Prognostic Effect of Bundle Branch Block Related to Coronary Artery Bypass Grafting

ALAN CHU, MD, ROBERT M. CALIFF, MD, DAVID B. PRYOR, MD, RAY A. McKINNIS, PhD, FRANK E. HARRELL, Jr., PhD, KERRY L. LEE, PhD,

STEVE E. CURTIS, PA, H. NEWLAND OLDHAM, Jr., MD, and GALEN S. WAGNER, MD

The incidence and prognostic effect of the develop- ment of new perioperative ventricular conduction abnormalities were examined in all patients under- going coronary artery bypass surgery at Duke Uni- versity Medical Center between 1976 and 1961. Cf the 913 patients included, transient (resolved before discharge) ventricular conduction abnormalities de- veloped in 156 (17 % ) and persistent (until dis- charge) changes developed in 126 (14% ). Com- plete right bundle branch block (BBB) was the most frequent type of new ventricular conduction abnor- mality, followed by left anterior hemibtock and in- complete right BBB (found in 60%, 26% and 9%) respectively, of all patients with transient changes and 29 % , 33 % and 26% of all patients with per- sistent changes). Development of new ventricular

conduction abnormalities was most strongly related to date of operation (p <O.OOOl, untvariate x2 = 122), increasing from 2% transient and 7% persfs- tent in 1976 to 36% transtent and 22% perststent in 1961. The incidence was also higher in otder pa- tients. Preoperative ejection fraction and number of diseased vessets were related to development of perioperattve ventricular conduction abnormaltties but were not independently related after adjustment for other baseline characteristks. Contrary to find- ings in other studies, development of new perioper- ative ventricular conduction abnormaltttes, including isolated new left BBB, did not worsen the survival rate in patients followed up to 3 years after surgery.

(Am J Cardld 1967;59:796-603)

V entricular conduction abnormalities have been ob- served with variable frequency (6 to 44%) after coro- nary artery bypass grafting (CABG).l-6 Their clinical

From the Departments of Medicine and Surgery, Duke Univer- sity Medical Center, Durham, North Carolina. This work was supported by Research Grant HL-17670 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland; NRSA Training Grant 5T 32 HL07101-09 from the National Institutes of Health; Research Grants HS- 03834 and HS-04873 from the National Center for Health Ser- vices Research, Hyattsville, Maryland; Training Grant LM-07003 and Grant LM-03373 from the National Library of Medicine, Bethesda; and grants from the Prudential Insurance Company of America, Newark, New Jersey, Kaiser Family Foundation, Palo Alto, California, and Andrew W. Mellon Foundation, New York, New York. Manuscript received Sep tember 2, 1986: revised manuscript received October 28, 1986, accepted October 30,1986.

Address for reprints: Galen S. Wagner, MD, Box 31211, Duke University Medical Center, Durham, North Carolina 27710.

799

significance is not well defined, although it has been suggested that these conduction abnormalities repre- sent ischemic myocardial damage.z6 In the few studies that have addressed their prognostic significance,2~6 left bundle branch block (BBB) was observed to be associated with a worse prognosis. These studies were limited by small sample sizes and failure to adjust for other important variables contributing to mortality rate such as ventricular function and extent of coronary disease. Furthermore, previous reports evaluated pa- tients who underwent CABG between 1973 and’ 1976.2v6 Major advances in surgical techniques and myocardial preservation have occurred since then. This study was designed to examine the incidence and clinical significance of ventricular conduction abnor- mality development in consecutive patients undergo- ing CABG between 1976 and 1981 at Duke University Medical Center. Patients with perioperative left BBB were also compared with those who had preoperative left BBB. We hypothesized that perioperative left BBB developed through a mechanism different from that of

April 1,1967 THE AMERICAN JOURNAL OF CARDIOLOGY Volume 59 799

TABLE I Deflnition of Conduction Abnormalities

Right BBB

Left BBB

QRS 20.12 second, terminal forces anterior

QRS 20.12 second, terminal forces posterior

Left anterior hemiblock QRS between -45’ and - 120’ with axis abnormality not a result of Q

