- Email: [email protected]
The Association of Stroke and Coronary Heart Disease: A Population Study
mj Vol. 62
y& Proceedings ROCHESTER, MINNESOTA
DECEMBER 1987
The Association of Stroke and Coronary Heart Disease: A Population Study
DONN D. DEXTER, Jr., M.D., Department of Neurology; JACK P. WHISNANT, M.D., Departments of Neurology and Health Sciences Research; DANIEL C. CONNOLLY, M.D., Division of Cardiovascular Diseases and Internal Medicine; W. MICHAEL O'FALLON, Ph.D., Section of Biostatistics In a 20-year population-based study (1960 through 1979), we estimated the cumulative probability of (1) the occurrence of stroke after a diagnosis of angina pectoris, (2) the occurrence of stroke after a diagnosis of myocardial infarction, and (3) the occurrence of myocardial infarction or sudden unexpected death after a diagnosis of cerebral infarction. In patients in whom angina had been diagnosed, no significant difference was noted between the observed and the expected probability of stroke throughout 10 years of follow-up. In patients with a diagnosis of myocardial infarction, a significant difference was noted between observed and expected probabilities of stroke at 1 month and at 2 months. This early excess in probability of stroke was especially pronounced in the subgroup of patients with transmural myocardial infarction but not evident in those with subendocardial myocardial infarction. Among patients with a diagnosis of cerebral infarction, the difference between observed and expected probabilities of myocardial infarction or sudden unexpected death was not significant until 5 years after the cerebral infarction and showed no change thereafter. We conducted this study to determine whether, in a defined population, an excess risk of having a stroke was present after the onset of coronary heart disease or an excess risk of myocardial infarction (MI) or sudden unexpected death ex isted after a first cerebral infarct. The Mayo Clinic medical record index and record linkage system has been used to study stroke and cor-
This study was conducted in the Mayo Cerebrovascular Re search Center and was supported in part by Grants AM 30582, HL 24326, and NS 06663 from the National Institutes of Health, Public Health Service. Address reprint requests to Dr. J. P. Whisnant, Department of Neurology, Mayo Clinic, Rochester, MN 55905. Mayo Clin Proc 62:1077-1083, 1987
onary heart disease over several decades in the population of Rochester, Minnesota. 1 5 Existing data sets were used in this study to estimate (1) the probability of stroke after the diagnosis of angina pectoris, (2) the probability of stroke after the diagnosis of MI, and (3) the probability of MI or sudden unexpected death after the diagnosis of cerebral infarction. Life-table analyses were used and in each instance were compared with expected rates for the Rochester population by using subjects of the same age and sex as the patients. BACKGROUND Little has been published about the association of angina pectoris and stroke, but a population1077
1078
STROKE AND CORONARY HEART DISEASE
based study of transient cerebral ischemic at tacks revealed no evidence that previous angina pectoris, MI, valvular heart disease, or cardiac arrhythmia, individually or in combination, in fluenced the probability of the occurrence of stroke in patients with transient ischemic attacks. 6 Interest in the occurrence of stroke after MI was evident in the medical literature 5 decades ago and is still a subject of continuing investi gations. 7 " 21 In autopsy studies, the frequency of stroke associated with acute MI was reported to range from 1.2 to 9%.7"9 Left ventricular mural thrombi were found at autopsy in 37% of patients who died of acute MI.8 In another study, 35% of patients with left ventricular mural thrombi had a pathologic diagnosis of systemic arterial occlu sion, and almost half of these presumably embolic lesions were found in the cerebral circula tion. 9 In more recent prospective clinical studies in which follow-up of patients was conducted for up to 6 weeks after acute MI, the frequency of stroke reportedly ranged from 1 to 8.6%.10"12 Cardiogenic emboli have also been reported as an important cause of stroke after acute MI in several clinical studies.10"13'18·20'21 A review of the literature regarding stroke as a late complication of MI revealed that five studies in which 2,662 patients who had had MI and were under surveillance for 2 to 10 years reported no mention of stroke as a complication, whereas in five other studies involving 1,094 such patients, stroke was listed as the cause of 1 to 8% of the deaths. A 5-year follow-up of 260 pa tients with MI who had been admitted to a cor onary care unit and were subsequently dismissed from the hospital noted a 4% frequency of stroke plus a 2% frequency of transient ischemic attacks detected at the time of assessment 5 years after hospital dismissal. 13 An autopsy series of 117 patients with healed (older than 6 weeks) MI disclosed left ventricular thrombi in 24% and systemic arterial occlusion in 26% of patients with mural thrombi in the left ventricle. The frequency of stroke was 17% overall in the 117 patients with healed MI. The investigators noted, however, that cerebral infarction was equally as likely to be found in patients without as in those with left ventricular mural thrombi, a finding that suggests the decreasing importance of em bolism and the increasing prominence of shared risk factors in the causation of stroke as the time after acute MI increased. 8,9
Mayo Clin Proc, December 1987, Vol 62
In most reported cases in which the onset of cardiovascular and cerebrovascular disease could be dated, the cardiac problem occurred first.14 Several published reports have described concur rent myocardial and cerebral infarctions, but the temporal relationship between the two conditions has often been undetermined. 15 " 20 In one study of concurrent myocardial and cerebral infarction, two of nine patients probably had a stroke first and MI shortly thereafter, and the importance of shared risk factors relating these two conditions was noted. 17 METHODS The unique capability for population-based epidemiologic research in Rochester, Minnesota, is facilitated by the detailed medical records avail able for the residents of this city. The Mayo Clinic has served as the major health-care provider for local residents since the turn of the century, and an index of diagnoses and surgical procedures for all medical contacts is maintained in the medical record of each patient. A central file of medical records containing an index of diagnoses and surgical procedures entered by other health-care providers for local residents is indexed to the Mayo record system for retrieval. The result is a comprehensive record of the health care re ceived by the population of Rochester, Minnesota. The population of Rochester, Minnesota, was 39,012 in 1960, 52,629 in 1970, and 56,447 in 1980 as determined by census statistics. More than 99% of the population in this community are Caucasian. The Rochester population is some what younger and has a slightly higher mean income and level of education than the US white population in general. 1 The autopsy rate for deaths among Rochester residents has been ap proximately 60%. This system of medical records has been used to study the incidence of coronary heart disease in residents of Rochester, Minnesota, for the period Jan. 1, 1950, through Dec. 31, 1979.1"3 The diagnosis of coronary heart disease was accorded to three major initial manifestations: angina pec toris, MI, and sudden unexpected death. In ad dition to the strict diagnostic criteria, patients must have been a resident of Rochester for at least 1 year before the initial diagnosis of cor onary heart disease (in order to exclude persons who might have moved to Rochester because of their disease). The patient must also have had
