Aspirin use and survival after coronary bypass surgery

Volume

123

Number

3

CABG and PTCA

15. Crea F, Kaski JC, Fragasso G, et al. Usefulness of Holter monitoring to improve the sensitivity of exercise testing in determining the degree of myocardial revascularisation after coronary artery bypass grafting for stable angina pectoris. Am J Cardiol 1987;60:40-3. 16. Droste C, Lemmen S, Nitsche R, Betz P, Birnbaum D, Roskamm H. ST segment monitoring before, three weeks and six months after aortocoronary bypass surgery. Eur Heart J 1988;9(suppl N):169-75. 17. Kennedy HL, Seiler SM, Sprague MK, et al. Relation of silent myocardial ischemia after coronary artery bypass grafting to angiographic completeness of revascularization and long-term prognosis. Am J Cardiol 1990;65:14-22. 18. Josephson MA, Nademanee K, Intarachot V, Lewis H, Singh BN. Abolition of Holter monitor detected silent myocardial ischemia after percutaneous transluminal coronary angioplasty. J Am Co11 Cardiol 1987;10:499-503. 19. Gohlke-Barwolf C, Droste C, Roskamm H. Silent myocardial ischemia after bypass surgery and percutaneous transluminal coronary angioplasty. In: Kellerman JJ, Braunwald E, eds. Silent myocardial ischemia: a critical appraisal. New York, Basel: S Karger, 1990:288-96.

Aspirin use and survival surgery

on total ischemic burden

20. Deligonul U, Vandormael MG, Younis LT, Chaitman BR. Prognostic significance of silent myocardial &hernia detected by early treadmill exercise after coronary angioplasty. Am J Cardiol 1989;64:1-5. 21. Gottlieb SO, Weisfeldt ML, Ouyang P, Mellits ED, Gerstenblith G. Silent ischemia as a marker for early unfavourable outcomes in patients with unstable angina. N Engl J Med 1986;314:1214. 22. Gottlieb SO, Weisfeldt ML, Ouyang P, Mellits ED, Gerstenblith G. Silent ischemia predicts infarction and death during 2-year follow-up of unstable angina. J Am Co11 Cardiol 1987; 10:756-60. 23. Tzivoni D, Gavish A, Zin D, et al. Prognostic significance of ischemic episodes in patients with previous myocardial infarction. Am J Cardiol 1988;62:661-4. 24. Gottlieb SO, Gottlieb SH, Achuff SC, et al. Silent ischemia on Holter monitoring predicts mortality in high-risk post-infarction patients. JAMA 1988;259:1030-5. 25. Nabel EG, Barry J, Rocco MB, et al. Variability of transient myocardial ischemia in ambulatory patients with coronary artery disease. Circulation 1988;78:60-7.

after coronary

bypass

From Jan. 1, 1966, through Dec. 31, 1961, 5618 patients in a single cardiovascular surgery practice underwent coronary bypass surgery. Detailed clinical information was obtained during hospitalization and entered into a data base. All surviving patients were surveyed for aspirin use four times: in 1984, 1985, 1986, and 1987. A subgroup of 2395 patients gave consistent snswers on every returned survey. The group that answered consistently “yes” had a 5-year survival rate of 79%; those who answered consistently “no” had a B-year survival rate of 67%. Stepwise logistic regression showed that, after adjusting for patient characteristics, the relative risk of death was 58% (confidence interval 47% to 70%) for consistent aspirin users compared with consistent non-aspirfn users. This study provides evidence that regular aspirin use after bypass surgery may enhance survival. (AM HEART J 1992;123:603.)

