Myocardial Infarction and Cardiac Mortality in the Bypass Angioplasty Revascularization Investigation (BARI) Randomized Trial
Background Cardiac mortality and myocardial infarction (MI) rates are used to evaluate the efficacy of coronary artery bypass grafting (CABG) and percutaneous transluminal coronary angioplasty (PTCA). We compared 5-year cardiac mortality and MI rates in 1829 patients with multivessel disease randomized to CABG or PTCA.
Methods and Results The 5-year cardiac mortality rate was 8.0% in patients assigned to PTCA compared with 4.9% in those assigned to CABG (relative risk [RR] of 1.55 with a 95% confidence interval [CI] of 1.07 to 2.23; P=.022). In a subgroup of 1476 nondiabetic patients, there were no significant differences between treatment groups in cardiac mortality either overall (4.6% versus 4.2%; RR=1.04, 95% CI, 0.65 to 1.66; P=.908) or in subgroups based on symptoms, left ventricular function, number of diseased vessels, or stenotic proximal left anterior descending artery. The two treatment groups had similar event rates for the combined end point of cardiac death or MI. The RR for cardiac mortality in 264 patients who sustained an MI compared with those who did not was 5.9 (P<.001). MIs were more common after CABG during index hospitalization (P=.004), but in the PTCA group, they were more common after discharge (P<.001).
Conclusions The Bypass Angioplasty Revascularization Investigation (BARI) trial indicates 5-year cardiac mortality in patients with multivessel disease was significantly greater after initial treatment with PTCA than with CABG. The difference was manifest in diabetic patients on drug therapy. There were no significant differences overall for the composite end point of cardiac mortality or MI between treatment groups or for cardiac mortality in nondiabetic patients regardless of symptoms, left ventricular function, number of diseased vessels, or stenotic proximal left anterior descending artery.
The NHLBI-sponsored BARI multicenter trial was designed to test the hypothesis that the long-term clinical outcome of patients with multivessel coronary disease suitable for treatment with either PTCA or CABG is not compromised when PTCA is chosen as the initial treatment strategy.1 2 3 4 5 6 Cumulative total mortality, the primary end point in BARI, was similar for both procedures after an average of 5.4 years of follow-up.6 Total mortality was significantly higher in a subgroup of treated diabetic patients assigned to PTCA.7 Cardiac mortality and MI (Q-wave and non-Q wave) were important secondary end points in BARI that have not been reported previously.
In prior randomized trials comparing PTCA with CABG,8 9 10 11 12 13 the impact of the type of coronary revascularization procedure on the incidence of major cardiac events such as cardiac mortality or MI has been based on <3 years of follow-up and inadequate standardized criteria for MI. In this report, we compare the impact of the initial treatment strategy (PTCA versus CABG) on the 5-year rate of cardiac events in the total BARI randomized population and in patient subgroups by use of centrally classified cardiac events and validated criteria for MI.
A detailed description of the BARI study design, protocol, and clinical characteristics of the patients at entry has been published previously.1 2 3 4 5 6 Briefly, patients who had multivessel coronary artery disease and severe angina or ischemia were randomly assigned to an initial treatment strategy of CABG or PTCA. The primary clinical indications for coronary revascularization included the following: (1) non–Q-wave MI or unstable angina stabilized for 4 hours to 6 weeks; (2) stable class III or IV angina pectoris; (3) stable class I or II angina with either severe ischemia on noninvasive testing, Q-wave MI stabilized for 24 hours to 30 days, or resting ejection fraction <50%; or (4) no angina during the 6 weeks before entry but objective evidence of severe ischemia on noninvasive testing and either prior Q-wave MI or history of prior angina. Patients were considered ineligible if any of the following criteria were present: single-vessel disease, age <17 or ≥80 years, prior coronary revascularization procedure, left main coronary stenosis ≥50%, or noncardiac illness expected to limit survival. From August 1988 through August 1991, 1829 patients consented to the randomization process; 915 were assigned to PTCA and 914 to CABG. There were no significant differences in baseline characteristics between the two treatment groups (Table 1⇓). The protocol was approved by the institutional review boards at the participating institutions.
