Patients with myocardial infarction (MI) generally present with chest pain or pressure at rest or minimal exertion and have associated electrocardiographic changes and/or elevation of the biomarkers of myocardial necrosis. A subset of patients, however, experience little chest discomfort or do not present to medical attention despite experiencing symptoms. Unrecognized MI might be detected using electrocardiographic or imaging techniques, such as echocardiography, nuclear imaging, or cardiovascular magnetic resonance imaging. Unrecognized MI is a common clinical entity, with an incidence as great as 35% in high-risk populations. Moreover, the risk of a subsequent major adverse cardiovascular event might be similar to the risk after a clinically apparent MI. In the present review, we examined the incidence of unrecognized MI across broad groups of subjects and the subsequent risk of adverse cardiovascular events. Finally, we explored the potential role of including unrecognized MI as a major adverse outcome in randomized clinical trials of agents aimed at reducing cardiovascular morbidity.
The clinical presentation of acute coronary syndrome varies widely. Patients with recognized myocardial infarction (MI) usually present with chest discomfort at rest, electrocardiographic (ECG) changes, and/or elevation of cardiac biomarkers and thus receive the appropriate triage and management. In contrast, some patients will have so-called silent MI, with no or mild symptoms, and thus do not present to medical attention. The incidence and prevalence of unrecognized MI could be as great as 35% and 44%, respectively, and such patients have outcomes similar to, or worse than, those with recognized MI. Moreover, clinical trials evaluating novel techniques and pharmacologic agents among patients with stable coronary artery disease (CAD) or acute coronary syndrome frequently assess clinically evident MI as an end point. The addition of unrecognized MI as a clinical end point has the potential to decrease enrollment and shorten the length of clinical trials so as not to expose patients to undue harm from experimental therapies for a more prolonged period. The goal of the present report was to review the definition, prevalence, and clinical consequences of unrecognized MI and explore the role of identifying unrecognized MI as a clinical end point in long-term randomized trials.
Definition
Unrecognized MI is defined as MI that initially is undetected and is eventually discovered by surveillance electrocardiography, myocardial imaging techniques, or pathologic findings on autopsy. Traditionally, the appearance of objective ECG findings has been used to classify unrecognized MI. There has, however, been a large degree of heterogeneity in the ECG diagnostic criteria defining unrecognized MI. The result is a wide variability in prevalence estimates across study populations.
In general, unrecognized MI is defined as the appearance of new and persistent pathologic Q waves confirmed by ECG findings. The Framingham group first defined “unequivocal evidence of an interim MI” as new Q waves ≥0.04 second or the loss of R waves. Other studies, including the Israeli Heart Attack Study and the Honolulu Heart study, defined an unrecognized MI as Q waves ≥0.03 second, a loss of R waves, and more specific ST-segment and T-wave criteria.
Beginning with the Reykjavik group, studies have focused on using the Minnesota Code for standardization of ECG interpretations. The Minnesota Code was initially designed for standardization of ECG interpretations across the cohorts for large epidemiologic studies. The code consists of a hierarchical numerical coding system and further improves the diagnostic accuracy of Q waves by excluding those caused by other conditions, such as left ventricular hypertrophy, bundle branch blocks, chronic obstructive pulmonary disease, or obesity, which can cause false-positive results.
The sensitivity and specificity of both visual and computerized interpretations using the Minnesota Code is difficult to determine because no reference standard, prospective ECG–autopsy correlation study is available. Ammar et al examined these characteristics of recognized MI and extrapolated the findings to unrecognized MI. Their findings were that ECG findings have a sensitivity of 20% to 70%, with a specificity of 69% to 99.6%, with the greatest specificity observed with an emphasis on Q-wave criteria. Q waves thus have the greatest positive predictive power of interim MI. Their review suggested that unrecognized MI should be classified as “definite” with the appearance of intermediate or major Q waves, “probable” with minor Q waves accompanied by major ST-segment or T-wave abnormalities, and “possible” with minor Q wave changes. It follows that the inclusion of ST-segment or T-wave abnormalities improves the sensitivity for the detection of unrecognized MI. Regardless, the existence of Q waves in patients with CAD but no clinical evidence of previous MI is independently predictive of adverse cardiovascular events and were thus the foundation for the classification of unrecognized MI.
Incidence and Prevalence of Unrecognized MI
The incidence and prevalence estimates of unrecognized MI were derived from large cohort studies beginning in the 1960s with the Western Collaborative Study Group. Overall, the incidence of unrecognized MI has ranged from 22% to 44% of all documented MI events ( Table 1 ). Of note, the Western Collaborative, Israeli Heart, and Honolulu Heart studies only included men with a mean age of 46 to 55 years. The wide range of incidence estimates likely resulted from the variations in the populations studied, duration and frequency of follow-up, and differences in the ECG classifications of unrecognized MI.