W. Dudley Johnson, MD, Kenneth L. Kayser, MS, Arthur J. Hartz, MD, PhD, and Saed F. Saedi, MD. Milwaukee, Wis.

Extensive recent research has dealt with the effect of aspirin on many aspects of cardiovascular disease. Most published studies show a reduction in cardiovascular deaths in subjects taking aspirin. Research From the Department of Cardiovascular Diseases,St. Mary’s Hospital, and the Department of Biostatistics and Epidemiology, Medical College of Wisconsin. Received for publication July 15, 1991; accepted Sept. 5, 1991. Reprint requests: W. Dudley Milwaukee, WI 53211. 4/l/34406

Johnson,

MD,

2315

N. Lake

Dr., Suite

1007,

reports and commentary articles number into the thousands. Review articles have been written on a wide variety of different aspects. Most studies of aspirin use after coronary bypass surgery have dealt with graft patency. l-21 An assortment of regimens, doses, and use of other agents has been reported. A few groups have reported negative results.4, 5, 12p16*2o All reports of no patency improvement used relatively high (975 to 1000 mg/day) doses of aspirin. Theoretically, higher doses of aspirin could be less effective than lower doses.22 603

604

Johnson

et al.

Table

I. Independent

American

variables

for logistic

regression Aspirin

Total subjects Gender Male Female Multiple surgeries Yes No Left ventricular dysfunction None Moderate Severe Unknown Age at survey (yr) 60 60-65 65-70 >70 Mammary graft Yes No Yr of surgery Before 1976 1976-1978 After 1978 Grafted system 1 2 3 Coronary endarterectomy Yes No *Univariate

categories

by aspirin

March 1992 Heart Journal

use

use (n)

No

Yes

% Yes

1399

996

41.6

1161 238

865 131

42.7 35.5

0.010

134 1265

121 875

47.5 40.9

0.044

642 400 227 130

476 306 129 85

42.6 43.3 36.2 39.5

0.116

450 299 314 336

382 222 227 165

45.9 42.6 42.0 36.2

<0.0001

689 710

517 479

42.9 40.3

0.200

462 439 498

265 325 406

36.4 42.5 44.9

0.0021

110 392 897

76 283 637

40.9 41.9 41.5

0.963

218 1181

180 816

45.2 40.9

0.107

p Value*

chi square, aspirin versus no aspirin.

Zichner and Weihrauch23 have reported on the use of aspirin for prevention of myocardial infarction after coronary bypass surgery. In a randomized trial of aspirin (990 mg/day) and dipyridamole, Gershlick et a1.4found no difference in survival (or anything else) a mean of 6.6 years after bypass surgery. Pfisterer et a1.2found that dipyridamole plus aspirin (50 mg/day) was as effective as anticoagulant therapy for preventing vein graft closure. Complications were less with antiplatelet therapy. They also found that therapy continued for 1 year was superior to therapy discontinued after 3 months. Lorenz et a1.14 reported improved patency rates 4 months after surgery in patients treated with 100 mglday aspirin. They also reported that this low dose of aspirin was effective in blocking thromboxane formation and thromboxanesupported aggregation on collagen. They observed no side effects throughout their trial. Investigators from the Second International Study of Infarct Survival (ISIS-2)24 studied the use of streptokinase and aspirin in 17,000 cases of suspected myocardial infarc-

tion. Both agents were given during the acute phase and aspirin was continued indefinitely. The ISIS-2 investigators concluded that the effects of aspirin and streptokinase were additive and “very significant” compared with placebo. The Antiplatelet Trialists’ Collaboration25 compiled the results of 25 completed randomized trials of antiplatelet treatment of “vascular disease.” The trialists concluded that allocation to antiplatelet treatment reduced vascular mortality rates by 15 % . The trialists were unable to detect a significant difference between different types and doses. We contrast the trialists’ studies of aspirin with the subject of this report in the Discussion section. The purpose of this study was to analyze the association of aspirin use on later (2 to 20 years) survival after coronary bypass surgery. In the 5-year period from Jan. 1,1984, through Dec. 31,1988, we observed aspirin use and survival for 3864 subjects who underwent coronary bypass surgery from Jan. 1, 1968, through Dec. 31, 1981.

Volume Number

123 3

Long-term

aspirin

and survival after CABG

605

Table II. Summary of stepwise logistic regression: 2395

consistent subjects Variable

Step 1 2 3 4 5 6 7 Aspirin *Other

Ventricular function Age at study start Aspirin use Mammary graft Yr of surgery Grafted systems Coronary endarterectomy last*

variables

Aspirin

use

Improvement chi square 111.78 41.42 33.12 18.31 14.84 11.71 7.15 30.80

Consistent

p Value
3 .L fi

.8

I=

forced, aspirin allowed to enter last.