Left posterior hemiblock QRSbetween 120’ and 180’ without right ventricular hypertrophy, lateral myocardial infarction or car pulmonale

Right-axis deviation QRS between 90’ and 180’ Left-axis deviation QRSbetween -30° and -45’ or

between -30’ and -120’ with left BBB

Incomplete right BBB QRS>O.OE but 10.12 second either with R’in V, >0.06 mV or terminal forces anterior without right ventricular hypertrophy

lntravantricular conduction delay QRS >O.li second; includes peri-infarct block, but without right BBB. left BBB or left anterior hemiblock

Trifasclcular block I0 AV block, plus right BBB plus either left anterior hemiblock or left posterior hemiblock

AV = atrioventricular; BBB = bundle branch block.

preoperative left BBB and that perioperative BBB would carry a better prognosis in patients undergoing CABG.

Methods All patients undergoing cardiac catheterization for

chest pain and surviving until hospital discharge after subsequent CABG at Duke between January 1976 and June 1981 were entered into the study. Patients under- going concomitant valvular surgery, repeat CABG, an- eurysmectomy, surgery for ventricular septal defect repair, surgical ablation of conduction abnormalities or diagnosed to have preexcitation syndrome were excluded.

The computerized information system used in this study has been describeda At the time of cardiac cath- eterization, a complete medical history, physical ex- amination, chest x-ray and standard l&lead electro- cardiogram were obtained. The data were itemized into 48 characteristics that formed the noninvasive preoperative database in this study. Twenty invasive hemodynamic, angiographic and ventriculographic descriptors were obtained by cardiac catheterization.* Eighteen selected descriptors were used in this study.

All patients underwent biplane left ventriculog- raphy and coronary angiography. Results were re- viewed by at least 2 senior angiographers. Significant coronary artery disease was defined as 75% or more diameter narrowing of a major coronary artery, in- cluding the left main, left anterior descending or large diagonal branch, left circumflex, large optional diago- nal branch or large marginal branch and right coro- nary artery. Left ventricular ejection fraction was cal- culated using the modified area-length method.9

TABLE II Subgroups of Conduction Abnormalltles and Severity Score

Group A Incomplete right BBB, left-axis deviation, left anterior

hemiblock, lefl posterior hemiblock, right-axis deviation Group B

Left BBB, right BBB, intraventricular conduction delay, trifascicular block

Severity score 0 = No conduction abnormalities 1 = Group A transient 2 = Group B transient 3 = Group A persistent 4 = Group B persistent

Serial electrocardiograms were recorded 1 day be- fore surgery, within 2 hours postoperatively, routinely after operation until discharge and during follow-up visits. All available electrocardiograms from each pa- tient were reviewed by a senior cardiologist. The pres- ence and type of ventricular conduction abnormalities were determined according to the criteria of Rosen- baum,lO and Hecht and their co-workers11 (Table I). Ventricular conduction abnormalities were further classified as old if present before operation; new if they developed perioperatively; transient if they re- solved before discharge: and persistent if they persist- ed until discharge. Development of a new transient conduction abnormality did not preclude other tran- sient abnormalities; all of them were included in the analysis. In the event that a new persistent conduction abnormality or abnormalities developed, transient ab- normalities were not included. Conduction abnormal- ities were separated into 2 groups and assigned a se- verity score (Table II].

The operative technique has been described.12*13 Direct revascularization of the involved coronary arte- ries was performed using reverse saphenous vein or internal mammary artery bypass grafts, cardiopulmo- nary bypass and mild to moderate systemic hypother- mia (28 to 32’C). Early in the series, induced but not maintained electrical fibrillation, topical saline slush and hypothermia were used. As of 1978, increased protection with crystalloid cardioplegia was used in some patients. By 1981, all patients had crystalloid car- dioplegic arrest in addition to hypothermia and saline topical slush.

Follow-up data on survival status were obtained at 6 months and 1 year after operation and yearly there- after as reported in a previous study.s Cardiovascular death is defined as either sudden death without rela- tion to noncardiovascular events, or as death resulting directly from a cardiovascular event including myo- cardial infarction, cerebrovascular accidents and re- fractory congestive heart failure.

