Unrecognized Myocardial Infarction: Epidemiology, Clinical Characteristics, and the Prognostic Role of Angina Pectoris: The Reykjavik Study

The clinical manifestations of coronary heart disease vary considerably. Since it was first described by James B. Herrick in 1912, clinically unrecognized myocardial infarction [1] has been extensively researched and debated. Epidemiologic studies have shown that silent, atypical, or unrecognized myocardial infarctions constitute between 20% and 60% of all myocardial infarctions [2-7]. Unrecognized myocardial infarction is diagnosed objectively using thallium perfusion imaging, radionuclide angiography, or echocardiography; it is most often diagnosed from typical, unequivocal changes on the electro-cardiogram of a patient with symptoms so atypical that neither patient nor physician recognizes the problem as an infarction. Atypical and silent myocardial infarction have traditionally been grouped together as unrecognized myocardial infarction. Patients with silent myocardial infarction seemingly have no symptoms. Half of all patients with unrecognized myocardial infarction recall no symptoms and have therefore had silent myocardial infarctions; the remainder have had atypical myocardial infarctions [8, 9]. Silent myocardial ischemia as an important manifestation in patients with coronary heart disease has been studied in recent decades and clearly affects prognosis unfavorably [10, 11]. Because the prognosis for patients with unrecognized myocardial infarction seems to be as serious as that for patients with recognized myocardial infarction [12, 13], practicing physicians face considerable diagnostic and therapeutic challenges when dealing with the many patients with this condition. Not only is it difficult to choose methods with which to identify these patients, it is also difficult to make decisions about secondary prevention and medical treatment. Detailed knowledge about this disease entity is therefore important and must include a thorough understanding of which patient subgroups are especially vulnerable. We report the results of a long-term study of a population-based cohort participating in the Reykjavik Study. Our purpose was to determine the incidence, prevalence, and prognosis associated with unrecognized myocardial infarction. We evaluated the risk factor profile for patients with unrecognized myocardial infarction compared with that of patients with recognized myocardial infarction, as well as the prognostic role of angina pectoris in persons with unrecognized myocardial infarction. Methods The design of the Reykjavik study has been described previously [6], and only a brief description is included here. The study is a large population-based cohort study that started in 1967. Men living in the Reykjavik area who were born between 1907 and 1934 were invited to participate. The study has been conducted in several stages: 1967-1968, 1970-1971, 1974-1975, 1979-1980, and 1983-1987. The response rate has varied from 64% to 75%. Since 1969, women have also participated in the study, but their results will be the subject of a separate report. Each participant answered a questionnaire that included the Rose chest pain questionnaire used by the London School of Hygiene and Tropical Medicine [14]; was examined by a physician; and had a standardized 12-lead electrocardiogram recorded and evaluated according to the Minnesota Code [15]. A total of 9141 men participated in the study at least once. Since 1981, data on the incidence of myocardial infarction have been collected as part of the World Health Organization MONICA Project [16]. Hospital records for persons who had had myocardial infarctions before 1981 were reviewed and evaluated according to criteria used in the MONICA study. Causes of death were determined from all death certificates from the start of the study until 31 December 1987. All autopsy records were also reviewed (autopsy rate, 55%). Diagnostic categories were defined as follows: 1. Recognized myocardial infarction: Patients who fulfilled the MONICA criteria for definite myocardial infarction [16] were placed in this category. These criteria include electrocardiographic changes (Minnesota codes 1.1.1-1.2.8); typical, atypical, or inadequately described symptoms together with probable electrocardiographic changes and abnormal enzyme levels; and typical symptoms and abnormal enzyme levels with ischemic or noncodable electrocardiographic results. 2. Unrecognized myocardial infarction: Participants in this category had no history or symptoms of heart attack but had electrocardiographic changes that fulfilled the criteria for definite myocardial infarction results (Minnesota codes 1.1.1-1.2.8). 3. Angina pectoris with electrocardiographic manifestations of myocardial ischemia: Participants who fulfilled the Rose questionnaire criteria for angina pectoris and who had either ischemic electrocardiographic results (Minnesota codes 1.3.1-1.3.6, 4.1-4.4, 5.1-5.4) or normal resting electrocardiographic results with a positive exercise stress test result ( 0.2 mV horizontal or down-sloping ST depression) were placed in this category. 