+l S.E.

METHODS

All patients undergoing bypass surgery by Milwaukee Heart Surgery Associates(a private practice of cardiovascular surgery) are entered into the Milwaukee Heart Surgery Data base. This data base contains information on preoperative characteristics, details of the operation, recatheterization, postoperative symptoms,subsequentprocedures,and survival. Once entered, all patients are followed up for life. Surgical technique, myocardial preservation, and perfusion technique are reported elsewhere.25v 26 The study group was derived as follows: From Jan. 1, 1968,through Dec. 31,1981,5618 patients underwent coronary bypass surgery. Every patient who underwent at least one coronary bypass procedure of any type is included. Of these, 1296patients died before the observation period. Further excluded were 258 patients residing outsidethe United States and 52 patients who required valve replacementat the time of bypasssurgery.This leaves4012 potential study subjects. Each subject was surveyed four times or until death: 1984,1985,1986,and 1987.One hundred forty-eight (3.7%) patients were lost to or gave no follow-up responseto any survey, leaving 3864study subjects. Each year all patients were sent questionnairesby mail. Those not respondingto the first mailing were sent a second questionnaire. Those not responding to the second mailing were contacted by telephone. Each survey questionnaire asked, “Do you take aspirin regularly?” For every patient in the study, werecordedeither yes, no, or blank for each survey year. We alsorecorded all deaths that occurred in the study period (Jan. 1, 1984,through Dec. 31, 1988).All deaths were recorded, and we madeno attempt to separatecardiac or vascular from other causesof death. In any singlesurvey we wereunable to contact about 6 % of the subjects.These 6 % were then submitted to the National Death Index (NDI). Deaths reported from the ND1 wererecorded.If the ND1 searchfor a patient wasnegative, the patient was consideredalive on Dec. 31, 1988. If the ND1 report left someuncertainty as to the identity of a subject (e.g., slightly different birth date or different middle initial), we obtained a copy of the death certificate. In a separate check of our method we submitted 200 known deathsto the NDI; 198wereidentified correctly. Therefore we consider the ND1 check to be about 99% effective.

1984

1985 1986 1987 Year of Study

1988

Fig. 1. Actuarial survival rates for 1399 “consistent no aspirin” and 996 “consistent aspirin” responders.Consistent aspirin responders answered “yes” to the aspirin question on all returned questionnaires.Consistentno-aspirin respondersansweredsimilarly “no.” The ordinate is Jan. 1, 1984.

To ascertain whether reported aspirin usewas a significant predictor of survival, we performed stepwiselogistic regressions.The dependent variable was survival through the study period. The independent variables are listed in Table I. In the logistic regression,ageand year of surgery were treated as continuous variables and the rest were treated as categorical, as shownin Table I. Consistent subgroup. We first analyzed the 2395subjects who answeredconsistently “yes” or consistently “no” to the aspirin question, hereafter referred to asthe consistent subgroup. The “consistent yes” subjects answered “yes” to all returned questionnaires.The “consistent no” subjectssimilarly answered“no.” For these 2395subjects our best estimate is that they were either taking or not taking aspirin throughout the period of observation. For this group we plotted the actuarial survival curves and performed astepwiselogisticregressionfor survival through the 5-year study period. To evaluate population differences between the aspirin and no-aspirin groups, we performed the regressionwithout aspirin in the model. We then computed the expected survival in each group with these regressioncoefficients. This gives the expected survival basedon all significant predictors except aspirin. Entire study group, separate surveys. For the entire study group of 3864patients, we did the regressionon each survey separately and calculated one combinedlogistic regressioncoefficient and its standard error for the four surveys. Clinical material and follow-up. As defined above, the study group has 3864subjects.At the start of the observation period, 382(9.9%) subjectshad undergoneat leasttwo coronary bypasssurgeries.The averageagewas62.4 years;

March

606

Johnson et al.