Multivariable analysis was performed to assess the association between perioperative ventricular con- duction abnormality development and other baseline characteristics using the logistic regression model for ordinal response variables.14*15 The dependent var- iable is the ordinal severity score defined above. Development of each perioperative ventricular con-

600 BYPASS-RELATED BUNDLE BRANCH BLOCK

TABLE III Selected Basellne Characteristics of Patients

Conduction Abnormalities

Characteristic None Transient

(n = 631) (n = 156) Persistent (n = 126)

Men

Age (yr)’ Diabetes mellitus 0 wave on ECG Ejection fraction (%)’ No. of CAs narrowed 275 % l

1 2 3

No. of grafts’ 1 n L

23 Emergency surgery

65% 51 f 6

10% 32%

54f 12 2.29 f 0.61

19% 31% 50%

2.50 f 0.96 16% 32% 52%

2%

76% 55 f 6

13% 37%

56 f 12 2.42 f 0.70

12% 33% 55%

3.04 f 1.19 6%

27% 67%

1%

66% 56 f 6

6% 30%

56f 11 2.65 i 0.56

4% 27% 69%

3.26 f 1.04 2%

17% 61%

0%

‘Values are mean f standard deviation. CAs = coronary arteries; ECG = electrocardiogram.

TABLE IV fncldence’

New VCDs Left BBB IVCD

Right BBB

Trifasc. Block

Inc. Right BBB LAD LAH RAD LPH

Transient (n = 156) 17 0 93 2 14 15 40 10 4 Persistent (n = 126) 6 0 36 3 33 16 41 9 0

Development of 1 type of abnormality does not exclude the concomitant existence or subsequent conversion to other types. Each type is treated as an individual incidence.

BBB = bundle branch block; IVCD = intraventrlcular conduction delay; LAD = left-axis deviation; LAH = left arterior hemiblock: LPH = left posterior hemiblock; n = number of patients: RAD = right-axis deviation; Trifasc. = trifascicular; VCDs = ventricular conduction defects.

1976 ,

1978 1979 1980 1981

Year

FIGURE 1. Relation between development of perloperatlve ventrlc- ular conduction abnormalltles (PGCA) and date of operation. Dots, transient; frlangles persistent; squares, combined.

duction abnormality was treated as an individual inci- dence. Analyses for the relation between survival time and various patient characteristics were performed us- ing Breslow’s formulation of the Cox model,i6-*8 taking into consideration the severity score. Univariable as- sociations were determined by Rao’s efficient score statisticsi

Results Between January 1976 and June 1981,965 consecu-

tive patients underwent CABG for the first time at Duke. Fifty-two did not meet selection criteria; the remaining 913 comprise our final study group. Base- line characteristics of the 913 patients are shown in Table III. The patients were mostly men, average age older than 50 years, and had relatively normal preop- erative ejection fraction. At least half of the patients had 3-vessel disease and received at least 3 grafts. The incidence of the individual types of perioperative ven- tricular conduction abnormalities are listed in Table IV.

The univariable associations of selected baseline characteristics with new conduction abnormality de- velopment are reported in Table V. In this analysis, the severity score (Table II) was taken into consideration. The severity score was related most strongly to the date

April 1. 1967 THE AMERICAN JOURNAL OF CARDIOLOGY Volume 59 801

TABLE V Selected Baseline Characteristics Related to Development of Perloperatlve Ventricular Conduction Abnormalities’

Unadjusted Adjusted

x*t x2%

Date of surgery 122 86.2

Age 42 12.2 Sex 0.4 1.1 Diabetes mellitus 0.01 1.0 Hypertension 0.7 0.3 Peripheral vascular disease 2.3 0.9 Typical angina 0.01 0.0 Previous MI 0.5 0.1 Preop ventricular conduction abnormality 0.13 0.5 Ejection fraction 9.9 1.4 No. of diseased vessels 23 2.9 Significant left main disease 12 0.9 Significant left anterior descending disease 1 5.4 Significant circumflex disease 21 4.0 Significant right coronary disease IO 0.0

x2 > 3.64, p <0.05; >6.64, p <O.Ol; >10.81, p <O.OOl; >15.14, p <0.0001.