4. Angina pectoris with normal resting electrocardiographic results: Participants with normal resting electrocardiographic results who either had normal exercise test results or had had no exercise test were assigned to this category if Rose questionnaire results indicated angina pectoris and if the examining physician could confirm the diagnosis. 5. Angina pectoris by the Rose questionnaire only: Participants fulfilling the Rose questionnaire criteria for angina pectoris, if their electrocardiograms did not indicate ischemia and the investigating physician could not confirm the diagnosis, were placed in this category. 6. No coronary heart disease: Participants in this category had no history or electrocardiographic manifestations of myocardial infarction. Patients were classified as having recognized myocardial infarction (category 1) on the basis of hospital records. Classification into the other diagnostic categories, including that of unrecognized myocardial infarction, was based on data collected at set intervals when patients attended the study clinic. Statistical Methods Three designs were used. In the first, a cross-sectional study, logistic regression was used to compute the prevalence odds of unrecognized myocardial infarction as a function of age and calendar year. It was also used to estimate the dependence of unrecognized and recognized myocardial infarction on simultaneous values of measured variables. In the second design, a prospective study, Poisson regression was used to compute the incidence of unrecognized myocardial infarction as a function of age and calendar year. It was also used to compute the predictive power of the measured variables for future unrecognized and recognized myocardial infarction. In these computations, consecutive visits were paired and each pair was used; pairs of visits were excluded if myocardial infarction had been diagnosed during the former visit. The risk period was the time elapsed between two visits (3 to 6 years), and age was the participant's age halfway between the visits. In the third design, a prospective study of survival, Cox regression was used to estimate the simultaneous predictive power for risk for death (cause-specific or from all causes). -coefficients were calculated to investigate the prognostic value of risk factors and to form a composite score for individual persons. The composite risk score was the product sum of -coefficients and values of risk factors. Significance testing was two-sided and based on a 5% probability level. The software package used was EGRET (Epidemiologic Graphics, Estimation and Testing) [17]. Results Prevalence and Incidence The overall prevalence of unrecognized myocardial infarction in the first stage of the study was 0.5% in 1968 and 0.4% in 1971, and it increased in later stages of the study. In 1975 and 1980 it was 1.0% and 1.3%, respectively, and in the last stage, in 1986, it was 2.8%. To adjust for changes in age in the participants during the study period, we used logistic regression in which the prevalence odds were modeled as a function of age and either stage number or calendar year. As shown in Figure 1, prevalence increased steeply after age 60 years; it was 0.5% at age 50 years but exceeded 5% at age 75 years. The odds ratio per year was 1.10 (95% CI, 1.07 to 1.12). There was no significant time trend when the computation of prevalence was limited to the first visit of each participant, thus eliminating the bias introduced by the diagnosis of the previously unrecognized infarction. Figure 1. The prevalence of unrecognized myocardial infarction as a function of age. The incidence rate was obtained from the prospective study using a Poisson regression. The only explanatory variables tried at this stage were age, age squared, and calendar time. The incidence rate did not depend significantly on calendar time. Figure 2 shows that incidence was almost zero before age 40 years and increased steeply from age 40 years to age 60 years, at which it exceeded 300 per year per 100 000 persons. After age 65 years, the incidence rate decreased with age. The odds ratio for age (per year) was 2.06 (CI, 1.23 to 3.46); for age squared it was 0.994 (CI, 0.990 to 0.999). This was a significant (P < 0.05) contribution by age squared to the explanation of the incidence rate and indicated a decrease in the incidence rate of unrecognized myocardial infarction after age 65 years (Figure 2). Figure 2. Incidence of unrecognized myocardial infarction as a function of age. Risk Factor Profile Table 1 compares the mean values of some of the baseline characteristics of the cohort that had unrecognized myocardial infarction with those of the cohort that had recognized myocardial infarction. Although more participants with recognized than with unrecognized myocardial infarction were treated for high blood pressure and diabetes mellitus, the differences were not significant. Age and cholesterol, triglycer