MI Ascertainment and Definition
The BARI protocol required that each patient have a 12-lead ECG at entry, before the procedure, 24 to 96 hours after the procedure, 4 to 14 weeks after randomization, and annually thereafter. Additional ECGs were required in patients who underwent subsequent coronary revascularization procedures (before and after the procedure) and in cases of suspected ischemic events.
All ECGs were interpreted at the Saint Louis University Central ECG and Myocardial Infarction Classification Laboratory. With the use of the Minnesota code criteria,14 each ECG was coded independently by trained central laboratory staff blinded to the patient’s clinical history and treatment assignment. Serial comparison of sequential tracings was performed with the use of a modified Novacode system to identify patients with new ECG changes in the Minnesota Code.15 16 The modified Novacode adjusts for nonsignificant Minnesota Code Q-wave changes that result from minimal biological or technical procedural variations in the QRS waveform.
Ischemic cardiac events related to the index hospitalization for coronary revascularization were examined. The index hospitalization was defined as the hospitalization associated with the initial revascularization procedure. Postprocedure MI was defined by the presence of a new, two-grade (Minnesota Code) worsening in the Q-wave on the postprocedure versus the preprocedure ECG.16 17 According to protocol, cardiac enzymes were not used to define MI within 96 hours of a revascularization procedure because of the recognized difficulties in comparing the clinical significance of cardiac enzyme elevations after CABG compared with after PTCA. Symptomatic ischemic cardiac events requiring hospitalization were classified as Q-wave MI, non–Q-wave MI, or neither. Symptomatic Q-wave MI was defined as a hospitalization with ischemic symptoms and a new, two-grade (Minnesota Code) worsening in the Q wave or new left bundle-branch block plus abnormal cardiac enzymes. Non–Q-wave MI was defined as a hospitalization with abnormal CK/CK-MB enzymes associated with either chest pain lasting >20 minutes or the appearance of new ECG changes.16 Silent Q-wave MI was defined as the new appearance of a two-grade (Minnesota Code) worsening in the Q wave on a regularly scheduled follow-up ECG.
As of June 5, 1995, the vital status was known for 1792 patients (98%). The mean follow-up interval was 5.4 years (range, 3.8 to 6.8 years). Of the 13 418 ECGs required by protocol, 11 705 (87.2%) were available for review. A baseline tracing was available for 99.8% of patients. Of the 1981 coronary revascularization procedures, 96% of the postprocedure ECGs were available for review. Of the 9608 follow-up ECGs required by protocol, 83% were available for analysis: 94.5% at 4 to 14 weeks, 92.3% at year 1, and 72.4%, 86.9%, 69.1%, and 80.1% at years 2, 3, 4, and 5, respectively. A clinic visit was not required at years 2 and 4. In 94.2% of the 1023 suspected ischemic events classified at the central ECG laboratory, an ECG was available for analysis. In the remaining 5.8%, the diagnosis was made solely on the basis of symptoms and enzyme changes. There were no significant differences in ECG availability between the two treatment groups.
Cause of death was classified by an independent Mortality and Morbidity Classification Committee as cardiac (direct or contributory), noncardiac but related to atherosclerotic disease, noncardiac medical cause (such as cancer or pulmonary disease), trauma, suicide, accident, other, or unknown.3 Cardiac death was defined as death within 1 hour after onset of cardiac symptoms or within 1 hour to 30 days after a documented or probable MI, death from intractable congestive heart failure or cardiogenic shock, or other documented cardiac cause. Cardiac death was considered contributory if cardiac dysfunction contributed to the death but it was unclear if it was the direct cause of death. Classification was based on the report from the clinical center’s principal investigator; death certificate; surgical and catheterization laboratory reports within 30 days; ECGs and enzyme data within 24 hours; patient’s baseline, procedural, and hospital study data; and, if available, a coroner’s report. Each case was reviewed by two committee members. Disagreements were resolved by consensus of the full committee.