* Dependent variable is severity score defined in Table II. T Unadjusted x2 obtained by testing each variable singly. $ Adjusted x2 obtained by testing the contribution of each variable after

adjusting for the other varlables.

of operation: The incidence increased dramatically from 2% transient and 7% persistent (9% combined) in 1976 to 36% transient and 22% persistent (58% com- bined) in 1981 (Fig. 11. After adjusting for other base- line characteristics, the date of operation remained most strongly related to development of conduction abnormalities, followed by age: The incidence was greater in older patients. The preoperative ejec- tion fraction and the number of diseased vessels were not independently related to development of conduction abnormalities after adjusting for other characteristics.

Figure 2 shows the relation between survival and development of conduction abnormalities. Develop- ment of perioperative ventricular conduction abnor- malities was not related to a worse prognosis com- pared with control patients (group O]. The independent prognostic value was further tested using the Cox mod- el. In this evaluation, the severity (group A < group B) and the duration [transient < persistent] were exam- ined separately as 2 variables. No significant prognos- tic information was added before or after adjusting for other known independent prognostic indicators, in- cluding descriptors of history, physical examination, rest electrocardiogram, coronary anatomy, ejection fraction and hemodynamic data (unadjusted x2 with 2 d.f. = 1.65, p = 0.44; adjusted x2 = 1.34, p = 0.51). The independent prognostic value of development of peri- operative left BBB (n = 23) was also tested. Again, development of new perioperative left BBB was not associated with a worse prognosis (x2 = 0.03, p = 0.9).

When the independent prognostic value of preop- erative left BBB (n = 13) and intraventricular conduc- tion delay (n = 28) was examined after adjusting for other known independent prognostic indicators, pre- operative left BBB was associated with survival (ad-

TABLE VI Selected Characteristics of the 13 Patients Who Underwent Coronary Artery Bypass Grafting with Preoperative Left Bundle Branch Block

% Diameter Narrowing

Age Pt Year (yr) & Sex LAD Right LCx LM EF (%) Survival

1’ 2’ 3’ 4 5 6 7 8 9

10 11 12 13

1977 1981 1976 1976 1961 1977

1977 1979 1980 1981 1961

57 F 95 100 100 0 37 PD 65 F 100 95 75 50 38 PD 64 F 95 95 50 50 62 PD 59 M 95 100 75 0 51 36 mo 46 F 75 75 100 0 42 24 mo 51 M 95 100 100 0 26 31 mo 52 M 95 0 0 0 40 72 mo 51 M 75 25 95 0 52 58 mo 53 M 0 95 100 0 49 Alive 64 F 95 95 25 0 68 Alive 42 M 25 95 75 0 25 Alive 53 M 95 75 0 0 57 Alive 50 M 100 95 95 0 61 Alive

* Not included in the study group of 913 patients because of failure to survive surgery.

EF = ejection fraction; LAD = left anterior descending coronary artery; LCx = left circumflex coronary artery: LM = left main coronary artery; PD = perioperative death; Year = year of surgery.

justed x2 = 22.3, p <O.OOOl) but preoperative intraven- tricular conduction delay was not (adjusted x2 = 0.76, p = 0.38). Of the 13 patients [Table VI) who had preoper- ative left BBB and underwent CABG between January 1976 and June 1981, only 10 survived the operation and were included in our study group. Of these 10 patients, 3 (30%) died within 3 years after the operation.

As a group, these 13 patients had a mean ejection fraction of 47% and mean number of diseased vessels of 2.38. All 3 patients who died perioperatively were women, 2 of whom had ejection fractions of less than 40% and 3-vessel disease, with 1 patient having an additional 50% left main stenosis. The third patient had a normal ejection fraction, &vessel disease plus a 50% left main stenosis. The 3 patients who survived the operation but died within 3 years all had 3-vessel

L lz I 9 0

t

60)

OO I I 1 I 2 3

Years post-op

FIGURE 2. Relation between survival and development of perloper- atlve ventricular conductlon abnormalltles. The conductlon abnor- malltles were dlvlded Into group A and group B and assigned a severity score (Table II).

