Declining Severity of Myocardial Infarction From 1987 to 2002
The Atherosclerosis Risk in Communities (ARIC) Study
Background— Death rates for coronary heart disease have been declining in the United States, but the reasons for this decline are not clear. One factor that could contribute to this decline is a reduction in the severity of acute myocardial infarction (MI). We hypothesized that for those patients hospitalized in the Atherosclerosis Risk in Communities (ARIC) Study with acute incident MI, there was a decline in MI severity from 1987 to 2002.
Methods and Results— The community surveillance component of the ARIC Study consisted of tracking residents 35 to 74 years of age with hospitalized MI or fatal coronary heart disease in 4 diverse communities. For incident, hospitalized MI, a probability sample of hospital discharges was validated and an MI classification was assigned according to an algorithm consisting of chest pain, ECG evidence, and cardiac biomarkers. Severity indicators were chosen from abstracted hospital charts validated as a definite or probable MI. With few exceptions, the MI severity indicators suggested a significant decline in the severity of MI during the period of 1987 to 2002. The percent of MI cases with major ECG abnormalities decreased as evidenced by a 1.9%/y (P=0.002) decline in the proportion of those with initial ST-segment elevation, a 3.9%/y (P<0.001) decline in those with subsequent Q-waves, and a 4.5%/y (P<0.001) decline in those with any major Q wave. Maximum creatine kinase and creatine kinase-MB values declined (5.2% and 7.6%; P<0.001, P<0.001 per year, respectively), although in the later years, maximum troponin I values remained stable (1.1%/y decline; P=0.66). The percent with shock declined (5.7%/y; P<0.001), although those with congestive heart failure remained stable. A combined severity score, the Predicting Risk of Death in Cardiac Disease Tool (PREDICT) score, also declined (0.2%/y; P<0.001). Results for blacks paralleled those of the entire group, as did results for women.
Conclusions— Evidence from ARIC community surveillance suggests that the severity of acute MI has declined among community residents hospitalized for incident MI. This reduction in severity may have contributed, along with other factors, to the decline in death rates for coronary heart disease.
Received January 30, 2007; accepted October 17, 2008.
Mortality rates for coronary heart disease (CHD) have been declining in the United States.1 Data from the Atherosclerosis Risk in Communities (ARIC) Study reflect national trends; CHD death rates for men, women, blacks, and whites have all declined, although black men show the least decline.2 Investigators from the Framingham Heart Study, on the basis of adjudicated events, found that CHD death from heart disease decreased by 59% from 1950 to 1999,3 and data from death certificates in Olmsted County, Minnesota, showed that age- and sex-adjusted CHD death had an annual decline of 3.3% from 1970 to 2003 in this population.4
Editorial p 489
Clinical Perspective p 514
Several factors may contribute to the observed decline in the rates of CHD death. Declining incidence may be a contributing factor. National data are not available for trends in incident CHD events; however, the ARIC Study provides validated data from 4 geographically diverse communities showing that the incidence of acute myocardial infarction (MI) remained steady for all groups except blacks, who show an increase.2 With steady incidence, other reasons for the decline in death rates must be investigated. One in particular may be the lessened severity of incident MI. For incident MI, severity may be influenced by both primary prevention efforts, which might result in less severe events from the earliest stages, and acute care, which might preserve cardiac tissue and function after treatment initiation. For recurrent MI, secondary prevention efforts might also play a role. From a public health perspective, it is important to determine specific reasons for the decline in death rates and the individual factors that influence the decline so that effective strategies can be reinforced.
The ARIC Study is well poised to examine the question of whether persons who are hospitalized for acute incident MI have less severe events than in past years. In a previous analysis of ARIC surveillance data, Goff et al5 examined trends in indexes of severity in patients hospitalized for MI from 1987 to 1994. They found mixed results to support a decrease in MI severity. Hemodynamic indicators were stable, ECG indicators were worsening, and enzymatic indicators were improving. With 8 more years of data to 2002, we examined the trends in MI severity to expand on the previous work to provide insight into why death from CHD has decreased in the face of stable incidence. This added time covers a period when there were many changes in diagnostic and treatment patterns in care for MI.6,7 We hypothesized that for adjudicated incident MIs in the ARIC communities, severity of the event, both at presentation and during the hospitalization, has declined over time.
The ARIC Study
The ARIC Study was designed to study cardiovascular disease through 2 methods: a surveillance component in 4 geographically defined communities and a cohort component sampled from those communities. The surveillance component was wholly observational with no participant contact, whereas the participants in the cohort component were contacted regularly by telephone and physical examinations. The present article is based on data collected in the surveillance component.
The community surveillance component of the ARIC Study monitored hospital discharges of acute MI in 35- to 74-year-old participants using targeted International Classification of Diseases (ICD) discharge codes for all residents presenting to area catchment hospitals of the 4 ARIC communities of Forsyth County, North Carolina; Washington County, Maryland; the suburbs of Minneapolis, Minn; and Jackson, Miss. The Jackson community was ≈50% black; the Forsyth community was ≈20% black; and the remaining 2 communities were primarily white. Data presented here cover 1987 through 2002. According to the 2000 US Census Bureau, these regions had a combined population of black and whites 35 to 74 years of age of ≈374 000, of whom 40% were white women, 37% were white men, 13% were black women, and 10% were black men. Forsyth County, geographically the largest community, accounted for 34% of the total ARIC surveillance population. The suburban Minneapolis, Jackson, and Washington County areas accounted for 28%, 19%, and 16% of the total population, respectively. The institutional review boards at the university associated with each field center approved the ARIC study protocol.