Table

HI. Results of the logistic regression

American

Patient

Logistic coefficient

characteristic Ventricular dysfunction None Moderate Severe Unknown Age at start of study, continuous Mammary graft No Yes Yr of surgery, continuous No. of grafted systems 1 2 3 Coronary endarterectomy No Yes Aspirin use No Yes Constant *95% Confidence interval. tWald statistic. $Relative risk for a lo-year $Relative risk for a 5-year

age increment. date of surgery

Upper bound*

Lower bound*

p Value7
0 0.2549 1.1756 0.1804 0.0279

1.00

1.29 3.24 1.19 1.323

1.03 2.49 0.76 1.25

1.62 4.22 1.88 1.40

1.00 0.57 0.54s

0.47 0.49

0.70 0.62

0 0.1354 0.4542

1.00 1.15 1.57

0.77 1.71

1.69 2.32

0.006 0.500 0.021

0 0.3882

1.00 1.47

1.06

1.86

0.007

1.00 0.58

0.47

0.70


0 -0.5557 -0.1202

0 -0.5493 6.304

0.029
0.433


increment.

593 (15.5%) subjectswere women. Ventricular function is not known for the first 338 patients operated on; of the remaining 3526,553(15.7%) had severeventricular dysfunction at the time of surgery. Grafting to all three vesselsystems was required in 2498 (64.4%) patients, and 2044 (52.9%) underwent a mammary bypass. The average interval between surgery and the start of the observation period was6.7 years. All referencesto surgery date refer to the first surgery by us.At the end of the observation period the survival status was known, with certainty, for all but four (0.1% ) subjects;857 (22.2% ) subjectsdied during the study period. RESULTS Consistent

Retative risk

1992

Heart Journal

subgroup. Fig. 1 shows actuarial survival rates for the two subgroups. As stated previously, some of the difference in the curves of Fig. 1 can be attributed to differences in age and date of surgery. To assessthe effects of all significant variables except aspirin, we performed a logistic regression similar to the one reported below except that all significant variables except aspirin were forced into the model. In this logistic model the no-aspirin group had a predicted probability of death of 0.2949; in the aspirin group it was 0.2686. Therefore about three percentage points of the difference demonstrated by Fig. 1 can be attributed to differences in predictors other than aspirin use, particularly age and date of surgery. Table I shows the variables tested for inclusion in the logistic model and the breakdown by aspirin use.

Categories are obvious except for the following: left ventricular dysfunction: moderate, distinct contraction abnormality of one major wall or sluggish motion in two major walls; and severe, distinct contraction abnormality in two or more major walls or sluggish motion throughout the ventricle. All judgments were made from the right anterior oblique, contrast-injection ventriculogram. It is noteworthy that men, patients who underwent multiple surgeries, patients under 60 years, and subjects operated on after 1978 were more likely to have taken aspirin. Women, patients who underwent one operation, subjects over 70 years, and patients operated on before 1970 were less likely to have taken aspirin. Tables II and III show a summary of the forward stepwise logistic regression analysis, with aspirin use defined either “consistent yes” or “consistent no.” Variables are the same as in Table I except that age and year of surgery were treated as continuous variables. In the regression, aspirin use was the third variable to enter the model. The logistic coefficient for aspirin was -0.55, SE was 0.10, and p < 0.000001 (approximate t test). The relative risk of death (adjusted for all listed prognostic factors) for subjects taking aspirin was 0.58 (95 % confidence limits 0.47 to 0.70). The improvement chi square with aspirin entered last was 30.80 (p < 0.000001). If the regression is run with age and year of surgery as categoric vari-

Volume

123

Number

3

ables, the results are essentially the same except that aspirin enters the model second. Entire study group, separate surveys. Table IV shows the total responses to and deaths after each individual survey. Death in a survey period was defined as having occurred after the survey was answered but before the next survey was answered. Overall aspirin use increased from 33 % in the 1984 survey to 61% in the 1987 survey. In every survey period, subjects taking aspirin had a lower percentage of deaths. Table V shows the summary of four separate regressions. In each regression, aspirin use was detined as the answer for that year’s survey. The four separate regressions were then combined by calculating the weighted mean logistic coefficient and pooled variance for the four analyses. The weighted mean coefficient is 0.2486, SE is 0.0752, and p < 0.001. This corresponds to a relative odds for aspirin users (adjusted for all listed prognostic factors) of 0.78. This risk is higher than the risk for consistent aspirin users versus consistent nonusers.