BYPASS-RELATED BUNDLE BRANCH BLOCK

TABLE VII Different Patient Selection Crlteria and Incidence In Previous Reports of Perioperative Ventricular Conduction Abnormalities

Study No. of Year of

PtS Surgery Valve Surgery

Included Time (hr) of 1st

Postop ECG Incidence (% )

Satinsky et al’ 280 1970-1972 + Not specified 8 Zeldis et al* 200 1973-1975 0 1 20 Bantea et al3 322 1975-1978 0 15 8 O’Connell et al4 111 1980 0 1 19 Rippe et al6 286 1978-1977 Not specified 1 37

1978-1980 Bateman et al* 227 1975-1978 0 2 28

ECG = electrocardiogram.

disease and a mean ejection fraction of 40%. The re- maining patients had a mean ejection fraction of 50% and a mean of 2 diseased vessels.

Discussion The major findings in this study are: Perioperative

ventricular conduction abnormalities are developing with increasing frequency in patients undergoing CABG, from 9% (2% transient, 7% persistent) in 1976 to 58% (36% transient, 22% persistent) in 1981. How- ever, contrary to previous reports2s6 there is no associ- ated prognostic effect on patient survival either when all perioperative conduction abnormalities are consid- ered as an entire group or when considering patients in whom new left BBB develops.

Our patient population is by far the largest ever included in studies of this kind. The overall incidence of new perioperative conduction abnormalities in our patients was 31% (17% transient, 14% persistent). This is consistent with the incidence in previous studies (Table VII). The slight variability in the incidence may be due to the different types of conduction abnormali- ties included, the interval between surgery and the first electrocardiogram, zig the number of patients, concomitant valvular surgery in some studies, and most likely also due to the year in which the surgery was done. The incidence in our patients progressively increased, from 9% in 1976 to 58% in 1981 (Fig. 1). This was true even after correction for other selected base- line patient characteristics. This interesting observa- tion was similar to that reported in the study by Rippe et al5 (18% between 1976 and 1977 and 44% between 1978 and 1980). Right BBB was the most frequently developed perioperative conduction abnormality in our study and in those reported previously.l-6 The rea- son for this higher incidence of right BBB is unclear.

The observation of a progressive 6- to 7-fold in- crease in the incidence of perioperative ventricular conduction abnormalities in recent years despite im- proving survival in CABG patients is itself a strong indication that these abnormalities do not worsen sur- vival. Results from our study also indicate that devel- opment of new perioperative left BBB was not associ- ated with shortened survival in patients who undergo CABG. The latter finding was contrary to previous reports2s6 In the study by Zeldis et aL2 patients with left BBB, left anterior hemiblock or both, had a higher mortality rate (19% vs 7%) than control patients at a

follow-up of 13 to 39 months. They concluded that these conduction disturbances developing after CABG identified a subset of patients with more extensive ischemic heart disease, although supporting evidence such as poor ejection fraction and multivessel disease was not presented. Bateman et al6 reported that 38% of patients with new perioperative left BBB or intraven- tricular conduction delay died during the follow-up period of 66 months, compared with 3% for the other CABG patients. Their high risk group had significantly more collateralized regions on preoperative angio- grams (77% vs 47%). more grafts placed (3.6 vs 3.1) and a significantly higher incidence of perioperative in- farction (77% vs 19%), although their mean preopera- tive ejection fraction was only slightly lower (61% vs 63%). However, when comparing survivors (n = 14) to nonsurvivors (n = 21) at 1 year after revascularization in a larger group of patients including those with pre- operative left BBB or intraventricular conduction de- lay and 15 additional CABG patients with similar con- duction abnormalities who died within 1 year of surgery, the nonsurvivors were older, had worse pre- operative angina, poorer left ventricular function, more extensive coronary artery disease and 43% had had perioperative infarction.