Study | Span | Sample Size (n) | Mean Age (y) | Men (%) | Total MI (n) | Unrecognized MI (%) |
---|---|---|---|---|---|---|
Framingham study | 1948–1990 | 5,070 | 55 | 44% | 363 | 30% |
Western Collaborative Group Study | 1960–1964 | 3,524 | 46 | 100% | 73 | 37% |
Israeli Heart Attack Study | 1963–1968 | 9,509 | NA | 100% | 427 | 40% |
Honolulu Heart Study | 1965–1971 | 7,331 | 55 | 100% | 135 | 22% |
Reykjavik Study in Men | 1967–1987 | 9,141 | 60 | 100% | 237 | 35% |
Reykjavik Study in Women | 1968–1991 | 13,000 | NA | 0% | 641 | 33% |
Multiple Risk Factor Intervention Trial | 1972–1982 | 12,866 | 46 | 46% | 460 | 25% |
Bronx Aging Study | 1980–1988 | 390 | 79 | 36% | 72 | 44% |
Atherosclerosis Risk in Communities | 1987–1998 | 12,843 | NA | NA | 508 | 20% |
Cardiovascular Health Study | 1989–1995 | 5,888 | 72 | 42% | 901 | 22% ∗ |
Rotterdam Study | 1990–1993 | 3,272 | 67 | 39% | 141 | 36% |
Heart and Estrogen/progestin Replacement Study Trial | 1993–1998 | 2,763 | 67 | 0% | 256 | 4% |
Irbesartan Diabetic Nephropathy Trial | 1996–1999 | 1,387 | 60 | 65% | 99 | 14% |
Fenofibrate Intervention and Event Lowering in Diabetes Study | 1998–2005 | 9,795 | 63 | 68% | 406 | 37% |
Heart and Soul Study | 2000–2008 | 462 | 67 | 79% | 141 | 36% |
Fremantle Diabetes Study | 1993–1996 | 126 | 66 | 50% | 114 | 44% ∗ |
The best-known, large-cohort study comes from the Framingham group, which documented an incidence of unrecognized MI of 26% and 34% of all MIs in men and women, respectively. Roughly half of affected patients reported no symptoms, and the other half experienced nonspecific symptoms not initially attributed to an MI. These results were similar to those from the Reykjavik group, which reported an incidence of 33% during 23 years of follow-up in women and 35% during 20 years of follow-up in men. The Atherosclerosis Risk in Communities (ARIC) study was the first to include black subjects and followed a cohort of men and women aged 45 to 65 years without CAD or ECG evidence of an MI at baseline. Of the 508 MIs, 20% were unrecognized, with no significant difference observed between men and women. Subsequently, the Rotterdam study group followed up a large cohort of men and women aged >55 years from 1990 to 2000 and demonstrated an incidence of unrecognized MI in men of 33% and in women of 54% of all documented MIs.
Several studies have examined the incidence of unrecognized MI in patients with diabetes mellitus. The Irbesartan Diabetic Nephropathy Trial observed patients with type 2 diabetes mellitus, hypertension, and nephropathy with mean follow-up of 2.5 years and found that 14% of all first nonfatal MIs were unrecognized. The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study followed up patients with diabetes aged 50 to 75 years. The observed incidence of unrecognized MI was 36.8% of all MIs, with a greater proportion of those being inferior infarctions.
The Cardiovascular Health Study (CHS) and Fremantle Diabetes Study reported a 22% and 44% prevalence, respectively, of unrecognized MI at baseline. The CHS study included men and women aged >65 years, and the Fremantle study included a community cohort of patients with diabetes mellitus, a population in which unrecognized MI would be expected to have a greater prevalence.
It is worth noting that most of these large cohort studies likely underestimated the incidence and prevalence of unrecognized MI for several reasons. As discussed previously, the ECG criteria used to classify MI has varied, and those studies that emphasized Q waves will not capture many non–Q-wave MIs. This might represent a significant proportion of MI, because it has been reported that only 15% of patients who present with non–ST-segment elevation MI will develop Q waves. In addition, it has been estimated that the indicators of MI on the electrocardiograms disappear within 2 years in 10% of patients with an anterior infarction and 25% of patients with an inferior infarction. Thus, the diagnosis can be missed if the electrocardiogram is not taken shortly after an MI.