Mayo Clin Proc, December 1987, Vol 62
no prior diagnosis of coronary heart disease and no clinical evidence of congestive heart failure, congenital or rheumatic heart disease, noncoronary cardiomyopathy, or valvular heart disease before inclusion in the study. A continuing study of the incidence of stroke in Rochester, Minnesota, for the period Jan. 1, 1945, through Dec. 31, 1979, provided the infor mation about the patients with stroke in this report.4,5,22 During this period, 2,133 new cases of stroke occurred during 1.5 million person-years of observation in residents of Rochester. The neurologic history and examination record have been standardized for much of the 70-year period that the Department of Neurology at the Mayo Clinic has been in existence. This standardiza tion, as well as the high autopsy rate (58% of patients in the community who died of stroke from 1970 through 1974 had a complete post mortem evaluation), adds to the reliability of the neurologic observations and categorization of stroke. Strokes have been categorized as cerebral and cerebellar or brainstem arterial thrombosis, cerebral and cerebellar or brainstem embolus, intracerebral and intracerebellar or brainstem hemorrhage, primary subarachnoid hemorrhage, or type unknown. The criteria used for the var ious diagnostic categories of stroke have been published previously. 4 In the current study, we used data gathered and computerized for the aforementioned studies to analyze the association of coronary heart disease and stroke. During the 20-year period from Jan. 1, 1960, through Dec. 31, 1979, 2,622 patients fulfilled the diagnostic criteria for the coronary heart disease study—1,015 patients with an ini tial diagnosis of angina, 1,183 with an initial diagnosis of MI or concurrent angina and MI, and 424 with an initial diagnosis of sudden un expected death. Among the patients having an initial diagnosis of angina, 319 cases had sub sequent MI. These cases were combined with the patients who had an initial diagnosis of MI or concurrent angina and MI; thus, the total number of patients with MI was 1,502. The patients with MI were also divided into the subgroups of transmural and subendocardial (also called nontransmural or non-Q-wave) MI. Patients who had a stroke before the initial diagnosis of angina or MI were excluded from the analysis; the resulting groups were 991 patients with an initial diag nosis of angina, 1,127 with an initial diagnosis
STROKE AND CORONARY HEART DISEASE
1079
of MI or concurrent angina and MI, and 1,419 in the combined MI group. There were 1,253 patients from the stroke study in the period covered in this report. This number included 1,043 patients with cerebral infarction or stroke of uncertain type, which is considered primarily infarction. Stroke occurring after the diagnosis of MI included all cases of cerebral infarction, stroke of uncertain type, and intra cerebral hemorrhage but did not include patients with subarachnoid hemorrhage. The patients with stroke who were under surveillance for sub sequent MI included only those 943 with cerebral infarction or stroke of uncertain type who did not have a MI before a stroke. In seven patients, the stroke and the MI oc curred on the same day, and which event occurred first could not be determined from the medical record. These patients were excluded from the subsequent analyses. Because both data sets included all cases from 1950 through 1979 and the incidence cases were selected beginning in 1960, detection of the cases to be excluded—because of prior stroke in pa tients with angina or MI or because of prior MI in patients with cerebral infarction—should be almost complete. The Kaplan-Meier life-table method 23 was used to estimate (1) the cumulative probability of stroke after a diagnosis of angina pectoris by using the date of diagnosis of angina as the beginning date, the date of stroke as the event date, and the date of last follow-up or of death from a cause other than stroke as the withdrawal date; (2) the cumulative probability of stroke after a diagnosis of MI by using the date of diagnosis of MI as the beginning date, the date of stroke as the event date, and the date of last follow-up or of death from a cause other than stroke as the withdrawal date; and (3) the cumulative prob ability of MI or sudden unexpected death after a cerebral infarction by using the date of cerebral infarction as the beginning date, the date of MI or sudden unexpected death as the event date, and the date of last follow-up or of death from any other cause as the withdrawal date. Separate analyses were performed on the entire group of patients with MI as well as the subgroups of transmural (N = 782) and subendocardial (N = 404) MI. The incidence rates for stroke and MI in age- and sex-matched subjects in the Rochester population were used to determine the expected cumulative
1080
STROKE AND CORONARY HEART DISEASE
probability of stroke or MI and sudden unex pected death for each of the aforementioned situations. Observed and expected rates were compared by using a one-sample log-rank test. RESULTS During the period studied, 1,015 patients had an initial diagnosis of angina pectoris. The 24 pa tients (2.4%) who had a stroke before the initial diagnosis of angina were excluded. In 53 pa tients, stroke occurred after the initial diagnosis of angina pectoris during the period of study. The mean age of patients with angina without prior stroke was 64 years (60 years for men and 67 years for women). The observed 10-year cumu lative probability of the occurrence of stroke in patients after the initial diagnosis of angina was 5.8%. No significant difference was noted be tween the observed and the expected cumula tive probability of stroke throughout 10 years of follow-up (Fig. 1). During the period studied, 1,502 patients had the diagnosis of MI, concurrent angina and MI, or angina followed by MI. In 80 of these patients (5.3%), a history of stroke before the initial diag-
Fig. 1. The cumulative probability of stroke 5 after initial diagnosis of angina pectoris (N = 991), given survival. Ex pected cumulative probability of stroke is for an age- and sexmatched population and is based on incidence rates for stroke in Rochester, Minnesota, for the period 1960 through 1979.22 No significant difference was noted between observed and expected cumulative probabilities of stroke throughout 10 years of follow-up.