Several factors may contribute to the discrepancy between the prognostic significance in our patients with perioperative BBB and the others.296 Our patient population is the largest group reported: 913 vs 200~ and 227.6 Although the number of patients in whom perioperative left BBB and left anterior hemiblock de- veloped is still small-23 and 81, respectively-it is still more than twice those reported earlier.2Jj The smaller number in previous series did not allow for multivariable analysis for assessment of independent contribution of BBB to survival. The observed poor prognosis in those patients may be a result of other stronger underlying descriptors, such as poorer ven- tricular function and more extensive coronary lesions.6 In our study, however, perioperative devel- opment of any of these ventricular conduction abnor- malities, with or without adjustment for other vari- ables, did not lead to worse prognosis. The date of surgery may play a key role in explaining the different outcome. In our study, there was a progressive in- crease in incidence (Fig. 1). In the other 2 studies in which the prognostic significance was addressed,2p6 the operations were performed between 1973 and

April 1, 1967 THE AMERICAN JOURNAL OF CARDIOLOGY Volume 59 803

1976. With improvement of surgical technique or myo- cardial protection with hypothermic cardioplegia, perioperative myocardial damage may have been re- duced in the more recent years.lg~zO It has been sug- gested that perioperative ventricular conduction ab- normalities, which probably resulted from myocardial ischemia/injury in CABG done in earlier years,2,6 rep- resents local effects of cardioplegic solutions on the conduction system or increased cooling during CABG performed after 19774,s; the combination of a reduction of injury-related and an increase in cardioplegia/hy- pothermia-related conduction abnormalities could lead to the discrepancy between our study and others. To assess the individual contribution of injury vs car- dioplegia, accurate assessment of perioperative myo- cardial infarction must be made. Although most agree that perioperative infarction is occurring less fre- quently, accurate diagnosis and quantitation of perio- perative infarction remain difficult. Other factors such as the increased widespread use of p-blocking and calcium channel-blocking drugs in the late 1970s in patients with coronary artery disease may also con- tribute to these differences. Contributions of these agents to improved survival were not systematically examined.

Preoperative left BBB contributed significantly to survival when its independent prognosic value was examined using the Cox model and adjusting for other known independent prognostic indicators (adjusted x2 = 22.3, p <O.OOOl). Of the 13 patients with preoperative left BBB (Table VI), 10 survived operation. This amounted to 23% operative mortality rate, much high- er than the 4.6% average operative mortality during the entire study period. Of these 10 patients, 3 (30%] died within 3 years; this rate is comparable to the 38% rate reported by Bateman et a1.6 The patients with preoperative left BBB in our study were all women, had lower ejection fraction (47%) and more significant coronary disease. All of the patients with at least 50% left main disease in this group died perioperatively. The three patients who died within 3 years postopera- tively from cardiovascular causes all had more severe coronary lesions and a worse ejection fraction than the other survivors. Myocardial and conduction system damage is the most likely cause of the left BBB for this group. The prognosis of these patients was clearly worse than those with new perioperative left BBB and other abnormalities, again suggesting that the develop- ment of new perioperative left BBB is due to factors such as hypothermia/cardioplegia rather than myo- cardial ischemia/injury. Further prospective con- trolled studies must be done to evaluate the mecha- nism of conduction abnormalities in patients undergoing CABG.

Clinica implications: Although we are observing a dramatic increase in perioperative ventricular con- duction abnormalities in patients undergoing CABG, these conduction abnormalities do not worsen survival risk despite their ominous appearance on postopera- tive electrocardiograms. Contrary to studies done in patients undergoing CABG before 1977296 and unlike patients with preoperative left BBB, our data suggest

that development of new perioperative left BBB or other BBB in CABG patients does not identify a subset with more extensive myocardial disease and does not warrant closer follow-up. However, we did not exam- ine the prognostic effect of perioperative ventricular conduction abnormalities on other clinically signifi- cant cardiovascular events, such as recurrent myocar- dial infarction, high-grade (type II, second- or third- degree) atrioventricular block or congestive heart failure. Further studies must be performed to define these outcomes.

Acknowledgemenh We thank Joseph C. Green- field, Jr., MD, for helpful guidance in the design and conduct of this study and critical review of the manu- script. We also thank Sousin Leggett and staff for tech- nical assistance and Gail McKinnis and Sylvester Cherry for secretarial assistance.