Incidence and Prevalence of Unrecognized MI
The incidence and prevalence estimates of unrecognized MI were derived from large cohort studies beginning in the 1960s with the Western Collaborative Study Group. Overall, the incidence of unrecognized MI has ranged from 22% to 44% of all documented MI events ( Table 1 ). Of note, the Western Collaborative, Israeli Heart, and Honolulu Heart studies only included men with a mean age of 46 to 55 years. The wide range of incidence estimates likely resulted from the variations in the populations studied, duration and frequency of follow-up, and differences in the ECG classifications of unrecognized MI.
Study | Span | Sample Size (n) | Mean Age (y) | Men (%) | Total MI (n) | Unrecognized MI (%) |
---|---|---|---|---|---|---|
Framingham study | 1948–1990 | 5,070 | 55 | 44% | 363 | 30% |
Western Collaborative Group Study | 1960–1964 | 3,524 | 46 | 100% | 73 | 37% |
Israeli Heart Attack Study | 1963–1968 | 9,509 | NA | 100% | 427 | 40% |
Honolulu Heart Study | 1965–1971 | 7,331 | 55 | 100% | 135 | 22% |
Reykjavik Study in Men | 1967–1987 | 9,141 | 60 | 100% | 237 | 35% |
Reykjavik Study in Women | 1968–1991 | 13,000 | NA | 0% | 641 | 33% |
Multiple Risk Factor Intervention Trial | 1972–1982 | 12,866 | 46 | 46% | 460 | 25% |
Bronx Aging Study | 1980–1988 | 390 | 79 | 36% | 72 | 44% |
Atherosclerosis Risk in Communities | 1987–1998 | 12,843 | NA | NA | 508 | 20% |
Cardiovascular Health Study | 1989–1995 | 5,888 | 72 | 42% | 901 | 22% ∗ |
Rotterdam Study | 1990–1993 | 3,272 | 67 | 39% | 141 | 36% |
Heart and Estrogen/progestin Replacement Study Trial | 1993–1998 | 2,763 | 67 | 0% | 256 | 4% |
Irbesartan Diabetic Nephropathy Trial | 1996–1999 | 1,387 | 60 | 65% | 99 | 14% |
Fenofibrate Intervention and Event Lowering in Diabetes Study | 1998–2005 | 9,795 | 63 | 68% | 406 | 37% |
Heart and Soul Study | 2000–2008 | 462 | 67 | 79% | 141 | 36% |
Fremantle Diabetes Study | 1993–1996 | 126 | 66 | 50% | 114 | 44% ∗ |
The best-known, large-cohort study comes from the Framingham group, which documented an incidence of unrecognized MI of 26% and 34% of all MIs in men and women, respectively. Roughly half of affected patients reported no symptoms, and the other half experienced nonspecific symptoms not initially attributed to an MI. These results were similar to those from the Reykjavik group, which reported an incidence of 33% during 23 years of follow-up in women and 35% during 20 years of follow-up in men. The Atherosclerosis Risk in Communities (ARIC) study was the first to include black subjects and followed a cohort of men and women aged 45 to 65 years without CAD or ECG evidence of an MI at baseline. Of the 508 MIs, 20% were unrecognized, with no significant difference observed between men and women. Subsequently, the Rotterdam study group followed up a large cohort of men and women aged >55 years from 1990 to 2000 and demonstrated an incidence of unrecognized MI in men of 33% and in women of 54% of all documented MIs.
Several studies have examined the incidence of unrecognized MI in patients with diabetes mellitus. The Irbesartan Diabetic Nephropathy Trial observed patients with type 2 diabetes mellitus, hypertension, and nephropathy with mean follow-up of 2.5 years and found that 14% of all first nonfatal MIs were unrecognized. The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study followed up patients with diabetes aged 50 to 75 years. The observed incidence of unrecognized MI was 36.8% of all MIs, with a greater proportion of those being inferior infarctions.
The Cardiovascular Health Study (CHS) and Fremantle Diabetes Study reported a 22% and 44% prevalence, respectively, of unrecognized MI at baseline. The CHS study included men and women aged >65 years, and the Fremantle study included a community cohort of patients with diabetes mellitus, a population in which unrecognized MI would be expected to have a greater prevalence.
It is worth noting that most of these large cohort studies likely underestimated the incidence and prevalence of unrecognized MI for several reasons. As discussed previously, the ECG criteria used to classify MI has varied, and those studies that emphasized Q waves will not capture many non–Q-wave MIs. This might represent a significant proportion of MI, because it has been reported that only 15% of patients who present with non–ST-segment elevation MI will develop Q waves. In addition, it has been estimated that the indicators of MI on the electrocardiograms disappear within 2 years in 10% of patients with an anterior infarction and 25% of patients with an inferior infarction. Thus, the diagnosis can be missed if the electrocardiogram is not taken shortly after an MI.