A detailed description of the methods used in ARIC community surveillance has previously been reported;8–10 a brief description is given here. Deaths and hospital discharges were validated and classified with standardized criteria.8 A stratified sample selection criterion based on sex, age, residence, date of discharge, and diagnosis code (ICD-9-clinical modification codes 402, hypertensive heart disease; 410 to 414, acute MI, other acute and subacute ischemic heart disease, old MI, angina pectoris, and other forms of chronic ischemic heart disease; 427, cardiac dysrhythmias; 428, heart failure; and 518.4, acute edema of the lung) was used to select hospital records for abstraction. In 1994, the sampling probabilities were adjusted to reduce the variance of event rate estimations while stabilizing the overall number of records requiring abstraction. This approach and the sampling probabilities have previously been detailed.10 Study manuals and forms can be viewed at http://www.cscc.unc.edu/aric/studform.
Determination of Acute Myocardial Infarction
Sampling criteria were used to select cases for investigation and thorough abstraction of medical records. Trained staff abstracted eligible medical records for information on presenting symptoms, the presence of chest pain, medical history, and levels of cardiac biomarkers during the first 4 days after the event.
A computerized algorithm was applied to abstracted information to determine a computer diagnosis of MI. Cases with multiple hospitalizations (within 28 days) were linked and considered a single event. Biomarkers considered were lactate dehydrogenase and its subfractions; creatine kinase (CK) and its myocardial fraction, MB; and troponins I and T. Copies of up to 3 ECGs were made and sent to the University of Minnesota ECG Reading Center for classification according to the Minnesota Code.11 The ARIC Mortality and Morbidity Classification Committee reviewed events with certain discrepancies between hospital discharge codes and computer diagnosis.
All eligible hospitalized events were classified into 1 of 4 categories of acute MI: definite, probable, suspect, and or MI. In ARIC, hospitalized events classified as definite or probable MI are combined.8 The ARIC definition of hospitalized MI includes events that meet one of the following criteria: (1) evolving diagnostic ECG pattern, (2) diagnostic ECG pattern and abnormally elevated cardiac biomarkers (≥2 times the upper limit of normal), (3) cardiac pain and abnormal biomarkers, (4) cardiac pain and slightly elevated biomarkers (between the upper limit of normal and twice the upper limit of normal) and evolving ST-T pattern or diagnostic ECG pattern, and (5) abnormally elevated biomarkers and evolving ST-T pattern. Eight combinations of pain and ECG will place a case in the definite or probable category without abnormal biomarkers (Figure 1).
Severity indicators were chosen from information abstracted from hospital charts and included those examined by Goff et al5 and Jacobs et al.12 Before consideration of the biomarkers CK-MB and troponin, 90% of patients had data for all 18 severity indicators. Considering CK-MB and troponin as well, all patients had data in their chart for at least 10 of 20 possible indicators; 91% had data for at least 18; and 20% had data for all 20 indicators.
Systolic blood pressure and pulse were those first measured on presentation to the hospital. Systolic blood pressure was considered abnormal if the reading was <100 mm Hg. Pulse was considered abnormal if it was <60 or >100 bpm.
ECGs were evaluated using the Minnesota Code, resulting in classification for ST elevation as either initial (first ECG obtained) ST-segment elevation or subsequent (new) ST-segment elevation, initial (first ECG obtained) Q wave, a subsequent (new) Q wave, any major Q wave, any minor Q wave, and diagnostic (of MI) and evolving diagnostic ECG. ST-segment elevation (initial or new), initial or new Q wave, and diagnostic (of MI) categories were defined by standard criteria as outlined in the Minnesota Code. To fit an evolving diagnostic ECG pattern, changes must occur within lead groups (ie, lateral, inferior, or anterior) and be confirmed for all codes by serial ECG comparison. The ECG series is assigned the highest category for which criteria are met; ie, evolving diagnostic is greater than diagnostic ECG.11
We have made a distinction between ST-segment elevation MI (STEMI) and non-STEMI (NSTEMI); however, some research suggests that the morbidity and mortality rates from STEMI and NSTEMI do not differ.13,14 Others have found that STEMI has been associated with a higher likelihood of in-hospital death.15 It also has been suggested that NSTEMI patients do worse than STEMI patients because of differences in baseline characteristics and decreased use of therapies in the acute setting.16,17
Use of biomarker determination of MI has evolved over the duration of the ARIC Study. Initially, CK, CK-MB, and lactate dehydrogenase were used. Beginning in 1996, troponins (both T and I) were used by hospitals in the diagnosis MI. Fewer than 1% of the ARIC hospitals used troponin T; the remaining used troponin I.
The percent of MIs without biomarkers ranged from a low of 3.9% in 1994 to a high of 8.3% in 1987 and 8.2% in 1999. The percent of people with events who had CK-MB, CK/lactate dehydrogenase, or troponin has been >60% since 1998. Use of troponin exceeded 80% in 1998 and has exceeded 90% since 1999. Since 1997, ≈20% of patients with troponins did not have CK-MB, and ≈4% did not have CK. Biomarkers were considered abnormal if they were ≥2 times the upper limit of normal by the standard or reference for each hospital.
