 8. 32. Hannan EL, Racz MJ, McCallister BD, et al. A comparison of three-year survival after coronary artery bypass graft surgery and percutaneous transluminal coronary angioplasty. J Am Coll Cardiol 1999;33:63–72. 33. Hirose H, Amano A, Takahashi A. Off-pump coronary artery bypass grafting for elderly patients. Ann Thorac Surg 2001;72: 2013–9.