The data were analyzed by intention to treat, except for comparisons of the index hospitalization for which only patients who actually received the assigned procedure were compared. Treatment differences for index hospital MI rates after the initial revascularization procedure were compared by use of the Fisher exact test. The Kaplan-Meier method was used to estimate 5-year cumulative event rates for mortality, cardiac mortality, the composite end point of cardiac mortality and Q-wave MI, and the composite end point of cardiac mortality and MI.18 MI rates included fatal and nonfatal events. We compared Kaplan-Meier curves using the log-rank test, with stratification according to clinical center to examine treatment differences.19 RRs in patient subgroups were estimated by Cox regression with terms for treatment, subgroup, and a treatment-by-subgroup interaction.20 The interaction term was used to test if treatment differences varied by subgroup level, eg, by number of diseased vessels. Cox regression analyses with a time-dependent variable to indicate presence of a follow-up MI were used to estimate the RR of cardiac death after an MI.
Patient subgroups analyzed in the present report include those defined a priori by symptomatic status, left ventricular function, number of diseased vessels, and presence of a type C lesion on the basis of the American Heart Association/American College of Cardiology consensus panel definition.21 A patient subgroup analysis of diabetics was requested and monitored early in the study by the Safety and Data Monitoring Board. To account for multiple subgroup comparisons, we used a significance level of .01 for the a priori subgroups and .005 for other subgroups.
There was no statistically significant difference between the two treatment groups in cumulative total mortality (111 deaths among CABG patients versus 131 among PTCA patients) after an average 5.4 years of follow-up ( RR=1.19, P=.19 by log-rank test). A cardiac cause of death was determined in 47 patients assigned to CABG compared with 72 assigned to the PTCA group. Other causes of death were attributed to 57 patients in the CABG group and 51 in the PTCA group, leaving 7 and 8 deaths, respectively, that could not be classified. Thus, the cumulative cardiac mortality rate at 5 years was 4.9% in the CABG group versus 8.0% in the PTCA group (RR=1.55; P=.022 by log-rank test) (Fig 1⇓). Cumulative rates for the composite end point of cardiac mortality or MI were not significantly different between the two treatment groups (RR=1.15; P=.23 by log-rank test). Cardiac mortality or MI rates at 5 years were 17.5% in the CABG group and 20.2% in the PTCA group (Fig 1⇓). The RR for cardiac death or Q-wave MI in patients assigned to PTCA versus CABG was 1.06 (P=.69 by log-rank test).
MI During and After the Index Hospitalization
Table 2⇓ illustrates that MI events during the index hospitalization and after the initial procedure were greater after CABG than after PTCA (41 Q-wave MI events in the patients who received CABG versus 19 in the PTCA group; P=.004). Of the 19 MI events in the PTCA group, 12 occurred after a CABG procedure for a failed PTCA.
After the index hospitalization discharge, the MI event rate was greater in the PTCA group than in the CABG group (122 versus 75 MI events; P<.001). The higher MI rate in the PTCA patients after hospital discharge was manifested as both Q-wave and non–Q-wave MI events (P=.019 and P<.001, respectively, by log-rank test).
Impact of MI on Cardiac Mortality
The RR for cardiac mortality after MI was 5.9 (P<.001) in the 264 patients who sustained an MI during follow-up. The probability value for the treatment-by-MI interaction was .110. Thus, the impact of MI on cardiac mortality was comparable between the two treatment groups. The RR of cardiac mortality increased to 13.2 (P<.001) in the 42 patients who sustained a second MI (18 in the CABG and 26 in the PTCA group).
The impact of first MI on cardiac mortality by treatment assignment is analyzed further by classifying the type of first MI according to whether or not the MI event occurred within 30 days of a procedure and if not, whether the MI event was associated with symptoms or was detected by a routinely scheduled follow-up ECG (silent Q-wave MI) (Table 3⇓). Patients who sustained a postprocedure or symptomatic MI had a significantly increased risk of cardiac mortality regardless of treatment assignment. Compared with no MI, the RR of cardiac mortality was 4.4 (P<.001) for postprocedure MI and 10.6 (P<.001) for symptomatic MI. Silent Q-wave MI was significantly related to cardiac mortality in the PTCA group (P<.01), although the number of events was too small to draw any conclusions; 2 of the 3 deaths in this category occurred in diabetic patients. There was no cardiac mortality in the 30 patients in the CABG group who had a silent Q-wave MI. In general, while the MI incidence was significantly greater in diabetic versus nondiabetic patients (20.9% versus 14.0%; P=.004), the proportion of MI detected during routine follow-up was virtually identical in both diabetic and nondiabetic patients (21%).