As many as 9 measurements of cardiac biomarkers performed on the first 4 days after the event were recorded. The peak was the highest regardless of when it was recorded. If there was a linked event (an event with multiple hospitalizations within 28 days), then all were examined to find a peak.
If a patient had coronary revascularization followed by an abnormal CK or troponin, the classification of the biomarker elevation was downgraded to equivocal. Because troponin was largely unavailable before 1997, it was set to missing before 1997. Troponin, considered more sensitive than CK-MB, has resulted in greater detection of MI.18,19 In the ARIC algorithm, among those with both CK-MB and troponin measures, the addition of troponin biomarkers resulted in 12% more MIs than with CK biomarker alone.
We used a modification of the Predicting Risk of Death in Cardiac Disease Tool (PREDICT) score to create an index of severity. This score was developed as part of the Minnesota Heart Survey by Jacobs et al12 from a community-based study. It was used to provide a simple, long-term admission-day prognostic score for patients hospitalized for MI or unstable angina. The score was evaluated on the basis of the 30-day, 2-year, and 6-year mortality experience. Score components include shock (0 to 4 points), clinical history (MI, stroke, angina; 0 to 2 points), age (0 to 3 points), ECG findings (0 to 3 points), congestive heart failure (0 to 3 points), kidney function (0 to 3 points), and Charlson Comorbidity Index (0 to 6 points) for a maximum severity score of 24 points.12 Among incident MIs in the ARIC community surveillance, a 1-unit change in the PREDICT score was associated with a 17% increase in risk of 28-day death (P<0.0001).
The modification used by the ARIC Study was developed by Watkins et al20 and does not include kidney function because it was not collected in the ARIC Study. The score ranges from 0 to a maximum of 21 points. The PREDICT score has been tested and validated by Singh and coworkers.21 Coding changes over time do not appear to influence the score significantly.12,22
Because of the complex sample design, all analyses were conducted with the SUDAAN package (Research Triangle Park, NC). Sampling fractions outlined by White et al8 were used for events occurring from 1987 through 2002. Trends in dichotomous severity indexes (indexes defined by present/absent status, ie, systolic blood pressure <100 mm Hg, abnormal pulse, abnormal enzymes, adjudicated definite MI, new Q wave, diagnostic ECG, initial ST-segment elevation, subsequent ST-segment elevation, initial Q wave, any major Q wave, any minor Q wave, evolving diagnostic ECG, diagnostic ECG pattern, shock during hospital stay, and congestive heart failure during hospital stay) were examined with weighted Poisson regression with the severity index as the dependent variable and the continuous value of year as the independent variable. Trends in peak CK ratio, peak CK-MB, troponin ratio, and PREDICT score were examined with a weighted linear regression of the log-transformed variable with year as a continuous independent variable. To examine the potential for nonlinear responses over time, analyses were repeated by testing a quadratic time term. All analyses were adjusted for age, sex, and race, and the average annual percent change in continuous indexes (peak CK, CK-MB, troponin, and the PREDICT score) were further adjusted for thrombolytic therapy. Potential sex and race effect modification of trends was tested with an interaction term with year of event.
Associations of MI severity indexes with 28-day death were examined with proportional-hazards models. Each indicator was tested separately after adjustment for age, race, sex, and year of incident MI. A separate model also was run without any MI severity indicators and in a multivariate fashion with selected MI severity indicators to examine the impact on 28-day case fatality trends resulting from changes in MI severity.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agreed to the manuscript as written.
Surveillance Population Characteristics
From January 1, 1987, through December 31, 2002, 17 215 discharges for definite or probable MI were identified in the ARIC Study (24 585 weighted discharges after taking the sample scheme into consideration). Of these discharges, 10 577 were incident events, 5332 were recurrent events, and 1306 could not be classified on MI history. Of the incident events, 244 were excluded because of community/race other than Forsyth County blacks, Forsyth County whites, Jackson blacks, Jackson whites, Minneapolis whites, or Washington County whites. Thirty-two events were excluded for age outside the range of 35 through 74 years; 16 were excluded for missing data regarding the sampling weight or sex. A total of 10 285 (13 713 weighted) discharges for incident definite or probable MI were included in subsequent analyses. Women made up approximately one third of the study population.
The characteristics of the incident MI patients are shown in Table 1. Incidence rates were computed by dividing the weighted number of MIs by the population denominator from census information. Rates specific for sex, race, and age were computed on the basis of dynamic population estimates derived by interpolation from US census data. Women represented 34% of the events in 1987 to 1990 and 37% in 1999 to 2002. Blacks accounted for 15% of the events in 1987 to 1990 and 26% in 1999 to 2002, a statistically significant increase over time of 4.7%/y (95% CI, 3.4 to 6.0) that was observed in both Jackson and Forsyth County. Age- and sex-adjusted incident rates for MI increased for blacks in both Forsyth County and Jackson, whereas MI incident rates increased for whites only in Jackson.