Long-term

aspirin

and survival after CABG

607

Table IV. Mortality rate and aspirin useby year: four surveys, 3864subjects Aspirin

use

No (n/%) Yes (n/%3) % Yes Survey yr nl%, Number

Survey 252118.7 124216.7 33.0 1984

response 178716.4 1558h.5 46.6 1985

of respondents/percent

and mortality 146718.2 170015.2 53.7 1986

1102/6.2 172514.5 61.0 1987

mortality.

V. Resultsof the logistic analysisby year for all 3864 study subjects Table

Survey 1984 1985 1986 1987 All surveys

Coefficient -0.2214 -0.1054 -0.4040 -0.2674 -0.2487

*

SE* 0.1364 0.1506 0.1472 0.1746 0.0752

Coefficient/ SE 1.62 0.70 2.74 1.53 3.31

p Value j 0.15 0.40 0.01 0.15
DISCUSSION

*Logistic coefficient for aspirin use and its SE. tApproximate t test.

The subjects of this study had confirmed heart diseaseto the extent that bypass surgery was deemed appropriate. We observed them for a 5-year period after bypass surgery. All study subjects were operated on from 1968 through 1981 and observed from 1984 through 1988. Thus the observations were made 2 to 20 years after surgery. During this time, subjects who took aspirin regularly had significantly enhanced survival. We also found that voluntary aspirin use increased during the study period. It was disappointing to find that fewer older subjects took aspirin. Older patients are more likely to suffer stroke, and aspirin use might reduce the incidence. The Antiplatelet Trialists’ Study27 summarized the results of 25 randomized trials. Of these, 10 trials were in patients with myocardial infarction; eight of these trials used aspirin. There were two trials in patients with unstable angina; both used aspirin. This gives 10 trials 28-37of aspirin used in treating patients with known heart disease. In these 10 trials the treatment groups had 40.5 fewer vascular deaths and 7.9 more nonvascular deaths than would be expected by pure chance. Combining all deaths gives 38.8 fewer deaths in the treatment group. The authors compute an odds ratio of 0.89, a risk reduction of 11IL for the combined treatment groups. In the subject study of this article the odds ratio (calculated from the regression coefficient for aspirin, thus corrected for all listed prognostic factors) for the consistent subgroup was 0.58, a risk reduction of 42%; for the entire group it was 0.78, a risk reduction of 22 % . This is a nonrandomized study. We cannot exclude the possibility that there exist nonrecorded variables

that could bias the observations. Are patients who take aspirin more health conscious? Do they smoke less, eat better, or exercise more? Are patients with symptoms more likely to take aspirin? In addition to the aspirin question, the surveys did ask about weight, smoking, and angina. An analysis of these data failed to show any association of smoking or weight with aspirin use. In the last survey only, patients with angina were slightly (p = 0.01) more likely to take aspirin. This study is strongly supportive of previously reported studies showing the benefit of aspirin. The degree of benefit associated with aspirin is higher in this study than in any other. This may be because all the patients in this study had documented, severe coronary artery disease. The time interval for follow-up after coronary artery bypass surgery is also longer than in other studies. Because vein grafts are known to develop accelerated atherosclerosis, perhaps aspirin therapy is more effective for vein grafts than for native arteries. Aspirin is extremely low cost with very few undesirable side effects. A risk reduction of 42 % with such simple therapy is really quite remarkable. Conclusions. Among the patients in this study (1) use of aspirin after bypass surgery is increasing, (2) older patients and those operated on earlier were less likely to use aspirin, (3) aspirin use after surgery was a significant predictor of survival after bypass surgery, and (4) differences in survival between aspirin users and nonusers cannot be attributed to differ-

608

Johnson

et al.

ences in the analyzed characteristics groups.

American

of the sub19.