Mayo Clin Proc, December 1987, Vol 62
nosis of MI excluded them from further analysis. In 74 patients, stroke occurred after the initial diagnosis of MI. The mean age of patients with out prior stroke was 65 years (62 years for men and 70 years for women). The curves depicting the observed and the expected probability of stroke after MI (Fig. 2) were compared by using the log-rank test, and the difference in the curves was significant (P<0.01). The difference between the observed (0.9%) and the expected (0.05%) prob ability of stroke was significant at 1 month (P<0.01) and also at 2 months (1.25% versus 0.1%; P<0.001). No further change in the difference was noted through the first year of observation (Fig. 2). This early excess in the probability of stroke was also noted—and to a greater extent— in the subgroup of patients with transmural MI, in whom the incidence of stroke at 2 months was increased by a factor of 15 over that of the expected incidence (Fig. 2). No significant differ ence was observed in the cumulative probability of stroke after subendocardial MI in comparison with the expected incidence at any time during the first year of follow-up (Fig. 2) or during the subsequent 9 years. Cerebral infarction was diagnosed in 1,044 pa tients during the period studied. The 93 patients (8.9%o) who had a MI before the initial diagnosis of cerebral infarction were excluded from further study. MI or sudden unexpected death occurred in 90 patients at the time of or after the initial diagnosis of cerebral infarction during the years of follow-up. The mean age of those patients without prior MI was 73 years (70 years for men and 75 years for women). The cumulative prob ability of MI or sudden unexpected death after cerebral infarction was significantly increased in comparison with the expected probability throughout 10 years of follow-up (Fig. 3) (log-rank test, P<0.01). The observed and the expected cumulative incidences for MI or sudden unex pected death after cerebral infarction were 0.8% and 0.5% at 6 months, 1.4% and 1.0% at 12 months, 10.6% and 5.4% at 5 years (P<0.001), and 16.6% and 11.0% at 10 years (P<0.01). No signif icant difference was noted between the observed and expected probabilities of MI or sudden un expected death until 5 years after the first cere bral infarction; however, the curves for observed versus expected probabilities were divergent after the first 6 months and continued in this pattern through 5 years.
Mayo Clin Proc, December 1987, Vol 62
io
—
STROKE AND CORONARY HEART DISEASE
Expected SMI
8
Ml —
TMI
6
% 4
2
0 C
1
2
3
4
5
6
7
8
9
10
I
I
11
12
Months after first diagnosis of Ml
Fig. 2. The cumulative probability of stroke after initial diagnosis of myocardial infarction (MI) (N = 1,419) and transmural (TMI) (N = 782) and subendocardial infarction (SMI) (N = 404), given survival. Expected cumulative prob ability of stroke is for a population that was matched for age and sex (to the patients with MI) and is based on incidence rates for stroke in Rochester, Minnesota, for the period 1960 through 1979.22 A significant difference was noted between observed and expected probabilities of stroke at 1 month and at 2 months; this early excess in probability of stroke was especially pronounced in subgroup of patients with TMI.
DISCUSSION For the time period studied (1960 through 1979), this study of the Rochester, Minnesota, popula tion included essentially all persons with onset of angina pectoris and all those with a first MI who had not had a prior stroke; it also included essentially all persons with a first cerebral infarc tion who had not had a prior MI. For the assess ment of the probability of the occurrence of stroke after the onset of angina pectoris, 24 persons who had a stroke before the diagnosis of angina pec toris were excluded from analysis, and 80 persons who had a stroke before the onset of MI were excluded from the analysis of the probability of the occurrence of stroke after MI. In addition, 93 patients were excluded from the analysis of prob ability of MI or sudden unexpected death after a first cerebral infarction because those patients had MI before the cerebral infarct. In seven patients, we could not determine with certainty whether the MI or the cerebral infarc tion occurred first. The timing of the detection of the serum enzyme abnormalities suggested, however, that in most, if not all, of these patients the MI occurred first. Because of the uncertainty, we thought that the interpretation of the follow-
1081
up data would be less biased if these seven cases were not included in the life-table analyses. Patients with the diagnosis of angina pectoris had no evidence of an increased risk of stroke either soon after the diagnosis or at any time during the next 10 years. Those patients who had a diagnosis of MI had a significant increase in the probability of stroke. This increased risk was most notable at 1 month, at which time the risk was increased by a factor of 18, and it was still increased by a factor of 13 at 2 months (Fig. 2). When a person-years analysis was performed after the first 2 months, no significant difference was found in the ob served and the expected number of strokes except in one of the remaining months during the first year of observation. The increase in risk of stroke during the first 2 months after MI was even more evident among patients with transmural MI than among the overall MI group (Fig. 2). No signif icant difference was noted, however, between the observed and expected number of strokes at any point in the life-table analysis for patients with subendocardial MI (Fig. 2). These observations suggest that a direct caus ative factor relating to stroke is present soon after
Fig. 3. The cumulative probability of myocardial infarction or sudden unexpected death after initial diagnosis of cerebral infarction (N = 943), given survival. Expected cumulative probability of myocardial infarction or sudden unexpected death is for an age- and sex-matched population and is based on incidence rates for myocardial infarction and sudden unexpected death in Rochester, Minnesota, for the period 1960 through 1979.2 No significant difference was noted between observed and expected probabilities of myocardial infarction or sudden unexpected death until 5 years after the diagnosis of cerebral infarction.