The percent of patients with a history of hypertension increased over time from 57% to 64%; the percent of those with history of diabetes also increased from 26% (1993 to 1994) to 30% (1999 to 2002). Data were not available on diabetes history before 1993, and no data were available on dyslipidemia or preadmission medications. No change was observed in time from onset of symptoms to arrival at the hospital, with approximately two thirds of patients delaying ≥2 hours before hospital arrival throughout the study period. Revascularization within 24 hours of event onset (percutaneous coronary intervention, coronary artery bypass grafting, atherectomy, or thrombolytics) among MIs without ST elevations did not change significantly; however, there was a significant increase in revascularization among patients with an ST-segment elevation (on either the initial or last ECG) that occurred by the 1991 to 1994 period.
Screened codes validated as an MI also showed significant changes over time. The proportion of validated MIs screened from ICD-9 code 410 decreased over time, along with ICD-9 411. Validated MIs from ICD-9 codes 412 to 414 and the other screened codes tended to increase over time (P<0.001).
Treatment modalities for acute incident MI changed significantly over the years. In 1987 to 1990, 24% of patients received thrombolytic therapy, but by 1999 to 2002, this treatment was provided to only 10%. Those receiving percutaneous coronary intervention during the initial hospitalization increased from 16% of the population in 1987 to 90% to 34% in 1999 to 2002. Use of coronary artery bypass grafting was relatively stable during the course of the study.
Severity Indicators: Overall Results
The age-, sex-, and race-adjusted percent of incident MI cases with major ECG abnormalities decreased over time (Table 2 and Figure 2). The decline was significant in all categories of abnormalities and included an average change of −1.9%/y (95% CI, −2.6 to −1.1) in the proportion of events with an initial ST-segment elevation, −3.9%/y (95% CI, −5.2 to −2.5) in the proportion with a new Q wave, and −4.5%/y (95% CI, −5.5 to −3.4) in the proportion with any major Q wave. A statistically nonsignificant decrease in patients with a new subsequent ST-segment elevation was found (−1.5%/y). There was a nonsignificant increase of 2.7%/y in those with left bundle-branch block (P=0.124). By sex, there was an annual increase in left bundle branch of 7.3%/y for women (P=0.009) and a decrease of 1%/y for men (P=0.6510).
Overall, the percent of incident MI cases with abnormal biomarkers declined modestly but statistically significantly (−0.7%/y; 95% CI, −1.0 to −0.3). Maximum CK and CK-MB values declined significantly over time (−5.2%/y and −7.6%/y, respectively), whereas maximum troponin values showed a small decline that did not reach significance. Overall, the percent meeting criteria for definite MI declined significantly from 1987 to 2002 (−1.9%/y; 95% CI, −2.3 to −1.5), so correspondingly, the proportion with probable MI rose. The composite severity score, the PREDICT score, also declined modestly but statistically significantly (−0.2%/y; 95% CI, −0.2 to −0.1).
Previous analysis of the ARIC data suggested that the addition of troponin has resulted in ≈12% more MIs than with CK and CK-MB biomarkers; therefore, we ran our analysis again, this time excluding MIs detected only by biomarkers. The trends were slightly attenuated but remained significant, indicating that use of troponins did not influence our findings (see Tables 1b and 2b of the online Data Supplement).
There were mixed results for hemodynamic parameters. The percentage who had cardiogenic shock at presentation or at any time during the hospitalization declined over time (−5.7%/y; 95% CI, −8.4 to −2.9). Those with congestive heart failure at presentation or during hospitalization trended downward, but this trend did not reach statistical significance. Of note, the total number of MI cases with shock or congestive heart failure was relatively small. The percent with systolic blood pressure ≤100 mm Hg was unchanged, whereas the proportion with abnormal pulse rate (<60 or >100 bpm) at presentation increased moderately. A particular strength of the ARIC Study is that many codes were screened for a diagnosis of acute MI and allowed an accurate measure of case fatality. Table 2 shows that 28-day case fatality appeared to decline with borderline significance (−2.2%/y; P=0.05).
Sex and Race
Sex-specific models of incident MI indicated that the results for men and women largely paralleled the overall results of the study. Potential effect modification was found for trends in the proportion with abnormal enzymes (women, −0.1%/y; men, −1.0%/y; P=0.04 for effect modification). There was possible race effect modification for subsequent Q wave (blacks, −0.8%/y; whites, −4.5%/y; interaction P=0.104) and proportion adjudicated as a definite MI (blacks, −1.1%/y; whites, −2.1%/y; interaction P=0.139). All other indexes of MI severity tended to be largely similar for whites and blacks. Given the number of interactions tested (18 for sex, 18 for race in women, 18 for race in men), these potential interactions may be due to chance.
Several indexes of incident MI severity exhibited a nonlinear trend over time, particularly among the ECG-based indicators. After adjustment for age, sex, and race, the majority of the nonlinear trends showed an increase in the severity indicator early in the study period followed by a decrease. The concave-down pattern in the nonlinear trends suggests that the strictly linear trends tended to underestimate the more recent declines in indexes of MI severity (data not shown). MI ECG severity indicators with a nonlinear trend included ST-segment elevation, initial Q wave, diagnostic ECG, major and minor Q wave, evolving diagnostic ECG pattern, and diagnostic ECG pattern (P<0.01 for nonlinear trend). The proportion with definite MI (Figure 3) also exhibited a nonlinear trend; however, the nonlinear association was not as strong as that observed for the ECG-based indicators (P=0.012 for nonlinear trend).