REFERENCES

S, Copeland J, Moritz T, et al. Saphenous vein graft 1. Goldman patency one year after coronary artery bypass surgery and effects of antiplatelet therapy: results of a Veterans Administration Cooperative Study. Circulation 1989;80:1190-7. 2. Pfisterer M, Burkart F, Jockers G, et al. Trial of low-dose aspirin plus dipyridamole versus anticoagulants for prevention of aortocoronary vein graft occlusion. Lancet 1989;2:1-7. 3. Guiteras P, Altimiras J, Aris A, et al. Prevention of aortocoronary vein-graft attrition with low-dose aspirin and triflusal, both associated with dipyridamole: a randomized, double blind, placebo-controlled trial. Eur Heart J 1989;2:159-67. 4. Gershlick AH, Lyons JP, Wright JE, Sturridge MF, Layton CA, Balcon R. Long-term clinical outcome of coronary surgery and assessment of the henefit obtained with postoperative aspirin and dipyridamole. Br Heart J 1988;60:111-6. 5. Thaulow E, Frysaker T, Dale J, Vatne K. Failure of combined acetylsalicylic acid and dipyridamole to prevent occlusion of aortocoronary venous bypass graft. Stand J Thorac Cardiovast Surg 1987;21:215-201 6. Cairns JA. Clinical trials of antinlatelet drug theranv in acute myocardial infarction, unstable angina, and aortocoronary bypass surgery. Cardiovasc Clin 1987;18:231-46. JH, Fuster V. Pathogenesis and prevention of aor7. Chesebro tocoronary bypass graft occlusion. Int J Clin Pharmacol Res 1986;6:261-7. 8. Pirk J, Vojacek J, Kovac J, Fabian J, Firt P. Improved patency of the aortocoronarv bypass by antithrombotic drugs. Ann Thorac Surg 1986;42:3I214. 9. Verstraete M. Brown BG. Chesebro JH. Ekestrom S. Harker LA, Henderson AH, Jewitt DE, Oliver MF, Sleight P. Evaluation of antiplatelet agents in the prevention of aorto-coronary bypass occlusion. Eur Heart J 1986;7:4-13. 10. Rajah SM, Salter MC, Donaldson DR, Subba Rao R, Boyle RM, Partridge JB, Watson DA. Acetylsalicylic acid and dipyridamole improve the early patency of aortocoronary bypass grafts: a double-blind. nlacebo-controlled, randomized trial. J Thorac Cardiovasc S&g 1985;90:373-7. 11. Brown BG, Cukingnan RA, DeRouen T, et al. Improved graft patency in patients treated with platelet-inhibiting therapy after coronary bypass surgery. Circulation 1985;72:138-46. 12. Brooks N, Wright J, Sturridge M, et al. Randomised placebocontrolled trial of aspirin and dipyridamole in the prevention of coronary vein graft occlusion. Br Heart J 1985;53:201-7. 13. Meister W, von Schacky C, Weber M, et al. Low-dose acetylsalicylic acid (100 mg/day) after aortocoronary bypass surgery: a placebo-controlled trial. Br J Clin Pharmacol 1984;17:70311. 14. Lorenz RL, Schacky CV, Weber M, et al. Improved aortocoronary bypass patency by low-dose aspirin (100 mg daily): effects on platelet aggregation and thromboxane formation. Lancet 1984;1:1261-4. 15. Chesebro JH, Fuster V, Elveback LR, et al. Effect of dipyridamole and aspirin on late vein-graft patency after coronary bvnass onerations. N Engl J Med 1984:310:209-14. 16. S&ma GV, Khuri SF, Rosa M, Folland ED, Parisi AF. The effect of antiplatelet therapy on saphenous vein coronary artery bypass graft patency. Circulation 1983;68(suppl):II21821. 17. Chesebro JH, Clements IP, Fuster V, et al. A platelet-inhibitor-drug trial in coronary-artery bypass operations: benefit of perioperative dipyridamole and aspirin therapy on early postoperative vein-graft patency. N Engl J Med 1982;307:73-8. acetylsal18. Weber M, von Schacky C, Lorenz R, et al. Low-dose

20.

