Author + information
- Received January 22, 2018
- Revision received February 12, 2018
- Accepted March 6, 2018
- Published online June 4, 2018.
- Maria Odqvist, MDa,b,
- Per-Ola Andersson, MD, PhDa,b,
- Hans Tygesen, MD, PhDa,b,
- Kai M. Eggers, MD, PhDc and
- Martin J. Holzmann, MD, PhDd,e,∗ (, )@uniofgothenburg@karolinskainst
- aDepartment of Medicine, South Älvsborg Hospital, Borås, Sweden
- bDepartment of Medicine, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden
- cDepartment of Medical Sciences, Uppsala University, Uppsala, Sweden
- dFunctional Area of Emergency Medicine, Karolinska University Hospital, Huddinge, Stockholm, Sweden
- eDepartment of Internal Medicine, Solna, Karolinska Institutet, Stockholm, Sweden
- ↵∗Address for correspondence:
Dr. Martin J. Holzmann, Functional Area of Emergency Medicine, C1:63, Karolinska University Hospital, Huddinge, Stockholm 141 86, Sweden.
Background It remains unknown how the introduction of high-sensitivity cardiac troponin T (hs-cTnT) has affected the incidence, prognosis, and use of coronary angiographies and revascularizations in patients with myocardial infarction (MI).
Objectives The aim of this study was to investigate how the incidence of MI and prognosis after a first MI was affected by the introduction of hs-cTnT.
Methods In a cohort study, the authors included all patients with a first MI from the Swedish National Patient Registry from 2009 to 2013. Cox regression was used to calculate hazard ratios (HRs) with 95% confidence intervals (CIs) for risk of all-cause mortality, reinfarction, coronary angiographies, and revascularizations in patients with MI diagnosed using hs-cTnT compared with those diagnosed using conventional troponins (cTn).
Results During the study period, 47,133 MIs were diagnosed using cTn and 40,746 using hs-cTnT. The rate of MI increased by 5% (95% CI: 0% to 10%) after the introduction of hs-cTnT. During 3.9 ± 2.8 years of follow-up, there were 33,492 deaths, with no difference in the risk of all-cause mortality (adjusted HR: 1.00; 95% CI: 0.97 to 1.02). There were, in total, 15,766 reinfarctions during 3.1 ± 2.3 years of follow-up, with the risk of reinfarction reduced by 11% in patients diagnosed using hs-cTnT (adjusted HR: 0.89; 95% CI: 0.86 to 0.91). The use of coronary angiographies (adjusted HR: 1.16; 95% CI: 1.14 to 1.18) and revascularizations (adjusted HR: 1.13; 95% CI: 1.11 to 1.15) increased in the hs-cTnT group.
Conclusions In a nationwide cohort study including 87,879 patients with a first MI, the introduction of hs-cTnT was associated with an increased incidence of MI, although with no impact on survival. We also found a reduced risk of reinfarction alongside increased use of coronary angiographies and revascularizations.
Cardiac troponin assays, which have been used since the 1990s to diagnose myocardial infarction (MI), have enabled more accurate diagnosis, risk stratification, and clinical decision making than older cardiac biomarkers (1). However, because prolonged serial sampling is often needed to achieve diagnostic accuracy for determining or excluding MI, the effectiveness of using conventional cardiac troponin (cTn) assays has been questioned (2). High-sensitivity cardiac troponin (hs-cTn) assays, which have been increasingly used in clinical practice for some years, allow more rapid rule-in and rule-out of MI compared with cTn assays (3–5).
Nevertheless, it has been suggested that implementation of hs-cTn assays could diminish the clinical specificity for MI diagnosis and potentially lead to increased use of resources. However, in cohort studies of patients with chest pain in the emergency department, the introduction of hs-cTn assays has been associated with small or no increases in coronary angiographies and revascularizations, and a sharp decline in admission rates for chest pain (6–9). In patients admitted to coronary care units, the 1-year mortality rate was unchanged after the introduction of the high-sensitivity cardiac troponin T (hs-cTnT) assay (10). To the best of our knowledge, no large cohort study has investigated the impact of the introduction of hs-cTn assays on the risk of mortality, reinfarctions, and resource use in unselected MI patients. Therefore, we conducted a nationwide study comprising patients with MI during the years when the hs-cTnT assay was adopted widely in Sweden.
