Myocarditis Incidence, Complications and Mortality in Sweden: A Retrospective Registry Study From 2000-2014

Objectives: Investigate trends in incidence and prognosis of myocarditis in Sweden during 2000-2014. Background: Myocarditis is an inammatory heart disease, with scarce data concerning incidence and prognosis. Methods: Linking Swedish National Patient and Cause of Death Register, we identied individuals ≥ 16 years with rst-time diagnosis of myocarditis during 2000-2014. Reference population, matched for age and birth year (n=16 622) was selected from the Swedish Total Population Register. Results: Among the 8 679 cases, (75% men, 64% <50 years), incidence rate/100 000 inhabitants rose from 6.3 to 8.6, mostly in men and those <50 years. Incident heart failure/dilated cardiomyopathy occurred in 6.2% within 1 year after index hospitalization and in 10.2% during 2000-2014, predominantly in those ≥ 50 years (12.1% within 1 year, 20.8% during 2000-2014). In all, 8.1% died within 1 year, 0.9% (<50 years) and 20.8% ( ≥ 50 years). Hazard ratios (adjusted for age, sex) for 1-year mortality comparing cases and controls were 4.00 (95% condence interval 1.37-11.70), 4.48 (2.57-7.82), 4.57 (3.31-6.31) and 3.93 (3.39-4.57) for individuals aged <30, 30-<50, 50-<70, and ≥ 70 years, respectively. Conclusion: The incidence of myocarditis during 2000-2014 increased, predominantly in younger men. One-year mortality in the young was low, but fourfold higher compared with reference population.


Introduction
Myocarditis is an in ammatory disorder of the heart muscle, with either focal or diffuse involvement, and with different causes, clinical presentation and outcome 1 . Compared with other heart conditions, myocarditis is uncommon in Western populations, but also potentially frequently underdiagnosed.
Community-acquired myocarditis is considered to be predominantly of viral etiology, although con rmatory proof of this is often di cult to obtain 1 . No comprehensive data on the incidence of myocarditis in the population exist, but the proportion of myocarditis in various clinical populations has been reported 1,2 . For example, myocarditis was reported in 20% of sudden cardiac deaths in young military recruits 3 . In addition, it was one of the most common speci c causes in 1 278 patients referred to the Johns Hopkins Hospital with dilated cardiomyopathy (DCM) 4 .
Likewise, prognosis in myocarditis is poorly understood, especially for long-term survival and later development of other cardiac disorders such as cardiomyopathy or heart failure (HF). Biopsy-proven myocarditis is associated with a poor prognosis in up to 30% of the patients 1-10 , whereas mild forms of myocarditis are, in many patients, believed to resolve spontaneously 11 . Sweden has a universal healthcare system that provides low-cost hospital care to all Swedish residents, with all hospitals reporting discharge diagnoses to the Swedish National Inpatient Register and all causes of death reported to the National Death Register.
In full recognition of the many di culties and limitations inherent in the use of administrative data for this diagnosis, we identi ed all patients with a discharge diagnosis of acute myocarditis from 2000 to 2014 in an attempt to estimate trends in incidence and prognosis of myocarditis in Sweden.

Swedish National Patient Register and National Cause of Death Register
Registration of hospital discharge diagnoses and causes of death, coded in accordance with the International Classi cation of Diseases (ICD) system, is mandatory in Sweden, with national coverage of all discharge diagnoses since 1987. All individuals aged ≥ 16 years diagnosed with or who died from myocarditis (ICD codes I40 and I51.4) from 2000 to 2014 were identi ed from the Swedish National Patient Register (NPR) and the Cause of Death Register (NDR), and paired with two controls without myocarditis, matched for age and sex, from the Swedish Population Register. The NPR was used to retrieve information about diagnoses before, during, and after, the index hospitalization.
Data from the NPR and the NDR for cases and controls are linked to personal identity numbers, unique to each Swedish resident. This provides the possibility to combine registries and get exact data on prior disease, comorbidities and cause of death. We estimated 1-year incidence and mortality through the NPR and the NDR and compared these estimates to controls. Data for the whole population of Sweden were used to obtain standardized mortality rates.

Diagnostic validation
We performed a diagnostic validation of hospitalized cases diagnosed with myocarditis between 2000 and 2014. This include retrieving all electronic case records for all patients with a diagnosis of myocarditis, according to the ICD-10 at the Sahlgrenska University Hospital, Gothenburg, Sweden, during the study period. Two physicians reviewed manually the medical journals to verify the diagnosis of myocarditis at baseline. The diagnosis was considered incorrect if no criteria in the medical journals supported it (see a-c below). When con icting information, the diagnosis was deemed correct based on the following criteria, hierarchically. a) The clinical opinion of the treating physician who had done the documentation in the medical journal at the time when the diagnosis was established was that the it was a myocarditis; b) There was either biopsy veri ed myocarditis or imaging veri ed myocarditis (or both); c) There were typical symptoms and biomarkers corresponding to myocarditis 1 . In total, 507 cases (78.5% men) were systematically reviewed out of whom 421 (83.0%) could be veri ed as acute myocarditis. The Swedish Ethical Review Authority approved the study (Dnr: 026 − 16 and T826-17), and the study complies with the Declaration of Helsinki. The informed consent of the participants was waived by the Swedish Ethical Review Board because the data was coded, thus anonymous.