1082
STROKE AND CORONARY HEART DISEASE
the onset of MI but that it is not operative beyond the first 2 months after MI. Presumably, this factor is intracardiac mural thrombi that produce embolization. In this study, we do not have in formation about treatment with anticoagulants or other drugs from which a differential outcome might be judged. The results noted apply to all patients without regard to treatment. It is evi dent, however, that if treatment to prevent stroke is to be effective it must have its benefit during the first 2 months after MI, and there is little likelihood of benefit thereafter. Among the patients with the first cerebral in farction, only one clear case of subsequent MI occurred during the first month of follow-up. That patient had an anterior MI and cardiac arrest 4 days after the cerebral infarction. The difference between the observed and the expected prob ability of MI or sudden unexpected death after cerebral infarction was not significant at any observational point until 5 years after the onset of cerebral infarction. At 5 years, the risk of MI or sudden unexpected death was increased by a factor of 2, and the difference was significant (P<0.001) (Fig. 3). In only 1 of the next 5 years was the observed number of cases of MI or sud den unexpected death significantly greater than the expected number. This lack of difference from 5 to 10 years is apparent in the plotted curves (Fig. 3). The differences noted at 5 years do not consider risk factors other than the cerebral in farction that might have affected the occurrence of MI or sudden unexpected death. CONCLUSION In our study population, no significant difference was noted between the observed and the expected cumulative incidence of stroke in patients after the initial diagnosis of angina pectoris. We noted a significant increase in the cumula tive incidence of stroke at 1 and 2 months after MI but no change in this difference after 2 months, an indication that no further increase in risk existed after 2 months. The early excess of stroke was evident in the subgroup of patients with transmural MI but not in those with subendocardial MI. These observations suggest a causative factor related to embolization from intracardiac mural thrombi in patients with transmural MI for 2 months after the diagnosis of MI but not thereafter.
Mayo Clin Proc, December 1987, Vol 62
A significant increase was found in the cumula tive incidence of MI or sudden unexpected death after cerebral infarction at 5 years but no change in the difference after that time. The difference between the observed and expected probabilities of MI and sudden unexpected death gradually increased until 5 years after the first cerebral infarct; thus, the differences were likely due to shared risk factors for myocardial and cerebral infarction rather than a direct effect of cerebral infarction.
REFERENCES 1. Connolly DC, Oxman HA, Nobrega FT, Kurland LT, Kennedy MA, Elveback LR: Coronary heart disease in residents of Rochester, Minnesota, 1950-1975. I. Back ground and study design. Mayo Clin Proc 56:661-664, 1981 2. Elveback LR, Connolly DC, Kurland LT: Coronary heart disease in residents of Rochester, Minnesota. II. Mor tality, incidence, and survivorship, 1950-1975. Mayo Clin Proc 56:665-672, 1981 3. Connolly .DC, Elveback LR: Coronary heart disease in residents of Rochester, Minnesota. VI. Hospital and posthospital course of patients with transmural and subendocardial myocardial infarction. Mayo Clin Proc 60:375-381,1985 4. Whisnant JP, Fitzgibbons JP, Kurland LT, Sayre GP: Natural history of stroke in Rochester, Minnesota, 1945 through 1954. Stroke 2:11-22, 1971 5. Matsumoto N, Whisnant JP, Kurland LT, Okazaki H: Natural history of stroke in Rochester, Minnesota, 1955 through 1969: an extension of a previous study, 1945 through 1954. Stroke 4:20-29, 1973 6. Whisnant JP, Cartlidge NEF, Elveback LR: Carotid and vertebral-basilar transient ischemic attacks: effect of anticoagulants, hypertension, and cardiac disorders on survival and stroke occurrence—a population study. Ann Neurol 3:107-115, 1978 7. Dozzi DL: Cerebral embolism as a complication of cor onary thrombosis. Am J Med Sei 194:824-830, 1937 8. Jordan RA, Miller RD, Edwards JE, Parker RL: Thrombo-embolism in acute and in healed myocardial infarction. I. Intracardiac mural thrombosis. Circulation 6:1-6, 1952 9. Miller RD, Jordan RA, Parker RL, Edwards JE: Thrombo-embolism in acute and in healed myocardial infarction. II. Systemic and pulmonary arterial occlu sion. Circulation 6:7-15, 1952 10. Komrad MS, Coffey CE, Coffey KS, McKinnis R, Massey EW, Califf RM: Myocardial infarction and stroke. Neu rology 34:1403-1409, 1984 11. Fulton RM, Duckett K: Plasma-fibrinogen and thromboemboli after myocardial infarction. Lancet 2:1161-1164, 1976 12. Thompson PL, Robinson JS: Stroke after acute myocar dial infarction: relation to infarct size. Br Med J [Clin Res] 2:457-459, 1978 13. McAllen PM, Marshall J: Cerebrovascular incidents af ter myocardial infarction. J Neurol Neurosurg Psychi atry 40:951-955, 1977
Mayo Clin Proc, December 1987, Vol 62
14. Stemmermann GN, Hayashi T, Resch JA, Chung CS, Reed DM, Rhoads GG: Risk factors related to ischemic and hemorrhagic cerebrovascular disease at autopsy: the Honolulu Heart Study. Stroke 15:23-28,1984 15. Chin PL, Kaminski J, Rout M: Myocardial infarction coincident with cerebrovascular accidents in the elderly. Age Ageing 6:29-37,1977 16. Moles KW, Grant AP, Biggart JD: The coincidence of infarcts of the heart and brain: an analysis. Ir J Med Sei 151:145-150,1982 17. Von Arbin M, Britton M, de Faire U, Helmers C, Miah K, Murray V: Myocardial infarction in patients with acute cerebrovascular disease. Eur Heart J 3:136-141, 1982 18. Hellerstein HK, Martin JW: Incidence of thromboembolic lesions accompanying myocardial infarction. Am Heart J 33:443-452,1947
STROKE AND CORONARY HEART DISEASE
1083
19. Kagan AR: Atherosclerosis and myocardial lesions in subjects dying from fresh cerebrovascular disease. Bull WHO 53:597-600,1976 20. Badui E, Estaflol B, Garcia-Rubi D: Coincidence of cerebrovascular accident and silent myocardial infarction, Angiology 33:702-709,1982 21. Wolf PA, Dawber TR, Kännel WB: Heart disease as a precursor of stroke. Adv Neurol 19:567-576, 1978 22. Garraway WM, Whisnant JP, Drury I: The continuing decline in the incidence of stroke. Mayo Clin Proc 58:520523, 1983 23. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
y& Proceedings ROCHESTER, MINNESOTA
DECEMBER 1987
The Association of Stroke and Coronary Heart Disease: A Population Study