The proportion with abnormal biomarkers was the only indicator of incident MI severity to show a nonlinear pattern (P<0.0001 for nonlinear trend) in which the proportion initially declined but has since increased (Figure 4). This pattern was found for both men and women and in both blacks and whites. The concave-up pattern suggests that the overall decline in the linear trend does not reflect the more recent pattern of an increase in the proportion with abnormal enzymes. In women, the proportion with abnormal enzymes began to increase at an annual average rate of 1%/y in approximately 1995, whereas in men, an annual average increase of 0.8%/y began around 1997.
Associations of MI Severity Indicators With 28-Day Case Fatality
All indexes of MI severity were associated with 28-day case fatality except abnormal enzymes, definite MI (compared with probable MI), any minor Q wave, and maximum CK-MB (Table 3). Severity indexes with a hazard ratio >2 included a systolic blood pressure <100 mm Hg (hazard rate ratio [HRR], 3.1; 95% CI, 2.4 to 4.1), initial Q wave (HRR, 2.3; 95% CI, 1.8 to 3.0), shock during the hospital stay (HRR, 10.8; 95% CI, 8.5 to 13.7), and congestive heart failure during the hospital stay (HRR, 3.5; 95% CI, 2.8 to 4.3). Although a subsequent ST-segment elevation was significantly associated with higher 28-day death rates (HRR, 1.75; 95% CI, 1.23 to 2.50), an initial ST-segment elevation was associated with lower 28-day death rates (HRR, 0.78; 95% CI, 0.63 to 0.97) in these adjusted models. Before adjustment for indexes of MI severity, the year during which the MI occurred was marginally associated with lower 28-day death rates (HRR, 0.98; P=0.044; data not shown); however, after multivariate adjustment for selected MI severity indicators, the hazard ratio associated with year of MI increased to 0.99 (P=0.34), suggesting that about half of the change over time in 28-day case fatality may be due to the MI severity indicators.
The present observational study provides evidence that the severity of incident hospitalized myocardial infarction has declined over a 16-year period. Strengths of the ARIC Study include the use of multiple codes to screen for acute MI, information on incident events, and information regarding 28-day case fatality. Additionally, the ARIC Study is able to provide meaningful interpretation of data pertaining to blacks.
We found a general decline in indexes of MI severity for patients hospitalized for MI in the ARIC communities over the period of 1987 to 2002. Race- and sex-specific results were consistent with the results seen in all patients. The present report extends on the previous analysis of ARIC data by Goff et al,5 which found mixed support for decreases in severity, and may help to explain declining mortality rates for acute MI.
In their analysis of ARIC data from 1987 to 1994, Goff et al5 found that hemodynamic indicators were stable, enzymatic indicators were improving, and ECG indicators were worsening. With 8 more years of data, we confirm the findings of Goff et al of improvement in many indexes and provide further evidence of declining MI severity. Unlike Goff et al, we found that major ECG indexes, including percent with initial ST elevation, initial Q wave, new Q wave, and diagnostic ECG, were all less common.
Because severity of an acute event can be influenced by previous prevention efforts and acute care, it is important to disentangle these influences as much as possible so that effective strategies can be reinforced and other potentially effective strategies can be reemphasized. The severity indicators included in the present study that were least likely to have been influenced by acute care included initial blood pressure and pulse and ST-segment elevation and Q waves on the initial ECG. The evidence related to these indicators was mixed, with stable prevalence of abnormal blood pressure, increasing prevalence of abnormal pulse, and decreasing prevalence of ST-segment elevation and Q waves on the initial ECG. The evidence regarding other indicators that might be influenced by acute care and previous prevention efforts was more consistent in showing declining severity. These results may be interpreted as providing support for the effects of both previous prevention effort and acute care.
Information from other population studies tends to support our findings. An analysis from a predominantly white cohort in Olmsted County, Minnesota, showed a decline in indexes of MI severity. The proportion of persons with ST-segment elevation declined, as did the occurrence of Q waves and peak CK. Hemodynamic factors as measured by Killip class did not change. The decline in severity of MI was independent of time from first ECG to reperfusion therapy, which changed from 2.3 hours in 1983 to 1986 to 1.7 hours in 1991 to 1994 (P<0.001).23
Spencer et al24 reported on a population-based study with predominantly white participants in Worcester (Mass) from 1975 to 1995. There was a decline in percent with Q-wave MI and anterior MI in those patients who had heart failure after their MI. The proportion of acute MI patients developing heart failure significantly declined from 1975 to 1978 (38%) to 1993 to 1995 (33%). Their findings also suggest a decline in incidence of acute MI and case fatality.
Some studies, however, showed conflicting results. Masoudi et al25 investigated 20 550 Medicare patients hospitalized with acute MI between 1992 and 2001. Patients presenting in later time periods had lower systolic blood pressures and lower heart rates. The proportion with ST-segment elevation decreased, but those with left bundle-branch block increased. Over time, increasing proportions of patients presented with radiographic evidence of heart failure, and more patients experienced shock and heart failure during hospitalization. Reasons for this change could be the greater age of the patients (75 to 78 years versus 59 years in the ARIC communities), fewer details on severity indicators, and fewer years of surveillance.
