Guidelines for defining myocardial infarction (MI) have long recommended that events be classified using a combination of the clinical presentation, electrocardiogram (ECG) or cardiac imaging findings, and biomarkers [1–4]. Troponin is the preferred biomarker due to its high specificity for cardiac muscle [1–4]. The risk for mortality and cardiovascular events is increased with even modest troponin elevations [5–9]; therefore, for several years, American and European cardiology societies, the National Heart, Lung and Blood Institute, the Centers for Disease Control and Prevention, and others have recommended defining abnormal troponin as any value above the 99^th percentile for healthy individuals [1–4, 10–14], culminating in the publication of the Universal Definition of MI in 2007 [1–3].

A problem acknowledged by the Universal Definition and other preceding guidelines is that some assays may not be reliable at levels as low as the 99^th percentile. Troponin levels in normal reference populations are very close to zero, and recent advances in biomarker assays permit detection of levels as low as 0.006 μg/L [15]. However, precision at very low levels may not be acceptable [15]. Therefore, until assays achieve greater precision at very low levels and become more standardized, experts have recommended defining the threshold for abnormal troponin as either the 99^th percentile of a normal reference population or the level at which the assay achieves acceptable precision, defined by a coefficient of variation (CV) of 10% or better [1–4, 10–14]. All values of troponin above this threshold are recommended to be considered myocardial necrosis, which, together with a characteristic rising and/or falling pattern, is used in clinical decision-making to classify an event as an MI.

In order to apply this recommendation, the 99^th percentile or 10% CV level must be clearly identified on hospital laboratory troponin reports and labeled as the decision threshold. However, many hospitals have for years reported an “indeterminate range,” which was useful in interpreting troponin relative to creatine kinase (CK)-MB fractions when troponin was first introduced. The upper limit of this indeterminate range was the World Health Organization (WHO) cut-point corresponding to abnormal CK-MB, and many reports continue to provide interpretive guidelines recommending that troponin values above this WHO cut-point represent a high likelihood of MI. Clinicians interpreting such reports today may be challenged to identify the more current, lower guideline-recommended decision threshold. Furthermore, as ST elevation MI is declining while non-ST elevation MI remains common [16, 17], troponin levels are increasingly the deciding factor in determining whether acute coronary syndrome (ACS) is classified as an MI. In this context, the troponin laboratory report is critical for appropriately classifying and risk stratifying MI events. However, guideline concordance in hospital laboratory troponin reports has to our knowledge not been examined.

We studied US hospital laboratory troponin practices in the REasons for Geographic And Racial Differences in Stroke (REGARDS) study, a large national prospective cohort study. We also examined mortality risks associated with very low, or ‘microsize,’ troponin peak MIs relative to MIs with higher peak troponin levels to both confirm the risks that have been described in other populations and to highlight the potential clinical relevance of hospital troponin reporting practices.

The REGARDS study is prospectively following 30,239 individuals for cardiovascular events and mortality to better understand regional and racial influences on stroke and MI mortality. Details of the study are described elsewhere [18]. Briefly, recruitment was conducted from 2003–2007 using commercially available lists and a combination of mail and telephone contact to recruit English-speaking, community-dwelling adults aged 45 and older living in the 48 contiguous US. Baseline data collection included telephone surveys and in-home exams, and living participants are telephoned every six months and asked if they were hospitalized with subsequent medical records retrieval. Deaths were detected from reaching next of kin at a scheduled follow-up, online sources (e.g., Social Security Death Index), or the National Death Index. Medical records were reviewed by a team of experts, who adjudicated MI events using a standardized approach modeled on major epidemiologic studies [4]. The study protocol was reviewed and approved by the University of Alabama Institutional Review Board, and all participants provided informed consent.

We included the first hospitalization for ACS for each participant occurring between 2003 and 2009. ACS was defined as an urgent presentation of signs and symptoms suggestive of acute coronary ischemia resulting in hospitalization. For each case, we examined the hospital laboratory troponin report for documentation of the 99^th percentile for healthy adults or the 10% CV level for the assay utilized and whether either was clearly identified as the decision threshold for abnormal troponin. We also examined whether an indeterminate range was provided, a practice that is no longer recommended [1].

Expert adjudicators classified cases as ACS without MI or ACS with MI following approaches used in most large epidemiology studies [4]. For MI, medical records were examined for the presence of signs or symptoms suggestive of ischemia; a rising and/or falling pattern in cardiac troponin or creatine phosphokinase-MB over ≥6 hours with a peak value ≥ twice the upper limit of normal; and ECG changes consistent with ischemia or MI, guided by the Minnesota code [4]. By convention, adjudicators used as the troponin decision threshold the 99^th percentile or 10% CV value if clearly identified on the report, or if not clearly identified, twice the lowest listed upper limit of normal (ULN). MIs were adjudicated as definite, probable, or possible, and only definite or probable MI events were included in this study as MI events. The highest and lowest levels of troponin and upper limits of normal were also recorded.