References 1. Satinsky ID, Collins JJ Jr, Dalen ]E. Conduction defects after cardiac sur- gery. Circulation 1974;5O:suppI II:II-170-H-174. 2. Zeldis SM, Morganroth 1, Horowitz LN, Michelson EL, Josephson ME, Lozner EC, MacVaugh H, Kastor JA. Fascicular conduction disturbances after coronary bypass surgery. Am J Cardiol 1978;41:860-864. 3. Bantea C, Bleese N. Kalmar P, Krebber HJ, Radiger W, Rodewald G. Periooerativ entstqndenen komuletter rechtsschenkelhlock nach aorto-cor- onarer bypass-chisurgie. Herz i%n;6:m. 4. O’Connell JB, Wallis D, Johnson A, Pifarre R, Gunnar RM. Transient bundle branch block following use of hypothermic cardioplegia in coronary bypass surgery: high incidence without perioperative myocardial infarction. Am Heart J 1962;103:65-91. 5. Rippe JM. Browning C. Salm TV, Goldberg R, Alpert JS, Dalen JE. Fascicu- Iqr conduction disturbances following aortocoronary bypass surgery: the role of hypothermia vs potassium-arrest cardioplegio. J Cardiovasc Surg 1984;25: 456-461. 6. Bateman TM, Weiss MH. Czer LSC. Conklin CM, Kass RM, Stewart ME, Matloff JM, Gray RJ. Fascicular conduction disturbances and ischemic heart disease: .adverse prognosis despite coronary revascularization. JACC 1985; 5:632-639. 7. Rosati RA, McNeer JF. Starmer CF, Mittler BS, Morris JJ, Wallace AG. A new information system for medical practice. Arch Intern Med 1975;135: 1017-1024. 8. Harris PJ, Harrell FE, Lee KL, Behar VS, Rosati RA. Survival in medically treated coronary artery disease. Circulation 1979;60:1259-1269. 9. Rackley CE, Behar VS, Whalen RE, McIntosh HP. Biplane cineqngiograph- ic determinants of left ventricular fraction: pressure-volume relationships. Am Heart J 1979;74:766-779. 10. Rosenbaum MB, Elizari MV, Lazzari JO. The Hemiblocks: New Concepts of Intraventricular Conduction Based on Human Anatomical, Physiological and Clinical Studies. Oldsmar FL:Tampa Tracings, 1978. 11. Hecht HH, Kossman CE, Childers RW, Langendorf R, Lev M, Rosen KM, Pruitt RD, TruexRC, Uhley HN. Watt TB. Atrioventricularand introventricu- lar conduction: revised nomenclature and concepts. Am J Cardiol 1973;31: 232-244. 12. Oldham H, Kong Y. Bartel A, Morris J. Behar V, Peter R, Rosati R, Young G, Sabiston Il. Risk factors in coronary artery bypass surgery. Arch Surg 1972;105:9l8-923, 13. Rankin JS, Newton ]R Jr, Califf RM, Jones RH. Wechsler AS, Oldham HN lr. Wolfe WG. Lowe IE. Clinical characteristics and current manaaement of medically refractory’unstable angina. Ann Surg 1984;289:457-4657 14. Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrikq 1967;54:167-179. 15. Harrell FE. The LOGIST procedure. In: SUGI Supplemental Library User’s Guide, Version 5 Edition. Gary. NC:SAS Institute, Inc., 1986:269-293. 16. Cox DR. Regression models and life tables (with discussion) J R Stat Sot B 1972;34:187-220. 17. Breslow NE. Covariance analysis of censored survival data. Biometrics 1979;30:69-99. 16. Harrell FE. The PHGLM Procedure. In: SUGI Supplemental Library User’s Guide, Version 5 Edition. Gary, NC:SAS Institute, 1986:437-466. 19. Engelman RM. Ronson JH, Vertrees RA, Rohrer C, Anvil J. Safety of prolonged ischemic arrest using hypothermic cardioplegia. J Thorac Cardio- vast Surg 1980;79:705-712. 20. Kouchoukos NT, Oberman A. Kirklin JW, Russell RO Jr, Karp RB. Pacific0 AD, Zorn GL. Coronary bypass surgery: analysis of factors affecting hospital mortaIity. CircuIation 1980;62:suppl I-64-1-89