There is less information on severity differences between blacks and whites. Whittle et al26 looked at a Veterans Affairs population admitted with acute MI or unstable angina between 1989 and 1995. Among patients with MI who underwent coronary angiography during admission, blacks were more likely than whites to have no significant coronary obstructions, although no information is provided on other severity indicators.
One of the strengths of the ARIC population is that the communities represent 4 different geographic areas and a large portion of blacks. In the present study, we are able to provide data on time trends in blacks and whites. Our findings give strong evidence that incident MI severity has declined in blacks and in whites.
Several reasons may be contributing to the reduction in severity, including previous prevention efforts and improvements in acute care. Increasing knowledge about the benefits of primary prevention may have influenced the use of aspirin, β-blockers, and statins. Patients already using these regimens who go on to have an MI may present with a less severe MI.27,28 In the ARIC Study, the use of aspirin in the setting of hospitalized non-ST-elevation acute coronary syndrome increased from 48% of patients in 1987 to 92% in 2000. There is no information on use before hospital presentation or whether increased hospital use is associated with greater use in the community.20 In the United States, prevalence of hypercholesterolemia has decreased and use of cholesterol-lowering medication has increased.29,30 Although we do not have information on dyslipidemia, there was an increase in the proportion with a history of hypertension and diabetes. These increases did not appear to affect severity trends significantly. Of note, these increases may simply reflect that more patients were being diagnosed and treated for these risk factors. The treatments, especially for hypertension, may have included more use of β-blockers and angiotensin-converting enzyme inhibitors. Both of these medications can reduce the severity of MI.27,28 Because this report was from a surveillance study, the age of the participants was constrained and remained constant, averaging 59 years, which eliminated the influence of increasing age seen in the study by Masoudi et al.25
Severity may be influenced by the time frame within which a patient seeks treatment; however, prehospital delay does not appear to have improved. There has been little progress in the time between onset of symptoms and arrival to hospital or in the time from call to 9-1-1 to definitive treatment in the United States.31,32 In the ARIC Study, no significant decline was seen in the proportion of patients with prolonged prehospital delay from 1987 to 2000.33
Improvements in hospital care also appear to be a factor. There was a rapid increase in evidence-based therapies, interventions, and medications for acute MI over the period of the present study. Over time, ARIC community patients underwent more percutaneous coronary interventions and fewer coronary artery bypass graft surgeries and thrombolytics from 1987 to 2002. This shift could influence severity as reflected in such markers as peak CK or troponin, development of new Q waves, shock, and congestive heart failure. Other data from ARIC has been published investigating the trends in use of diagnostic testing during hospitalization for acute MI from 1987 to 2001. These data showed an increase in use of angiography, doubling from 34% in 1987 to >63% in 2001. Echocardiography also increased, as did right-heart catheterization.34 Masoudi et al25 found increases in coronary angiography and percutaneous coronary intervention. Rogers and colleagues35 investigated temporal trends from 1990 to 1999 and found an increase in angioplasty and use of β-blockers, aspirin, and angiotensin-converting inhibitors for acute MI.
Could reduced severity be an artifact explained by the change in biomarkers? With the use of troponins, which may be more sensitive than CK and CK-MB, less severe MIs may be differentially ascertained. However, in the ARIC Study, the algorithm for MI is strongly influenced by ECG criteria. There are 8 combinations of pain and ECG that will place a case in the definite or probable category without abnormal biomarkers. The ARIC Study requires twice the upper limit of normal to qualify as a positive troponin, which also helps to control for the added sensitivity of troponin. In a separate analysis excluding biomarker-driven MIs, we found that associations were attenuated but largely unchanged, suggesting that the change to troponin has not significantly influenced severity.
Screened codes validated as an MI also showed significant changes over time, with the proportion of validated MIs screened from ICD-9 code 410 and 411 decreasing over time and those from codes 412 to 414, as well as the other screened codes, tending to increase over time (P<0.001). It is unclear whether this shift to codes other than 410 and 414 reflects decreasing severity. Of note, Rosamond et al36 looked into the impact of variation in hospital discharge diagnosis. They analyzed 154 836 target codes that were screened for acute MI in the ARIC communities between 1987 and 2000 and found that the validity of the ICD-9 code 410 for identifying acute MI patients was stable.
The specific question as to which factor, better prevention or better hospital care, may play a role and to what extent has been investigated by others. Unal et al37 found declining CHD death rates in England and Wales between 1981 and 2000, when rates decreased by 62% in men and 45% in women. They concluded that 42% of the decline was attributed to medical and surgical treatments, including that for acute MI. In addition, 58% of the decrease in total mortality rate was thought to be due to change in risk factors, especially smoking, cholesterol levels, and blood pressure, which may influence MI severity and case fatality, as well as incidence.37 In a computer-simulation state-transition model of CHD, Hunink et al38 found that about one quarter of the decline in mortality rates in the United States from 1980 to 1990 was due to primary prevention. Most of the decline was explained by improvements in the management of patients with diagnosed CHD; however, the authors were not able to fully separate the impact of a potential reduction in CHD severity caused by prevention efforts from the impact of improved CHD management on postdiagnosis survival.