The treatment comparison of 5-year cardiac mortality rates by patient subgroups is illustrated in Fig 2⇓. The RR was similar among the patient subgroups with the exception of diabetics, for whom the probability value was .006 for treatment-by-subgroup interaction, which was close to the stringent .005 level of significance (Fig 2⇓, left). The RR of PTCA versus CABG was 3.12 for cardiac mortality (P<.001 by log-rank test) in the 353 diabetic patients receiving drug therapy at study entry. There were no significant differences among patient subgroups when the diabetic patients were removed from the analysis (Fig 2⇓, right). Overall, cumulative 5-year cardiac mortality rates were virtually identical for both treatment groups in the nondiabetic population (RR=1.04, P=.91; Fig 3⇓).
The analyses were repeated for the composite end point of cardiac mortality or MI. The RR was similar in the subgroups for the total population (Fig 4⇓, left) and among the 1476 patients remaining after exclusion of the diabetic group (Fig 4⇓, right).
Cumulative total mortality, the primary end point in BARI, was not significantly different between patients assigned to CABG and PTCA after an average of 5.4 years of follow-up, as previously reported.6 In the present report, we assess incidence of cardiac mortality and MI in both treatment groups, an important secondary end point of BARI. Cumulative cardiac mortality was significantly lower after CABG than after PTCA (P=.022) and was related to the significant treatment difference observed in the treated diabetic population (P<.001). A more detailed analysis of the treatment differences in the BARI randomized diabetic population is published elsewhere.7 In the nondiabetic patient population, there were no significant treatment differences in cardiac mortality, either overall or in any of the a priori subgroups examined. The two treatments were similar for the composite end point of cardiac mortality or MI.
Previous trials have not demonstrated significantly different cardiac mortality rates between patients assigned to CABG versus PTCA (2.0% versus 1.6% in RITA10 and 4.1% versus 3.5% in EAST11 ). These trials enrolled fewer patients than were enrolled in BARI, reported cardiac mortality at <3 years of follow-up, and included patients with single-vessel disease in the case of the RITA trial. Cardiac mortality accounted for 49% of the total mortality in BARI, similar to the 58% observed in EAST and the 53% seen in RITA.
The BARI experience currently reflects 5.4 years of follow-up compared with 1 to 3 years published in earlier trials and centrally classifies MI events by use of a more stringent definition (Table 4⇓).9 10 11 12 A comparison of 5-year MI rates in BARI, however, can be made with patients in the NHLBI PTCA Registry22 and with patients randomized to CABG in the CASS trial.23 In the BARI PTCA group, the 5-year MI rate was 13.1% and 20.2% in patients with two- and three-vessel coronary disease, respectively, similar to the 15.1% and 20.3% reported by Detre et al from the NHLBI PTCA registry.22 In the CASS randomized trial, the 5-year incidence of nonfatal Q-wave MI in patients assigned to CABG was 13% and 17% for patients with two- and three-vessel disease, respectively.23 The CASS protocol used a core ECG laboratory, and all ECGs were classified according to the Minnesota Code, with annual follow-up ECGs acquired in accordance with a protocol similar to that used in BARI.17 The 5-year rates of Q-wave MI in patients assigned to CABG in BARI were 12.7% and 10.4% in patients with two- and three-vessel coronary disease, respectively. The slightly lower rates of Q-wave MI in BARI than in CASS may reflect differences in inclusion criteria between the two studies, random variability, or perhaps endothelial stabilization with the availability of newer drug treatments such as HMG co-A reductase inhibitors and ACE inhibitors.