DONN D. DEXTER, Jr., M.D., Department of Neurology; JACK P. WHISNANT, M.D., Departments of Neurology and Health Sciences Research; DANIEL C. CONNOLLY, M.D., Division of Cardiovascular Diseases and Internal Medicine; W. MICHAEL O'FALLON, Ph.D., Section of Biostatistics In a 20-year population-based study (1960 through 1979), we estimated the cumulative probability of (1) the occurrence of stroke after a diagnosis of angina pectoris, (2) the occurrence of stroke after a diagnosis of myocardial infarction, and (3) the occurrence of myocardial infarction or sudden unexpected death after a diagnosis of cerebral infarction. In patients in whom angina had been diagnosed, no significant difference was noted between the observed and the expected probability of stroke throughout 10 years of follow-up. In patients with a diagnosis of myocardial infarction, a significant difference was noted between observed and expected probabilities of stroke at 1 month and at 2 months. This early excess in probability of stroke was especially pronounced in the subgroup of patients with transmural myocardial infarction but not evident in those with subendocardial myocardial infarction. Among patients with a diagnosis of cerebral infarction, the difference between observed and expected probabilities of myocardial infarction or sudden unexpected death was not significant until 5 years after the cerebral infarction and showed no change thereafter. We conducted this study to determine whether, in a defined population, an excess risk of having a stroke was present after the onset of coronary heart disease or an excess risk of myocardial infarction (MI) or sudden unexpected death ex isted after a first cerebral infarct. The Mayo Clinic medical record index and record linkage system has been used to study stroke and cor-
This study was conducted in the Mayo Cerebrovascular Re search Center and was supported in part by Grants AM 30582, HL 24326, and NS 06663 from the National Institutes of Health, Public Health Service. Address reprint requests to Dr. J. P. Whisnant, Department of Neurology, Mayo Clinic, Rochester, MN 55905. Mayo Clin Proc 62:1077-1083, 1987
onary heart disease over several decades in the population of Rochester, Minnesota. 1 5 Existing data sets were used in this study to estimate (1) the probability of stroke after the diagnosis of angina pectoris, (2) the probability of stroke after the diagnosis of MI, and (3) the probability of MI or sudden unexpected death after the diagnosis of cerebral infarction. Life-table analyses were used and in each instance were compared with expected rates for the Rochester population by using subjects of the same age and sex as the patients. BACKGROUND Little has been published about the association of angina pectoris and stroke, but a population1077
1078
STROKE AND CORONARY HEART DISEASE
based study of transient cerebral ischemic at tacks revealed no evidence that previous angina pectoris, MI, valvular heart disease, or cardiac arrhythmia, individually or in combination, in fluenced the probability of the occurrence of stroke in patients with transient ischemic attacks. 6 Interest in the occurrence of stroke after MI was evident in the medical literature 5 decades ago and is still a subject of continuing investi gations. 7 " 21 In autopsy studies, the frequency of stroke associated with acute MI was reported to range from 1.2 to 9%.7"9 Left ventricular mural thrombi were found at autopsy in 37% of patients who died of acute MI.8 In another study, 35% of patients with left ventricular mural thrombi had a pathologic diagnosis of systemic arterial occlu sion, and almost half of these presumably embolic lesions were found in the cerebral circula tion. 9 In more recent prospective clinical studies in which follow-up of patients was conducted for up to 6 weeks after acute MI, the frequency of stroke reportedly ranged from 1 to 8.6%.10"12 Cardiogenic emboli have also been reported as an important cause of stroke after acute MI in several clinical studies.10"13'18·20'21 A review of the literature regarding stroke as a late complication of MI revealed that five studies in which 2,662 patients who had had MI and were under surveillance for 2 to 10 years reported no mention of stroke as a complication, whereas in five other studies involving 1,094 such patients, stroke was listed as the cause of 1 to 8% of the deaths. A 5-year follow-up of 260 pa tients with MI who had been admitted to a cor onary care unit and were subsequently dismissed from the hospital noted a 4% frequency of stroke plus a 2% frequency of transient ischemic attacks detected at the time of assessment 5 years after hospital dismissal. 13 An autopsy series of 117 patients with healed (older than 6 weeks) MI disclosed left ventricular thrombi in 24% and systemic arterial occlusion in 26% of patients with mural thrombi in the left ventricle. The frequency of stroke was 17% overall in the 117 patients with healed MI. The investigators noted, however, that cerebral infarction was equally as likely to be found in patients without as in those with left ventricular mural thrombi, a finding that suggests the decreasing importance of em bolism and the increasing prominence of shared risk factors in the causation of stroke as the time after acute MI increased. 8,9
Mayo Clin Proc, December 1987, Vol 62
In most reported cases in which the onset of cardiovascular and cerebrovascular disease could be dated, the cardiac problem occurred first.14 Several published reports have described concur rent myocardial and cerebral infarctions, but the temporal relationship between the two conditions has often been undetermined. 15 " 20 In one study of concurrent myocardial and cerebral infarction, two of nine patients probably had a stroke first and MI shortly thereafter, and the importance of shared risk factors relating these two conditions was noted. 17 METHODS The unique capability for population-based epidemiologic research in Rochester, Minnesota, is facilitated by the detailed medical records avail able for the residents of this city. The Mayo Clinic has served as the major health-care provider for local residents since the turn of the century, and an index of diagnoses and surgical procedures for all medical contacts is maintained in the medical record of each patient. A central file of medical records containing an index of diagnoses and surgical procedures entered by other health-care providers for local residents is indexed to the Mayo record system for retrieval. The result is a comprehensive record of the health care re ceived by the population of Rochester, Minnesota. The population of Rochester, Minnesota, was 39,012 in 1960, 52,629 in 1970, and 56,447 in 1980 as determined by census statistics. More than 99% of the population in this community are Caucasian. The Rochester population is some what younger and has a slightly higher mean income and level of education than the US white population in general. 1 The autopsy rate for deaths among Rochester residents has been ap proximately 60%. This system of medical records has been used to study the incidence of coronary heart disease in residents of Rochester, Minnesota, for the period Jan. 1, 1950, through Dec. 31, 1979.1"3 The diagnosis of coronary heart disease was accorded to three major initial manifestations: angina pec toris, MI, and sudden unexpected death. In ad dition to the strict diagnostic criteria, patients must have been a resident of Rochester for at least 1 year before the initial diagnosis of cor onary heart disease (in order to exclude persons who might have moved to Rochester because of their disease). The patient must also have had