Two very recent studies also have attempted to explain the decline in CHD death. Using data from the Global Registry of Acute Coronary Events (GRACE) Study, Fox et al39 found a decline in rates of death and heart failure in patients with acute coronary syndromes. Patients had a significant decrease in the rates of in-hospital death, cardiogenic shock, recurrent MI, and heart failure. The authors attribute the decline to change in practice of both pharmacological and interventional treatments in these patients. Ford et al40 used a statistical model to determine how much of the decrease in US deaths from coronary disease could be explained by medical and surgical treatments as opposed to changes in cardiovascular risk factors. They found that approximately half of the decrease may be attributable to reductions in major risk factors and half to evidence-based medical therapies.
The present study reflects data garnered in the surveillance component of the ARIC Study. No direct contact was made with any participant in the community. Typical of large-scale community surveillance studies, information is limited to what can be abstracted from hospital records. We lacked other helpful information such as preadmission medications and preexisting medical conditions. In addition, we were not able to extract complete and detailed information about all medical therapies during hospital admission; these therapies also may have influenced some severity indicators.
We did not have the benefit of several specific severity indicators such as left ventricular ejection fraction. These measures are common now but were much less so in 1987 when ARIC started. Many of our severity indexes are therefore “proxy” indicators of disease severity. We cannot gauge the severity of MIs that occurred out of the hospital and resulted in sudden cardiac death. Data from Framingham and the greater Worcester (Mass) population show that out-of-hospital death rates are declining, perhaps representing declining severity of the initial event so that more people are now surviving to present to hospitals.39,41
Comparison with results from other studies is difficult because of the use of different severity indicators, various definitions of MI, and differing demographics and clinical characteristics. In addition, methods used to abstract and verify clinical data may differ from study to study.
Finally, analyses of racial and sex subgroups (ie, black women) are limited by smaller numbers. Data obtained from these groups in our study generally reflected those of the entire group, but precise differences could not be quantified.
Data from the community surveillance component of the ARIC Study show that the severity of hospitalized incident MI has declined from 1987 to 2002. Biomarker and ECG indicators improved in both whites and blacks, as did the prognostic PREDICT score. Hemodynamic indicators also showed some improvement. This reduction in severity would most likely affect, along with other factors, the decline in population-wide death rates for CHD.
Because most of the severity indicators were evaluated during hospitalization rather than at presentation, it is difficult to determine the extent to which previous prevention efforts and acute care influenced severity. Attributing the reduction in severity to specific causes will be an important next step so that effective strategies can be reinforced. Our data, taken together with recent studies13,41 looking at CHD death, suggest that there has been a favorable change in the status of and prognosis for patients with acute MI. Because there has been no improvement in prehospital delay, efforts to address this obstacle to treatment are needed. Finally, ongoing high-quality surveillance programs are needed to monitor progress in reducing morbidity and death from CHD.
The authors thank the staff and participants of the ARIC Study for their important contributions.
Sources of Funding
The ARIC Study is carried out as a collaborative study supported by the National Heart, Lung, and Blood Institute contracts NO1-HC-55015, NO1-HC-55016, NO1-HC-55018, NO1-HC-55019, NO1-HC-55020, NO1-HC-55021, and NO1-HC-55022.
Dr Myerson was an employee of the National Heart, Lung, and Blood Institute during the initial preparation of this manuscript. Dr Myerson served on the speakers’ bureau for GlaxoSmithKline and Daiichi Sankyo and received a research and educational grant from Abbott Laboratories. The other authors report no conflicts.
American Heart Association. Heart Disease and Stroke Statistics—2006 Update. Dallas, Tex: American Heart Association; 2006.
Rosamond WD, Folsom AR, Chambless LE, Wang C-H. Coronary heart disease trends in four United States communities: the Atherosclerosis Risk in Communities (ARIC) Study 1987–1996. IntJ Epidemiol. 2001; 30: S17–S22.
Fox CS, Evans JC, Larson MG, Kannel WB, Levy D. Temporal trends in coronary heart disease mortality and sudden cardiac death from 1950 to 1999: the Framingham Heart Study. Circulation. 2004; 110: 522–527.
Gerber Y, Jacobsen SJ, Frye RL, Weston SA, Killian JM, Roger VL. Secular trends in deaths from cardiovascular diseases. Circulation. 2006; 113: 2285–2292.
McGovern PG, Jacobs DR, Shahar E, Arnett DK, Folsom AR, Blackburn H, Luepker RV. Trends in acute coronary heart disease mortality, morbidity, and medical care from 1985 through 1997. Circulation. 2001; 104: 19–24.
Lucas FL, DeLorenzo MA, Siewers AE, Wennberg DE. Temporal trends in the utilization of diagnostic testing and treatments for cardiovascular disease in the United States, 1993–2001. Circulation. 2006; 113: 374–379.
White A, Folsom AR, Chambless LE, Sharret AR, Yang K, Convill D, Higgins M, Williams OD, Tyroler HA. Community surveillance of coronary heart disease in the Atherosclerosis Risk in Communities Study: methods and initial two years’ experience. J Clin Epidemiol. 1996; 49: 223–233.
ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. Am J Epidemiol. 1989; 129: 687–702.
Jacobs D, Kroenke C, Crow R, Deshpande M, Gu DF, Gatewood L, Blackburn H. PREDICT: a simple risk score for clinical severity and long-term prognosis after hospitalization for acute myocardial infarction or unstable angina: the Minnesota Heart Survey. Circulation. 1999; 100: 599–607.