After adjudication, cases were further classified into microsize MI or usual MI based on peak troponin levels. Microsize MI was defined as an adjudicated MI with peak troponin no more than five times the lowest listed ULN, and usual MIs were all other adjudicated MIs. We examined clinical characteristics of individuals and in-hospital, 28-day and 1-year age, sex, and race-adjusted mortality in each of the three ACS groups of no MI, microsize MI and usual MI.

We found major deviations from guideline recommendations for troponin laboratory reporting in our study. Laboratory reports only rarely identified a 99^th percentile or 10% CV level, and more than half of hospitals identified an indeterminate range classifying MI based on a higher threshold than the guideline-recommended approach [1–3, 20]. In this context, many microsize MIs may be difficult to recognize clinically.

Although our study was not able to examine the proportion of microsize MIs missed clinically by treating physicians as a result of hospital reporting practices, it is likely that some of these events were in fact not treated as MI. The potential misclassification of microsize MIs as “normal” has significant long-term prognostic implications. The higher long-term mortality risk we observed for those suffering microsize MIs was similar to past reports in which the in-hospital risks were low, but by one month and one year were similar to those with usual MIs [21, 22]. This finding underscores both the importance of long-term risk factor management in patients with microsize MIs and the relatively low short-term risks [1–4, 21, 23–26].

A recent report from Scotland illustrates the clinical importance of detecting microsize MIs [22]. Investigators at the Royal Infirmary in Edinburgh took advantage of the introduction of a more sensitive troponin assay to study clinical management and outcomes among individuals analogous to our microsize MI group. In their study, the focus was on individuals whose peak troponin elevations were above the threshold for the new assay (0.05) and below the threshold for the older, less sensitive assay (0.20), both before and after changes in lab reports. They found that, after introduction of the lower decision threshold on lab reports, secondary preventive measures increased by 15-30% in the microsize MI group but remained similar for those in the usual MI or no MI groups. They also reported reductions in clinical outcomes, including the rate of MIs and deaths at 3 and 12 months in the group with microsize MIs after introduction of the lower threshold on lab reports. While this is an observational study from a single site, it does provide evidence of the potential impact of appropriate troponin reporting to optimize clinical management.

These findings together with our results suggest that improved reporting of hospital troponin results in the US is warranted; the US now has a system of public accountability for the quality of health care that provides a ready infrastructure to enact rapid change. A quality indicator assessing whether the recommended decision threshold is clearly identified on laboratory reports and assessing whether indeterminate ranges are used could be added rapidly to standards for accreditation by the Joint Commission or to the Centers for Medicare and Medicaid Services’ pay-for-performance programs, such as the Premier Hospital Quality Incentive Demonstration [27, 28].

Our study included a large proportion of black participants, which has not been the case in previous reports stemming from non-US samples [21]. Our finding that black participants had 62% higher odds for microsize MIs raises concerns that some microsize MIs among black individuals may not be recognized clinically. Indeed, the last decade has demonstrated a concerning widening of the CHD mortality disparity between black and white people in the US, with white mortality rates declining faster than black rates [29]. Similar to past reports, we found that those with a past history of CHD were also more likely to be in the microsize MI group [21]. The clinical consequences of overlooking a microsize MI in individuals with a history of CHD may be minimized because many such patients may already be receiving secondary prevention, including aggressive smoking cessation counseling, lipid management, beta blockers, or anti-platelet therapy. On the other hand, as many as half of individuals on chronic disease preventive therapies discontinue their medications [30–32]; thus, an overlooked microsize MI may translate into a missed opportunity to redouble secondary preventive efforts in these individuals.

Our study’s limitations include our inability to examine how many cases classified as microsize MI by our rigorous adjudication methods were also recognized as MI by the treating physicians. We also did not have available the 99^th percentile or 10% CV value at each hospital at the time of the event. Like other large epidemiology studies, we had to rely on clinical laboratories from many hospitals to classify MI and, given the variations we described, it is possible that, while state-of-the art and rigorous, the adjudication process under-detected some microsize MIs. This challenge would bias our results towards finding less difference in mortality rates than may actually exist; thus, the mortality risks we describe may underestimate the true risks. Our team of experts only classified events as MI if there was a clear rising and/or falling pattern of troponin, and perennial low level elevations seen in conditions such as heart failure or renal failure would therefore not have been classified as MI; however, it is possible that some misclassification could have occurred. Unfortunately, this study was not designed to detect differences in clinical management in microsize and usual MIs, but recent reports suggest that this may be an important issue [33]. Future studies should examine what proportion of microsize MIs were not managed as acute MIs. Another limitation is the context of a national epidemiologic cohort study; thus, participants may not be representative of the US population. Although we included 649 hospitals, they may not be representative of all US acute care hospitals. In addition, event accrual may permit future analyses of CHD or cardiovascular disease deaths, in contrast to the all-cause mortality we used in this study.

Many US hospital laboratories did not report 99^th percentiles or 10% CV levels in their troponin reports as recommended by national and international guidelines, complicating the detection of microsize MIs. This may result in misclassification of some individuals with MI, preventing optimal clinical management. Our study confirms the excess risks associated with microsize MI described in other populations, emphasizing the importance of recognizing these cases clinically. Prompt attention to improving the quality of hospital laboratory reporting of patient troponin results is warranted.