Mayo Clin Proc, December 1987, Vol 62
no prior diagnosis of coronary heart disease and no clinical evidence of congestive heart failure, congenital or rheumatic heart disease, noncoronary cardiomyopathy, or valvular heart disease before inclusion in the study. A continuing study of the incidence of stroke in Rochester, Minnesota, for the period Jan. 1, 1945, through Dec. 31, 1979, provided the infor mation about the patients with stroke in this report.4,5,22 During this period, 2,133 new cases of stroke occurred during 1.5 million person-years of observation in residents of Rochester. The neurologic history and examination record have been standardized for much of the 70-year period that the Department of Neurology at the Mayo Clinic has been in existence. This standardiza tion, as well as the high autopsy rate (58% of patients in the community who died of stroke from 1970 through 1974 had a complete post mortem evaluation), adds to the reliability of the neurologic observations and categorization of stroke. Strokes have been categorized as cerebral and cerebellar or brainstem arterial thrombosis, cerebral and cerebellar or brainstem embolus, intracerebral and intracerebellar or brainstem hemorrhage, primary subarachnoid hemorrhage, or type unknown. The criteria used for the var ious diagnostic categories of stroke have been published previously. 4 In the current study, we used data gathered and computerized for the aforementioned studies to analyze the association of coronary heart disease and stroke. During the 20-year period from Jan. 1, 1960, through Dec. 31, 1979, 2,622 patients fulfilled the diagnostic criteria for the coronary heart disease study—1,015 patients with an ini tial diagnosis of angina, 1,183 with an initial diagnosis of MI or concurrent angina and MI, and 424 with an initial diagnosis of sudden un expected death. Among the patients having an initial diagnosis of angina, 319 cases had sub sequent MI. These cases were combined with the patients who had an initial diagnosis of MI or concurrent angina and MI; thus, the total number of patients with MI was 1,502. The patients with MI were also divided into the subgroups of transmural and subendocardial (also called nontransmural or non-Q-wave) MI. Patients who had a stroke before the initial diagnosis of angina or MI were excluded from the analysis; the resulting groups were 991 patients with an initial diag nosis of angina, 1,127 with an initial diagnosis
STROKE AND CORONARY HEART DISEASE
1079
of MI or concurrent angina and MI, and 1,419 in the combined MI group. There were 1,253 patients from the stroke study in the period covered in this report. This number included 1,043 patients with cerebral infarction or stroke of uncertain type, which is considered primarily infarction. Stroke occurring after the diagnosis of MI included all cases of cerebral infarction, stroke of uncertain type, and intra cerebral hemorrhage but did not include patients with subarachnoid hemorrhage. The patients with stroke who were under surveillance for sub sequent MI included only those 943 with cerebral infarction or stroke of uncertain type who did not have a MI before a stroke. In seven patients, the stroke and the MI oc curred on the same day, and which event occurred first could not be determined from the medical record. These patients were excluded from the subsequent analyses. Because both data sets included all cases from 1950 through 1979 and the incidence cases were selected beginning in 1960, detection of the cases to be excluded—because of prior stroke in pa tients with angina or MI or because of prior MI in patients with cerebral infarction—should be almost complete. The Kaplan-Meier life-table method 23 was used to estimate (1) the cumulative probability of stroke after a diagnosis of angina pectoris by using the date of diagnosis of angina as the beginning date, the date of stroke as the event date, and the date of last follow-up or of death from a cause other than stroke as the withdrawal date; (2) the cumulative probability of stroke after a diagnosis of MI by using the date of diagnosis of MI as the beginning date, the date of stroke as the event date, and the date of last follow-up or of death from a cause other than stroke as the withdrawal date; and (3) the cumulative prob ability of MI or sudden unexpected death after a cerebral infarction by using the date of cerebral infarction as the beginning date, the date of MI or sudden unexpected death as the event date, and the date of last follow-up or of death from any other cause as the withdrawal date. Separate analyses were performed on the entire group of patients with MI as well as the subgroups of transmural (N = 782) and subendocardial (N = 404) MI. The incidence rates for stroke and MI in age- and sex-matched subjects in the Rochester population were used to determine the expected cumulative
1080
STROKE AND CORONARY HEART DISEASE
probability of stroke or MI and sudden unex pected death for each of the aforementioned situations. Observed and expected rates were compared by using a one-sample log-rank test. RESULTS During the period studied, 1,015 patients had an initial diagnosis of angina pectoris. The 24 pa tients (2.4%) who had a stroke before the initial diagnosis of angina were excluded. In 53 pa tients, stroke occurred after the initial diagnosis of angina pectoris during the period of study. The mean age of patients with angina without prior stroke was 64 years (60 years for men and 67 years for women). The observed 10-year cumu lative probability of the occurrence of stroke in patients after the initial diagnosis of angina was 5.8%. No significant difference was noted be tween the observed and the expected cumula tive probability of stroke throughout 10 years of follow-up (Fig. 1). During the period studied, 1,502 patients had the diagnosis of MI, concurrent angina and MI, or angina followed by MI. In 80 of these patients (5.3%), a history of stroke before the initial diag-
Fig. 1. The cumulative probability of stroke 5 after initial diagnosis of angina pectoris (N = 991), given survival. Ex pected cumulative probability of stroke is for an age- and sexmatched population and is based on incidence rates for stroke in Rochester, Minnesota, for the period 1960 through 1979.22 No significant difference was noted between observed and expected cumulative probabilities of stroke throughout 10 years of follow-up.