We included all patients who were hospitalized for a first MI during the period from 2009 to 2013 in Sweden and were diagnosed using either cTn or hs-cTnT at hospitals where the 99th percentile value was used as the decision limit for MI.
We used the Swedish National Patient Registry, which includes all hospitalizations in Sweden, to retrieve the study population, their characteristics, and outcomes. We then added information about medication from the Prescribed Drug Registry, and deaths from the Cause-of-Death Registry. Information on cTn assays; and the locally applied cutoffs for the diagnosis of MI were obtained from the SWEDEHEART registry (the Swedish Web-based system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies), from local representatives of all hospitals, and by online search using the Internet. The registries and the construction of the database are described in detail in the Online Appendix.
From 2009 to 2012, the current iteration of the hs-cTnT assay (Roche Diagnostics, Basel, Switzerland) was the only higher-precision method used in Sweden. For this assay, the lowest concentration measurable with a 10% coefficient of variation is 13 ng/l, the limit of detection is 5 ng/l and the 99th percentile value among healthy controls is 14 ng/l (11). However, some hospitals in Sweden initially used higher diagnostic cutoffs, in the range of 30 to 40 ng/l, for MI (Online Table 1).
The index date was defined as the date when the patient was hospitalized for MI. Medication at baseline was defined as a minimum of 2 dispensed prescriptions during the 365 days preceding the index date. Medication at discharge was defined as at least 1 dispensed prescription during the 6 months following the index date. International Classification of Disease–9th and 10th Revisions were used to retrieve information about comorbidities (Online Table 2). Diabetes at baseline was defined as ongoing medication with any hypoglycemic agent. In multivariable models, the diabetes definition was extended to include patients who, in addition, had any hypoglycemic agent dispensed 6 months following the index date. The Anatomical Therapeutic Chemical classification codes for medication are listed in Online Table 3.
Follow-up for all-cause mortality started at the index date and ended at the time of death or March 25, 2016, whichever occurred first. For reinfarction, follow-up started at the time of discharge after MI and ended at the date of readmission for a subsequent MI, at the time of death, or December 31, 2014. For coronary angiography and revascularization, follow-up started at the index date and ended at the time of the outcome, at the time of death, or at 30 days after the index date to account for acute event rates. The patient registry is updated only yearly for hospital stays, accounting for the 1-year difference between follow-up for reinfarction and all-cause mortality, which is reported and registered by Swedish authorities within a few weeks of an individual’s death.
The primary outcome was all-cause mortality. Secondary outcomes were: 1) reinfarction; 2) coronary angiography; and 3) coronary revascularization. In the outcomes analyses, we excluded 9,747 patients who were diagnosed with MI at hospitals that used the hs-cTnT assay with a higher decision limit for MI than the 99th percentile value (Online Table 1) or were diagnosed with MI in health care settings other than acute care hospitals. In addition, for reasons of comparison, we calculated the national incidence of MI for the years 2007 to 2013 whereby MIs excluded in the outcomes analyses were also counted.
Baseline characteristics are described as frequencies and percentages for categorical variables, and means and standard deviations for continuous variables. In the Central Illustration, we calculated the number of events 90 days before and after the transition to the hs-cTnT assay. We assumed that the quota of the 2 would be large enough to reflect an estimate of the relative risk. The relative risks were reported for each hospital alone and for all hospitals together, alongside estimated 95% confidence intervals (CIs). Cox proportional hazards models were used to calculate the association, expressed as hazard ratios (HRs) with 95% CIs, between patients tested with cTn (reference), and patients tested with hs-cTnT for all-cause mortality, reinfarction, coronary angiography, and revascularization. For each outcome, 2 models were conducted: unadjusted and adjusted for covariates including age, sex, chronic kidney disease, hospital, heart failure, prior stroke, chronic obstructive pulmonary disease, diabetes, previous cardiac revascularization, and treatment with aspirin, P2Y12 inhibitors, beta-blockers, angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, and statins. Patients were followed until death, emigration, or end of follow-up, whichever came first. The proportional hazards assumption was met. One-year absolute risk was calculated as the proportion of patients that came under risk and experienced the specific outcome within 1 year.
SAS software version 9.4 of the SAS System for MS Windows (SAS Institute, Cary, North Carolina) was used for all statistical calculations.
In total, we included 87,879 patients with a first MI, of whom 47,133 (54%) were diagnosed using cTn and 40,746 (46%) were diagnosed using hs-cTnT (Table 1).