MI During and After the Index Hospitalization
The postprocedure MI rate in BARI patients was significantly greater in patients assigned to CABG than in those assigned to PTCA, confirming data reported from other randomized clinical trials of patients with multivessel coronary disease.9 11 The 4.6% Q-wave MI rate in BARI after CABG is less than the 8.1% rate reported in GABI and the 10.3% reported in EAST (Table 4⇑). The postprocedure Q-wave MI rate was 2.1% in BARI patients after PTCA, similar to the 2.3% to 3% rates reported in the GABI and EAST trials.9 11
The BARI definition for in-hospital postprocedure Q-wave MI required the new appearance of a two-grade (Minnesota Code) Q-wave worsening, a finding previously reported to be associated with increased mortality risk.17 The GABI and EAST trials defined new Q-wave MI descriptively as “new” or “pathologic” Q waves rather than using quantitative validated criteria such as the Minnesota Code. Thus, the relatively lower incidence of Q-wave MI after CABG in BARI compared with the other two trials may indicate less perioperative necrosis or may be due to differences in the definition of Q-wave MI.
CK isoenzymes were not routinely collected during the 96 hours after revascularization procedures in BARI. Early CK-MB enzyme elevations occur in 10% to 20% of patients who receive PTCA and are associated with a worse long-term outcome than in patients without enzyme elevation.24 25 Mild to moderate CK-MB enzyme elevations occur in 30% to 70% of patients who receive CABG but do not appear to have an impact on long-term outcome. All earlier randomized trials comparing PTCA with CABG, including BARI, underreport the true incidence of in-hospital postprocedure non–Q-wave MI because of the difficulties in assessing the significance of elevated serum markers after CABG compared with PTCA.24 25 Newer serum markers such as cardiac troponin I or T may be helpful in this regard but require further study in the immediate postrevascularization setting.26 27
The higher incidence of MI during the index hospitalization in BARI patients assigned to CABG was offset by an increased incidence of Q-wave and non–Q-wave MI in patients assigned to PTCA during the follow-up phase. The late increase in MI rate in the PTCA group may relate to the less complete revascularization obtained with PTCA compared with CABG. Additional studies and a longer duration of follow-up are necessary to address this issue.
Impact of MI on Cardiac Mortality
BARI clearly demonstrates that MI events had a significant adverse impact on cardiac mortality in both treatment groups. The risk was greatest in patients who sustained a postprocedure MI or a symptomatic MI. Silent Q-wave MI, detected during routine follow-up, posed less of a risk to the patient. The risk was only observed in the PTCA group, although the number of events was relatively small. The difference in mortality rates between symptomatic and silent MI after coronary revascularization may be explained in part by the fact that symptomatic patients tend to have larger MIs, leading to greater earlier cardiac mortality than for patients with silent MIs, who must survive the event to present for a follow-up ECG. Additional studies are required to more fully explore this hypothesis.
Routine collection of ECGs is necessary to ensure that silent Q-wave MIs are detected without bias between the two treatment groups. After randomization, 21% of all MIs were silent Q-wave events that were associated with a significant increase in cardiac mortality in the PTCA group. In the RITA trial,10 which randomized patients with single-vessel and multivessel coronary disease, 12% of MIs were silent after 2.5 years of follow-up. In the Reykjavik study,28 a population-based cohort study with a 4- to 20-year follow-up, approximately one third of all myocardial infarcts were clinically unrecognized, an observation similar to the Framingham data,29 and were associated with an adverse prognosis.
MI detection depends on the frequency of ECG and cardiac enzyme sampling. The procedure-related non–Q-wave MI rate in BARI is underestimated because cardiac enzymes were not sampled in the initial 96 hours after the procedure. The incidence of silent Q-wave MI may also be underestimated. In the POSCH study of 30- to 64-year-old patients with hyperlipidemia who sustained an MI, 32% had regression of diagnostic Minnesota Code Q-QS items after an average 2.2-year follow-up. Regression occurred more frequently for isolated inferior infarcts (43%) than anterior infarcts (23%).30 The data indicate that even with ECG sampling once a year, there is a risk of missing silent Q-wave MI events because of Q-wave regression. Nevertheless, the BARI trial is the first and largest randomized clinical trial comparing PTCA with CABG to report 5-year MI rates that required annual follow-up ECGs analyzed by use of validated Minnesota Code criteria and a central ECG and MI classification laboratory.