21.

22. 23.

24.

25.

26.

March 1992 Heari Journal

icylic acid (100 mg/day) following aortocoronary bypass operation. Klin Wochenschr 1984;62:458-64. Mayer JE Jr, Lindsay WG, Castaneda W, Nicoloff DM. Influence of aspirin and dipyridamole on patency of coronary artery bypass grafts. Ann Thorac Surg 1981;31:204-10. Pantely GA, Goodnight SH Jr, Rahimtoola SH, Harlan BJ, DeMots H, Calvin L, Rosch J. Failure of antiplatelet and anticoagulant therapy to improve patency of grafts after coronary-artery bypass: a controlled, randomized study. N Engl J Med 1979;301:962-6. McEnany MT, Salzman EW, Mundth ED, et al. The effect of antithrombotic therapy on patency rates of saphenous vein coronary artery bypass grafts. J Thorac Cardiovasc Surg 1982;83:81-9. Klotz L. Antiplatelet and anticoagulant therapy after coronary bypass (letter). N Engl J Med. 1980;302:866. Zichner R, Weihrauch TR. Optimal dosage of acetylsalicylic acid. I: Use in primary and secondary prevention of myocardial infarct and following aortocoronary bypass operation. Med Klin 1989;84:43-51. ISIS-2 (Second International Study of Infarct Survival) Collaborative Group. Randomised trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction: ISIS-2. Lancet 1988;2:34960. Johnson WD, Brenowitz JB, Kayser KL. Factors influencing long-term (ten-year to fifteen-year) survival after a successful coronary bypass operation. Ann Thorac Surg 1989;48:19-24. Johnson WD, Kayser KL, Pedraza PM. Angina pectoris and coronary bypass surgery: patterns of prevalence and recurrence in 3105 consecutive patients followed up to eleven years.

AM HEARTJ 1984;108:1190-7. 27. Antiplatelet Trialists’ Collaboration. Secondary prevention of vascular disease by prolonged antiplatelet treatment. Br Med J 1988,296:320-31. 28. Klimt CR, Knatterud GL, Stamler J, Meier P. Persantine-aspirin reinfarction study. Part II. Secondary coronary prevention with Persantine and aspirin. J Am Co11 Cardiol1986;7:5169. 29. Lewis HD Jr, Davis JW, Archibald DG, et al. Protective effects of aspirin against acute myocardial infarction and death in men with unstable angina: results of a Veterans Administration Cooperative Study. N Engl J Med 1983;309:396-403. 30. Cairns JA, Gent M, Singer J, et al. Aspirin, sulfinpyrazone, or both in unstable angina: results of a Canadian multicenter trial. N Engl J Med 1985;313:1369-75. 31. Aspirin Myocardial Infarction Study Research Group. A randomized controlled trial of aspirin in persons recovered from myocardial infarction. JAMA 1980;243:661-9. 32. Persantine-Aspirin Reinfarction Study Research Group. Persantine and aspirin in coronary heart disease. Circulation 1980;62:449-61. 33. Elwood PC, Sweetnam PM. Aspirin and secondary mortality after myocardial Lancet - 1979;11:1313-5. -. infarction.34. Coronary Drug Project Research tiroup. Aspirin in coronary heart disease. J Chron Dis 1976;29:625-42. 35. Vogel G, Fischer C, Huyke R. Prevention of reinfarction with acetylsalicylic acid. In: Breddin K, Leow D, Ueberla K, Dorndorf W, Marx R, eds. Prophylaxis of venous, peripheral, cardiac and cerebral vascular diseases with acetylsalicylic acid. Stuggart: Schattauer Verlag, 1981:123-8. 36. Elwood PC. Trial of acetylsalicylic acid in the secondary prevention of mortality from myocardial infarction [Letter]. Br Med J 1981;282:481. 37. Breddin K, Loew D, Lechner K, Ueberla K, Walter E. Secondary prevention of myocardial reinfarction: a comparison of acetylsalicylic acid, placebo and phenprocoumon. Haemostasis 1980;9:325-44.