Introduction of the hs-cTnT assay in Sweden
Only 4 of 71 (5.6%) hospitals in Sweden started using the hs-cTnT assay concurrently with the 99th percentile value as the decision limit for MI in 2009 (Central Illustration). Nine (13%) hospitals that introduced the hs-cTnT assay initially used a decision limit of 30 or 40 ng/l for the diagnosis of MI (Online Table 1). Most hospitals underwent the transition from cTn assays to the hs-cTnT assay during 2010 and 2011 (36 [51%]) (Central Illustration). At the end of 2013, 52 (73%) hospitals in Sweden were using hs-cTn assays, with only 1 using the hs-cTnI assay and all others using the hs-cTnT assay.
MI and unstable angina
Despite a growing population, there was a decline in the yearly number of MIs in Sweden, from 33,380 in 2007 to 25,787 in 2013 (Table 2). This decline was attenuated during the years when the hs-cTnT assay was adopted. The national annual MI incidence fell by 27% from 2007 to 2013. There was no change in the proportion of MIs coded in the primary position compared with other positions during the study period (82% vs. 18%). When we compared the 90 days that followed with the 90 days that preceded the introduction of the hs-cTnT assay at each hospital, we found a 5% increase (95% CI: 0% to 10%) in the number of MIs (Central Illustration). However, the adoption of the hs-cTnT assay led to different interpretations in different hospitals, with some having increased numbers of MIs, and others having decreased numbers. The introduction of the hs-cTnT assay did not affect the MI rate in hospitals with a higher decision limit than the 99th percentile value (Online Figure 1). There was a decline in length of hospital stay for MI from 8.8 ± 11.5 days to 7.3 ± 9.1 days from 2007 to 2013. The pattern for trends in number of cases and duration of hospitalization for patients with unstable angina were similar to those for MI patients (Table 2). For reasons of comparison, the numbers of MIs and unstable angina at hospitals that used the hs-cTnT assay with a higher decision limit or settings other than acute care hospitals are shown in Table 2.
During a mean follow-up of 3.9 ± 2.8 years (516,198 person-years), there were, in total, 33,492 (38%) deaths: 20,115 in the cTn group and 13,377 in the hs-cTnT group. When the whole period of 2009 to 2013 was pooled, there was no difference in the risk of all-cause mortality in the hs-cTnT group (adjusted HR: 1.00; 95% CI: 0.97 to 1.02). In sex-specific analyses, no differences in all-cause mortality were found between groups, for both men (adjusted HR: 0.98; 95% CI: 0.95 to 1.01) and women (adjusted HR: 1.02; 95% CI: 0.99 to 1.05). Yearly event rates, 1-year absolute risks, and HRs for all-cause mortality from 2009 through 2013 are shown in Table 3.
During a mean follow-up of 3.1 ± 2.3 years (364,886 person-years), there were, in total, 15,766 (18%) reinfarctions: 9,465 in the cTn group and 6,301 in the hs-cTnT group. There was an 11% relative risk reduction of reinfarctions in the hs-cTnT group compared with the cTn group (adjusted HR: 0.89; 95% CI: 0.86 to 0.91). In sex-specific analyses, a similar risk reduction was found both for men (adjusted HR: 0.88; 95% CI: 0.84 to 0.92) and women (adjusted HR: 0.90; 95% CI: 0.85 to 0.94). Yearly event rates, 1-year absolute risks, and HRs for reinfarction are shown in Table 3. Reinfarctions were diagnosed using the hs-cTnT assay in 17% of cases in the cTn group and in 81% of cases in the hs-cTnT group (Online Table 4).
Coronary angiography, revascularizations, and medication at discharge
When the years 2009 to 2013 were pooled, patients in the hs-cTnT group were 16% more likely (adjusted HR: 1.16; 95% CI: 1.14 to 1.18) to undergo coronary angiography and 13% were more likely (adjusted HR: 1.13; 95% CI: 1.11 to 1.15) to undergo revascularization within 30 days of MI compared with patients in the cTn group (Table 4). When we compared frequencies of coronary angiographies and revascularizations in the 90 days following the introduction of the hs-cTnT assay at each hospital with the 90 days preceding its introduction, we found a 3% (95% CI: 0% to 9%), and 4% (95% CI: 0% to 11%) nonsignificant increase in coronary angiographies and revascularizations, respectively (Online Figures 2 and 3). There were no differences in medication at discharge between groups, except for statins, which were dispensed slightly more often to patients in the hs-cTnT group (Online Table 5).