Montalescot G, Dallongeville J, Van Belle E, Rouanet S, Baulac C, Degrandasart A, Vicaut E, for the OPERA Investigators. STEMI and NSTEMI: are they so different? 1 Year outcomes in acute myocardial infarction as defined by the ESC/ACC definition (the OPERA Registry). Eur Heart J. 2007; 28: 1409–1417.
Allen LA, O'Donnell CJ, Camargo CA Jr, Giugliano RP, Lloyd-Jones DM. Comparison of long-term mortality across the spectrum of acute coronary syndromes. Am Heart J. 2006; 15: 1065–1071.
Muller-Bardorff M, Hallermayer K, Schroder A, Ebert C, Borgya A, Gerhardt W, Remppis A, Zehelein J, Katus HA. Improved troponin T ELISA specific for cardiac troponin T isoform: assay development and analytical and clinical validation. Clin Chem. 1997; 43: 458–466.
Singh M, Reeder G, Jacobsen SJ, Weston S, Killian J, Roger VL. Scores for post-MI risk stratification: a community perspective. Circulation. 2002; 106: 2309–2314.
Hellerman J, Reeder G, Jacobsen S, Weston SA, Killian JM, Roger VL. Longitudinal trends in the severity of acute myocardial infarction: a population study in Olmsted County, Minnesota. Am J Epidemiol. 2002; 156: 246–253.
Spencer FA, Meyer T, Goldberg R, Yarzebski J, Hatton M, Lessard D, Gore JM. Twenty year trends (1975–1995) in the incidence, in-hospital and long-term death rates associated with heart failure complicating acute myocardial infarction: a community-wide perspective. J Am Coll Cardiol. 1999; 34: 1378–1387.
Masoudi FA, Foody JM, Havranek EP, Wang Y, Radford MJ, Allman RM, Gold J, Wiblin RT, Krumholz HM. Trends in acute myocardial infarction in 4 US states between 1992 and 2001. Circulation. 2006; 114: 2806–2814.
Preventing heart attack and death in patients with atherosclerotic cardiovascular disease: 2001 update. Circulation. 2001; 104: 1577–1579.
Preventive cardiology: how can we do better? Proceedings of the 33rd Bethesda Conference, Bethesda, Maryland, USA, December 18, 2001. J Am Coll Cardiol. 2002;40:580–651.
Arnett DK, Jacobs DR, Luepker RV, Blackburn H, Armstrong C, Claas SA. Twenty-year trends in serum cholesterol, hypercholesterolemia, and cholesterol medication use. Circulation. 2005; 112: 3884–3891.
Luepker R, Racztnski J, Osganian S, Goldberg RJ, Finnegan JR Jr, Hedges JR, Goff DC, Eisenberg MS, Zapka JG, Feldman HA, Labarthe DR, McGovern PG, Cornell CE, Proschan MA, Simons-Morton DG. Effect of a community intervention on patient delay and emergency medical service use in acute coronary heart disease. JAMA. 2000; 284: 60–67.
Pearte CM, Myerson M, Coresh J, McNamara RL, Rosamond W, Taylor H, Manolio TA. Variation and temporal trends in the use of diagnostic testing during hospitalization for acute myocardial infarction by age, gender, race, and geography (the Atherosclerosis Risk in Communities Study). Am J Cardiol. 2008; 101: 1219–1225.
Rosamond WD, Chambless LE, Sorlie PD, Bell EM, Weitzman S, Smith JC, Folsom AR. Trends in sensitivity, positive predictive value, false-positive rate, and comparability ratio of hospital discharge diagnosis codes for acute myocardial infarction in four US communities, 1987–2000. Am J Epidemiol. 2004; 160: 1137–1146.
Unal B, Critchley JA, Capewill S. Explaining the decline in coronary heart disease mortality in England and Wales between 1981 and 2000. Circulation. 2004; 109: 1101–1107.
Fox CS, Evans JC, Larson MG, Kannel WB, Levy D. Temporal trends in coronary heart disease mortality and sudden cardiac death from 1950 to 1999. Circulation. 2004; 110: 522–527.
Death rates for coronary heart disease have been declining in the United States. From a public health standpoint, it is important to identify factors that are contributing to this decline. We investigated whether a reduction in severity of myocardial infarction may contribute to lower death rates. We tracked residents 35 to 74 years of age in the community surveillance component of the Atherosclerosis Risk in Communities (ARIC) Study who were hospitalized with acute incident myocardial infarction and looked at severity indicators abstracted from hospital charts. Included in these were ECG, biomarker, and hemodynamic indicators. With few exceptions, the myocardial infarction severity indicators suggested a significant decline in the severity of myocardial infarction during the period of 1987 to 2002. This reduction in severity may have contributed, along with other factors, to the decline in death rates. There may be several reasons for a decline in severity of incident myocardial infarction; the findings from the present study suggest that 1 factor may be better primary prevention and support ongoing research to determine how preventive care may further reduce coronary heart disease death.
Clinical trial registration information—URL: www.clinicaltrials.gov. Unique identifier: NCT00005131. Public ARIC Website—URL: http://www.cscc.unc.edu/aric/.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.107.693879/DC1.