Mayo Clin Proc, December 1987, Vol 62
nosis of MI excluded them from further analysis. In 74 patients, stroke occurred after the initial diagnosis of MI. The mean age of patients with out prior stroke was 65 years (62 years for men and 70 years for women). The curves depicting the observed and the expected probability of stroke after MI (Fig. 2) were compared by using the log-rank test, and the difference in the curves was significant (P<0.01). The difference between the observed (0.9%) and the expected (0.05%) prob ability of stroke was significant at 1 month (P<0.01) and also at 2 months (1.25% versus 0.1%; P<0.001). No further change in the difference was noted through the first year of observation (Fig. 2). This early excess in the probability of stroke was also noted—and to a greater extent— in the subgroup of patients with transmural MI, in whom the incidence of stroke at 2 months was increased by a factor of 15 over that of the expected incidence (Fig. 2). No significant differ ence was observed in the cumulative probability of stroke after subendocardial MI in comparison with the expected incidence at any time during the first year of follow-up (Fig. 2) or during the subsequent 9 years. Cerebral infarction was diagnosed in 1,044 pa tients during the period studied. The 93 patients (8.9%o) who had a MI before the initial diagnosis of cerebral infarction were excluded from further study. MI or sudden unexpected death occurred in 90 patients at the time of or after the initial diagnosis of cerebral infarction during the years of follow-up. The mean age of those patients without prior MI was 73 years (70 years for men and 75 years for women). The cumulative prob ability of MI or sudden unexpected death after cerebral infarction was significantly increased in comparison with the expected probability throughout 10 years of follow-up (Fig. 3) (log-rank test, P<0.01). The observed and the expected cumulative incidences for MI or sudden unex pected death after cerebral infarction were 0.8% and 0.5% at 6 months, 1.4% and 1.0% at 12 months, 10.6% and 5.4% at 5 years (P<0.001), and 16.6% and 11.0% at 10 years (P<0.01). No signif icant difference was noted between the observed and expected probabilities of MI or sudden un expected death until 5 years after the first cere bral infarction; however, the curves for observed versus expected probabilities were divergent after the first 6 months and continued in this pattern through 5 years.
Mayo Clin Proc, December 1987, Vol 62
io
—
STROKE AND CORONARY HEART DISEASE
Expected SMI
8
Ml —
TMI
6
% 4
2
0 C
1
2
3
4
5
6
7
8
9
10
I
I
11
12
Months after first diagnosis of Ml
Fig. 2. The cumulative probability of stroke after initial diagnosis of myocardial infarction (MI) (N = 1,419) and transmural (TMI) (N = 782) and subendocardial infarction (SMI) (N = 404), given survival. Expected cumulative prob ability of stroke is for a population that was matched for age and sex (to the patients with MI) and is based on incidence rates for stroke in Rochester, Minnesota, for the period 1960 through 1979.22 A significant difference was noted between observed and expected probabilities of stroke at 1 month and at 2 months; this early excess in probability of stroke was especially pronounced in subgroup of patients with TMI.
DISCUSSION For the time period studied (1960 through 1979), this study of the Rochester, Minnesota, popula tion included essentially all persons with onset of angina pectoris and all those with a first MI who had not had a prior stroke; it also included essentially all persons with a first cerebral infarc tion who had not had a prior MI. For the assess ment of the probability of the occurrence of stroke after the onset of angina pectoris, 24 persons who had a stroke before the diagnosis of angina pec toris were excluded from analysis, and 80 persons who had a stroke before the onset of MI were excluded from the analysis of the probability of the occurrence of stroke after MI. In addition, 93 patients were excluded from the analysis of prob ability of MI or sudden unexpected death after a first cerebral infarction because those patients had MI before the cerebral infarct. In seven patients, we could not determine with certainty whether the MI or the cerebral infarc tion occurred first. The timing of the detection of the serum enzyme abnormalities suggested, however, that in most, if not all, of these patients the MI occurred first. Because of the uncertainty, we thought that the interpretation of the follow-
1081
up data would be less biased if these seven cases were not included in the life-table analyses. Patients with the diagnosis of angina pectoris had no evidence of an increased risk of stroke either soon after the diagnosis or at any time during the next 10 years. Those patients who had a diagnosis of MI had a significant increase in the probability of stroke. This increased risk was most notable at 1 month, at which time the risk was increased by a factor of 18, and it was still increased by a factor of 13 at 2 months (Fig. 2). When a person-years analysis was performed after the first 2 months, no significant difference was found in the ob served and the expected number of strokes except in one of the remaining months during the first year of observation. The increase in risk of stroke during the first 2 months after MI was even more evident among patients with transmural MI than among the overall MI group (Fig. 2). No signif icant difference was noted, however, between the observed and expected number of strokes at any point in the life-table analysis for patients with subendocardial MI (Fig. 2). These observations suggest that a direct caus ative factor relating to stroke is present soon after
Fig. 3. The cumulative probability of myocardial infarction or sudden unexpected death after initial diagnosis of cerebral infarction (N = 943), given survival. Expected cumulative probability of myocardial infarction or sudden unexpected death is for an age- and sex-matched population and is based on incidence rates for myocardial infarction and sudden unexpected death in Rochester, Minnesota, for the period 1960 through 1979.2 No significant difference was noted between observed and expected probabilities of myocardial infarction or sudden unexpected death until 5 years after the diagnosis of cerebral infarction.