In this observational cohort study, which included patients with a first MI in Sweden during the period 2009 through 2013, we found an association between being diagnosed using hs-cTnT and a small increase in the incidence of MI. Although no change in mortality was found, we noticed a reduced risk of reinfarction alongside a small increase in coronary angiographies and revascularizations.
After the findings of an increased prevalence of MI from 18% to 22% in patients with chest pain in the emergency department after the shift from cTn to hs-cTnT (12), concerns were expressed that the introduction of hs-cTn assays would be paralleled by an increase in MIs. In a later study in which patients were randomized to testing with either cTn or hs-cTnT, no increase in the number of MIs was reported (13). In another study in which 12,485 patients with chest pain tested with cTnT from 2009 to 2010 were compared with 24,729 patients tested with hs-cTnT from 2011 to 2013, a small reduction in MI prevalence from 5.2% to 4.7% was observed (14). A recent study from the SWEDEHEART registry found that patients admitted to coronary care units had a 5% increased prevalence of MI during the first year the hs-cTnT assay was used (10). We found a similar 5% increase in the number of MIs during the first 3 months the hs-cTnT assay was used. The results from the SWEDEHEART study may not be completely comparable with ours, because only 70% to 80% of all nationwide MIs will be registered in the SWEDEHEART registry (15). Because of the rather sharp decline (by 27%) in MI incidence between 2007 and 2013, which was already evident in the years preceding the first year hs-cTnT was in use, the long-term effect of the transition to the hs-cTnT assay on MI incidence was difficult to evaluate. However, we found an attenuation of the decline in MI incidence during the 3 years when most hospitals introduced hs-cTnT.
Interestingly, in some hospitals, the introduction of the hs-cTnT assay was associated with a relatively large decline, whereas in others there was a relatively large increase in the number of MIs. This may have been related to initial difficulties with interpretation of hs-cTnT levels. Indeed, in a recently published Swedish study, it was found that 60% of patients with a discharge diagnosis of unstable angina had hs-cTnT levels above the 99th percentile value (16). Also, there were large variations in the proportion of patients with MI compared with unstable angina between different hospitals in that study. This emphasizes the need for educational measures paralleling the implementation of hs-cTn assays in order to avoid the misdiagnosis of MI as unstable angina.
The hs-cTn assays were introduced in clinical practice in 2009, after studies found an improved rule-in of MI and risk assessment in patients with chest pain (3,4). Later on, interest turned to the improved and faster rule-out of patients in the emergency department, where algorithms using a second troponin level, analyzed only 1 h after the admission troponin level, were found to be efficient (5,17). Instead of increases in admission rates for chest pain, as had been feared, some studies found no changes in admissions in the short term (7), whereas others found sharp reductions over the first few years of use, (9) without any or with only small increases in coronary angiographies and revascularizations (6–10,13,14).
In our study, which exclusively involved patients with MI, we found no change in mortality, but a reduced risk of reinfarctions in the hs-cTnT group. Furthermore, there were no major differences in medication at discharge between groups. However, there was an increase in coronary angiographies and revascularizations in the hs-cTnT group, which may have been related to the reduced risk of reinfarction. This increase in resource use is of potential concern, not only because it will lead to increased costs, but also because it will expose patients to risks of complications. Indeed, if this increase in coronary angiographies and revascularization were associated with an improved prognosis, the patient would benefit. The magnitude of increase in coronary angiographies and revascularizations associated with the introduction of the hs-cTnT assay was similar to that found in a cohort of chest pain patients in a recently published single-center Swedish study (14). When we compared the 3 months following to the 3 months preceding the introduction of the hs-cTnT assay, we found only a small, nonsignificant increase in the use of coronary angiographies and revascularizations. This may have been related to the rather short time period, which did not allow changes in clinical practice to be implemented. This is supported by findings in a previous study in which admission rates among chest pain patients during the first year after the introduction of the hs-cTnT assay were virtually unchanged, but fell sharply during the second and third years (9). Thus, changes in clinical practice associated with new tests, interventions, or treatments may not be immediate, but may take some time to implement.