The incidence of ischemic cardiac events in BARI will likely increase with a longer duration of follow-up, which may potentially influence differences observed between the two treatment groups. Atherosclerotic disease progression in saphenous vein grafts is more prevalent 5 to 10 years postoperatively than in the first 5 years. The long-term impact of saphenous vein graft attrition is less likely to be a problem in BARI than in earlier studies because 82% of BARI patients received internal mammary artery conduits.
New interventional devices such as coronary stents, which reduce restenosis rates, and platelet glycoprotein IIb/IIIa receptor blocking agents, which decrease acute occlusion rates, were not used during the initial revascularization procedure in BARI.31 32 To date, no study has demonstrated that the use of stents reduces the 5-year incidence of death or MI compared with standard PTCA.33 Long-term follow-up data with newer stent devices are unavailable, and randomized clinical trials comparing recently designed stents with CABG are just beginning.
The BARI results show that diabetic patients with multivessel coronary disease have a significant increase in cardiac mortality rates after 5.4 years of follow-up with an initial treatment strategy of PTCA compared with CABG. There are no significant differences in cardiac mortality rates or MI risk in nondiabetic patients with multivessel coronary disease. Additional revascularization procedures and hospital admissions, however, were more common when PTCA was selected as the initial treatment modality.6 The long-term impact of new catheter-based approaches, currently used in 30% to 70% of all PTCA procedures, needs to be investigated in the same rigorous manner as standard PTCA to provide the best information and optimal guidelines for clinicians.33
Selected Abbreviations and Acronyms
|BARI||=||Bypass Angioplasty Revascularization Investigation|
|CABG||=||coronary artery bypass graft|
|NHLBI||=||National Heart, Lung, and Blood Institute|
|PTCA||=||percutaneous transluminal coronary angioplasty|
BARI is funded by the NHLBI and is supported by grants (HL-38493, HL-38504, HL-38509, HL-38512, HL-38514-6, HL-38518, HL-38524-5, HL-38529, HL-38532, HL-38556, HL-38610, HL-38642, and HL-42145) from the NHLBI.
Reprint requests to Katherine Detre, MD, DrPH, BARI Coordinating Center, Graduate School of Public Health, University of Pittsburgh, 130 DeSoto St, Pittsburgh, PA 15261.
↵1 A complete list of the BARI investigators has been published in Circulation. 1991;84(suppl V):V-23-V-27.
- Received February 3, 1997.
- Revision received May 5, 1997.
- Accepted May 19, 1997.
- Copyright © 1997 by American Heart Association
Rogers WJ, Alderman EL, Chaitman BR, DiSciascio G, Horan M, Lytle B, Mock BM, Rosen AD, Sutton-Tyrrell K, Weiner BH, Whitlow PL, and the BARI Study Group. Bypass Angioplasty Revascularization Investigation (BARI): baseline clinical and angiographic data. Am J Cardiol. 1995;75:9C-17C.
Protocol for the Bypass Angioplasty Revascularization Investigation. Circulation. 1991;84(suppl V):V-1-V-27.
Alderman EL, Anrews K, Bourassa MG, Brooks MM, Chaitman BR, Detre K, Faxon DP, Follman D, Frye RL, Hlatky M, Jones RH, Kelsey SF, Rogers WJ, Rosen AD, Schaff H, Sellers MA, Sopko G, Sutton-Tyrell K, Williams DO, on behalf of the Bypass Angioplasty Revascularization Investigation (BARI) Investigators. Five-year clinical and functional outcome in a multicenter randomized trial comparing bypass surgery and angioplasty in patients with multivessel disease. JAMA. 1997;277:715-721.
Hlatky MA, Rogers WJ, Johnstone I, Boothroyd D, Brooks MM, Pitt B, Reeder G, Ryan T, Smith H, Whitlow P, Wiens RD, Mark DB, Rosen AD, Detre KM, Frye RL, for the Bypass Angioplasty Revascularization Investigation (BARI) Investigators. Medical care costs and quality of life after randomization to coronary angioplasty or coronary bypass surgery. N Engl J Med. 1997;336:92-97.