1082
STROKE AND CORONARY HEART DISEASE
the onset of MI but that it is not operative beyond the first 2 months after MI. Presumably, this factor is intracardiac mural thrombi that produce embolization. In this study, we do not have in formation about treatment with anticoagulants or other drugs from which a differential outcome might be judged. The results noted apply to all patients without regard to treatment. It is evi dent, however, that if treatment to prevent stroke is to be effective it must have its benefit during the first 2 months after MI, and there is little likelihood of benefit thereafter. Among the patients with the first cerebral in farction, only one clear case of subsequent MI occurred during the first month of follow-up. That patient had an anterior MI and cardiac arrest 4 days after the cerebral infarction. The difference between the observed and the expected prob ability of MI or sudden unexpected death after cerebral infarction was not significant at any observational point until 5 years after the onset of cerebral infarction. At 5 years, the risk of MI or sudden unexpected death was increased by a factor of 2, and the difference was significant (P<0.001) (Fig. 3). In only 1 of the next 5 years was the observed number of cases of MI or sud den unexpected death significantly greater than the expected number. This lack of difference from 5 to 10 years is apparent in the plotted curves (Fig. 3). The differences noted at 5 years do not consider risk factors other than the cerebral in farction that might have affected the occurrence of MI or sudden unexpected death. CONCLUSION In our study population, no significant difference was noted between the observed and the expected cumulative incidence of stroke in patients after the initial diagnosis of angina pectoris. We noted a significant increase in the cumula tive incidence of stroke at 1 and 2 months after MI but no change in this difference after 2 months, an indication that no further increase in risk existed after 2 months. The early excess of stroke was evident in the subgroup of patients with transmural MI but not in those with subendocardial MI. These observations suggest a causative factor related to embolization from intracardiac mural thrombi in patients with transmural MI for 2 months after the diagnosis of MI but not thereafter.
Mayo Clin Proc, December 1987, Vol 62
A significant increase was found in the cumula tive incidence of MI or sudden unexpected death after cerebral infarction at 5 years but no change in the difference after that time. The difference between the observed and expected probabilities of MI and sudden unexpected death gradually increased until 5 years after the first cerebral infarct; thus, the differences were likely due to shared risk factors for myocardial and cerebral infarction rather than a direct effect of cerebral infarction.
REFERENCES 1. Connolly DC, Oxman HA, Nobrega FT, Kurland LT, Kennedy MA, Elveback LR: Coronary heart disease in residents of Rochester, Minnesota, 1950-1975. I. Back ground and study design. Mayo Clin Proc 56:661-664, 1981 2. Elveback LR, Connolly DC, Kurland LT: Coronary heart disease in residents of Rochester, Minnesota. II. Mor tality, incidence, and survivorship, 1950-1975. Mayo Clin Proc 56:665-672, 1981 3. Connolly .DC, Elveback LR: Coronary heart disease in residents of Rochester, Minnesota. VI. Hospital and posthospital course of patients with transmural and subendocardial myocardial infarction. Mayo Clin Proc 60:375-381,1985 4. Whisnant JP, Fitzgibbons JP, Kurland LT, Sayre GP: Natural history of stroke in Rochester, Minnesota, 1945 through 1954. Stroke 2:11-22, 1971 5. Matsumoto N, Whisnant JP, Kurland LT, Okazaki H: Natural history of stroke in Rochester, Minnesota, 1955 through 1969: an extension of a previous study, 1945 through 1954. Stroke 4:20-29, 1973 6. Whisnant JP, Cartlidge NEF, Elveback LR: Carotid and vertebral-basilar transient ischemic attacks: effect of anticoagulants, hypertension, and cardiac disorders on survival and stroke occurrence—a population study. Ann Neurol 3:107-115, 1978 7. Dozzi DL: Cerebral embolism as a complication of cor onary thrombosis. Am J Med Sei 194:824-830, 1937 8. Jordan RA, Miller RD, Edwards JE, Parker RL: Thrombo-embolism in acute and in healed myocardial infarction. I. Intracardiac mural thrombosis. Circulation 6:1-6, 1952 9. Miller RD, Jordan RA, Parker RL, Edwards JE: Thrombo-embolism in acute and in healed myocardial infarction. II. Systemic and pulmonary arterial occlu sion. Circulation 6:7-15, 1952 10. Komrad MS, Coffey CE, Coffey KS, McKinnis R, Massey EW, Califf RM: Myocardial infarction and stroke. Neu rology 34:1403-1409, 1984 11. Fulton RM, Duckett K: Plasma-fibrinogen and thromboemboli after myocardial infarction. Lancet 2:1161-1164, 1976 12. Thompson PL, Robinson JS: Stroke after acute myocar dial infarction: relation to infarct size. Br Med J [Clin Res] 2:457-459, 1978 13. McAllen PM, Marshall J: Cerebrovascular incidents af ter myocardial infarction. J Neurol Neurosurg Psychi atry 40:951-955, 1977
Mayo Clin Proc, December 1987, Vol 62
14. Stemmermann GN, Hayashi T, Resch JA, Chung CS, Reed DM, Rhoads GG: Risk factors related to ischemic and hemorrhagic cerebrovascular disease at autopsy: the Honolulu Heart Study. Stroke 15:23-28,1984 15. Chin PL, Kaminski J, Rout M: Myocardial infarction coincident with cerebrovascular accidents in the elderly. Age Ageing 6:29-37,1977 16. Moles KW, Grant AP, Biggart JD: The coincidence of infarcts of the heart and brain: an analysis. Ir J Med Sei 151:145-150,1982 17. Von Arbin M, Britton M, de Faire U, Helmers C, Miah K, Murray V: Myocardial infarction in patients with acute cerebrovascular disease. Eur Heart J 3:136-141, 1982 18. Hellerstein HK, Martin JW: Incidence of thromboembolic lesions accompanying myocardial infarction. Am Heart J 33:443-452,1947
STROKE AND CORONARY HEART DISEASE
1083
19. Kagan AR: Atherosclerosis and myocardial lesions in subjects dying from fresh cerebrovascular disease. Bull WHO 53:597-600,1976 20. Badui E, Estaflol B, Garcia-Rubi D: Coincidence of cerebrovascular accident and silent myocardial infarction, Angiology 33:702-709,1982 21. Wolf PA, Dawber TR, Kännel WB: Heart disease as a precursor of stroke. Adv Neurol 19:567-576, 1978 22. Garraway WM, Whisnant JP, Drury I: The continuing decline in the incidence of stroke. Mayo Clin Proc 58:520523, 1983 23. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958