The main strength of this study was that the population was retrieved from the National Patient Registry, which allowed us to include all patients with MIs in Sweden during the study period. The large study population and high number of events led to high precision in our estimates. Unlike previous studies that compared time periods before and after the transition to hs-cTn assays, our outcome analyses were direct comparisons between hospitals using cTn and hs-cTnT. By doing this, we reduced the risk of confounding that may have been related to the time period.
We had no information about whether or not patients were cared for in a coronary care unit, which may have affected, not only the length of hospital stay, but also the frequency of coronary angiographies and revascularizations.
As with every observational cohort study, we cannot rule out that there was residual confounding present. However, considering that patients tested with cTn had characteristics virtually identical to those of patients tested with hs-cTnT, and that the testing took place during the same time period, we believe that the risk of residual confounding was minimal. The true magnitude of the change in incidence of MI was difficult to evaluate, as a trend with lower annual incidences of MI was already evident in the years before the introduction of the hs-cTnT assay. Because we knew how large the population at risk was only on a national level, and not on a hospital level, we were not able to estimate incidence rates of the outcomes for each hospital. We could not comment on type of MI because this information is not collected in the registries used for this study. However, we found that there was no association between being diagnosed with hs-cTnT and the proportion of MIs coded in the primary position. As type 2 MIs most often are coded in other positions than the primary position, there would likely be a proportionate increase in diagnosis of type 2 MI as well. Accordingly, we were not able to investigate whether the increase in resource use was mainly attributed to an increase in coronary angiographies and revascularizations in patients with type 2 MI. Furthermore, there may have been misclassification of the reinfarction outcome, mainly in hospitals that used the hs-cTnT assay, where some patients potentially were misdiagnosed with unstable angina instead of MI (16). This might have contributed to a reduced risk of reinfarctions in the hs-cTnT group. However, as this was not paralleled by an increase in the incidence of unstable angina, the overall importance of potential misclassifications on the main results of our analysis appears to be limited. Although differences in medication use after discharge were small, we did not have information about intensity of statin therapy, which may have been associated with the observed reduction in reinfarction.
In a nationwide cohort study comprising 87,879 patients with a first MI during the years when 73% of acute care hospitals in Sweden transitioned from cTn to hs-cTnT, the introduction of the hs-cTnT assay led to a decline in the annual incidence of MI, with no survival benefit, but with a reduced risk of reinfarction, alongside an increase in coronary angiographies and revascularizations.
COMPETENCY IN SYSTEMS-BASED PRACTICE: The introduction of hs-cTn assays for diagnosis of acute myocardial infarction is associated with increased resource utilization including coronary angiography and revascularization and a slight reduction in the risk of re-infarction, but no impact on survival.
TRANSLATIONAL OUTLOOK: As hs-cTn assays become widely available and clinicians gain experience interpreting the results, more work is needed to enhance clinical reasoning and implementation to improve patient outcomes.
Dr. Holzmann holds a research position funded by the Swedish Heart-Lung Foundation (grant 20150603). Dr. Andersson has been a member of the speakers bureaus of Roche, Gilead, and Janssen; and has been a consultant for Abbvie, CTI Bipharma, and GlaxoSmithKline. Dr. Eggers has received honoraria from Abbott Laboratories and AstraZeneca; and has served as a consultant for Abbott Laboratories and Fiomi Diagnostics. Dr. Holzmann has received consultancy honoraria from Actelion and Pfizer. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- cardiac troponin
- hazard ratio
- high-sensitivity cardiac troponin
- myocardial infarction
- Received January 22, 2018.
- Revision received February 12, 2018.
- Accepted March 6, 2018.
- 2018 The Authors
- Thokala P.,
- Goodacre S.,
- Collinson P.,
- et al.
- Sanchis J.,
- García-Blas S.,
- Mainar L.,
- et al.
- Bandstein N.,
- Ljung R.,
- Lundbäck M.,
- Johansson M.,
- Holzmann M.J.
- Chew D.P.,
- Zeitz C.,
- Worthley M.,
- et al.
- Bandstein N.,
- Wikman A.,
- Ljung R.,
- Holzmann M.J.
- ↵SWEDEHEART, Annual Report 2009. Available at: http://www.ucr.uu.se/swedeheart/arsrapport-2017/aeldre-arsrapporter-older-reports/arsrapporter-2007-2010. Accessed September 2, 2017.
- Eggers K.M.,
- Jernberg T.,
- Lindahl B.
- Reichlin T.,
- Twerenbold R.,
- Wildi K.,
- et al.