The BARI Investigators. The influence of diabetes on 5-year mortality and morbidity in a randomized trial comparing CABG and PTCA in patients with multivessel disease: the Bypass Angioplasty Revascularization Investigation (BARI). Circulation. In press.
Hamm CW, Reimers J, Ischinger T, Rupprecht HJ, Berger J, Bleifeld W, for the German Angioplasty Bypass Surgery Investigation. A randomized study of coronary angioplasty compared with bypass surgery in patients with symptomatic multivessel coronary disease. N Engl J Med. 1994;331:1037-1043.
Prineas RJ, Crow RS, Blackburn H. The Minnesota Code Manual of Electrocardiographic Findings: Standards and Procedures for Measurement and Classification. Wright J, ed. Boston, Mass: John Wright. PSG Inc; 1982:1-229.
Rautaharju PM, Calhoun HP, Chaitman BR. NOVACODE serial ECG classification system for clinical trials and epidemiologic studies. J Electrocardiol. 1992;24:179-187.
Chaitman BR, Zhou SH, Tamesis B, Rosen A, Terry AB, Zumbehl KM, Stocke K, Takase B, Gussak I, Rautaharju PM. Methodology of serial ECG classification using an adaptation of the NOVACODE for Q wave myocardial infarction in the Bypass Angioplasty Revascularization Investigation (BARI). J Electrocardiol. 1996;29:265-277.
Chaitman BR, Alderman EL, Sheffield LT, Tong T, Fisher L, Mock MB, Wiens RD, Kaiser GC, Roitman D, Berger R, Gersh B, Schaff H, Bourassa MG, Killip T, and Participating CASS Medical Centers. Use of survival analysis to determine the clinical significance of new Q waves after coronary bypass surgery. Circulation. 1983;67:302-309.
Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959;22:719-748.
Cox DR. Regression models and life-tables. J R Stat Soc (B). 1972;34:187-220.
Ryan TJ, Bauman WB, Kennedy JW, Kereiakes DJ, King SB, McCallister BD, Smith SC Jr, Ullyot DJ. Guidelines for percutaneous coronary angioplasty: a report from the American Heart Association/American College of Cardiology Task Force on Assessment of Diagnostic and Therapeutic Cardiovascular Procedures (Committee on Percutaneous Transluminal Coronary Angioplasty). Circulation. 1993;88:2987-3007.
Detre K, Yeh W, Kelsey S, Williams D, Desvigne-Nickens P, Holmes D, Bourassa M, King S, Faxon D, Kent K, for the Investigators of the NHLBI Percutaneous Transluminal Coronary Angioplasty Registry. Has improvement in PTCA intervention affected long-term prognosis? The NHLBI PTCA registry experience. Circulation. 1995;91:2868-2875.
Chaitman BR, Jaffe AS. What is the true periprocedure myocardial infarction rate? Does anyone know for sure? The need for clarification. Circulation. 1995;91:1609-1610.
Califf RM, Abdelmeguid A, Kuntz R, Popma JJ, Tardiff BE, Crenshaw B, Bauman RP, Davidson CJ, Cohen EA, Kleiman NS, Mahaffey KW, Topol EJ, Pepine CJ, Lipicky R, Granger CB, Harrington RA, Zuckerman B, Chaitman BR, Bittl JA, Ohman EM. Myonecrosis after revascularization procedures. J Am Coll Cardiol. In press.
Ohman EM, Armstrong PW, Christenson RH, Granger CB, Katus HA, Hamm CW, O’Hanesian MA, Wagner GS, Kleiman NS, Harrell FE, Califf RM, Topol EJ, for the GUSTO-IIa Investigators, Cardiac troponin T levels for risk stratification in acute myocardial ischemia. N Engl J Med. 1996;335:1333-1341.
Serruys PW, de Jaegere P, Kiemenij F, Macaya C, Rutsch W, Hendrickx G, Emanuelsson H, Marco J, Legrand V, Materne P, Belardi J, Sigwart U, Colombo A, Goy J, Van Den Heuvel P, Delca J, Morel M, for the Benestent Study Group. A comparison between balloon-expandable-stent implantation with balloon angioplasty in patients with coronary artery disease: Benestent Study Group. N Engl J Med. 1994;331:489-495.