Statistical analysis
All statistical analyses and data management were performed using SAS 9.3 (SAS Institute, Cary, NC, USA). Graphics were drawn in SAS 9.3 and R 3.1.3 (RStudio, Boston, MA, USA). Differences in baseline characteristics between the study groups were evaluated by the t test as well as the Wilcoxon nonparametric test for continuous variables and the chi-square test for dichotomous data. Standardized rates of incidence of myocarditis were estimated per 100 000 individuals and mortality rates per 1 000 person-years. Trends in incidence and mortality were determined by linear regression. The general difference in rates was tested as the point estimate under the assumption of the Poisson distribution. The Cochran-Armitage trend test veri ed changes in the proportion of deaths due to cardiovascular disease (CVD) during the study period. Variation in follow-up times was assessed with Kaplan-Meier survival analysis. A Cox proportional hazard model adjusted for age, sex and period of the study, and the log-rank test were used to compare hazard ratios (HRs). The signi cance level was set at p < 0.05.
The odds ratio for having a veri ed myocarditis case was calculated for year of hospitalization, age and sex. The univariate logistic regression showed that the probability of being assigned a valid diagnosis increased by 15% (95% con dence interval [CI]: 1.09-1.22) for each year since hospitalization, decreased by 8% with each year of age (95% CI: 0.91-0.94) and was twice as high for men (95% CI: 1.25-3.70). The sex difference disappeared when an interaction term of age and sex was introduced into the model along with sex. The overall frequency of valid diagnoses was 421 of 507. The incidence rates per 100 000 were then adjusted under the assumption that the number of valid diagnoses in the NPR followed the same distribution as in the reviewed subset.

Population characteristics, incidence rates and trend in 2000-2014
From the patient and death registers, we identi ed 8 679 individuals hospitalized with a rst-time diagnosis of myocarditis between 2000 and 2014 in Sweden, with a reference population of 16 622 matched for age and sex without myocarditis. Among patients with myocarditis, 74.8% were men and 63.7% were aged < 50 years. Before the index hospitalization for myocarditis, 7.6% had a prior diagnostic code of acute myocardial infarction, 10.2% HF or DCM, 3.4% pericarditis, 9.2% atrial brillation and 30.0% any prior diagnosis of CVD. The corresponding gures for the controls were 3.0%, 0.5%, 0%, 2.2% and 12.3%, respectively (Table 1). Table 1 Baseline data and comorbidities before the index hospitalization or death. Description of age, gender, incidence period and of relevant comorbidities prior to diagnosis in all individuals with myocarditis in Sweden from 2000 to 2014, as well as in their age and gender matched controls.

Myocarditis
Reference population P-value The overall incidence rate per 100 000 inhabitants increased from 6.3 in 2000 to 8.6 in 2014 (Fig. 1, Supplementary Table S1). However, while the standardized incidence in men has risen from 9.1 in 2000 to 13.5 in 2014 per 100 000, there was little change in women. When further subdivided by age, the most marked rise was in persons aged < 30 years (p < 0.0001) and 30-49 years (p < 0.0001). There was a slight increase in those aged 50-69 years (p = 0.003), whereas a decrease was seen in the age group ≥ 70 years (p = 0.0003) (Fig. 2). Rates in men aged < 50 years nearly doubled from 9.8 in 2000 to 18.6 per 100 000 in 2014 (p for trend < 0.0001), and similarly for women aged < 50 years, although at a much lower level, from 2.0 to 3.6 per 100 000 (p for trend < 0.0001) (Supplementary Table S2). There were downward trends in men and women ≥ 50 years.
In the subpopulation with a hospital diagnosis of myocarditis, validated through a review of medical case records, a similar rise in incidence was seen from 2000 to 2014 (from 5.3 to 7.8 per 100 000 inhabitants).
The increasing incidence for men was from 7.8 in 2000 to 12.4 in 2014 and 2.9 in 2000 to 3.2 in 2014 in women (Fig. 1, Supplementary Table S1).
Diagnoses of heart failure or dilated cardiomyopathy during follow-up From the index hospitalization, 6.4% were newly diagnosed with either HF or DCM during the 1-year follow up while 9.3% developed HF/DCM during the entire study period. (Table 2). Compared with those < 50 years old, a signi cantly higher incidence rate of HF/DCM was observed in those ≥ 50 years with a fourfold higher incidence (3.1% vs. 12.2%) during the rst year after the index hospitalization, and a vefold higher incidence (3.6 % vs. 19.2% ) during the entire study period. The risk for the development of HF/DCM was highest in the immediate post-discharge period after index hospitalization, especially in those ≥ 50 years (Fig. 3). Regardless of age and follow-up duration, the incidences of HF/DCM were much higher in individuals with myocarditis compared with the reference population.  Table S3). Moreover, the risk of death was highest in the immediate post-discharge period after index hospitalization, especially in those ≥ 50 years (Fig. 4). For cause of death, cardiovascular causes contributed to about 50% of the deaths in the myocarditis group. However, this percentage decreased from 70% in 2000 to 50% in 2014, whereas in the reference population < 50% of the deaths were due to cardiovascular causes, decreasing from 44-29% between 2000 and 2014.

Discussion
In this population-based cohort, established on hospitalization for myocarditis according to the Swedish NPR database, the following major ndings were observed: 1) A slightly increased incidence of myocarditis per 100 000 inhabitants (from 6.3 in 2000 to 8.6 in 2014); 2) A decline in 1-year mortality in patients with myocarditis during 2000-2014; 3) An observed reduction in the mortality during the study period in patients ≥ 50 years of age.
The incidence of myocarditis in the general population is at present unknown. Studies have shown suspected myocarditis in 3.5-5% of patients during outbreaks of Coxsackievirus infection 7,12 . The "golden standard" for diagnosis of myocarditis is endomyocardial biopsy. However, this is infrequently used, and has limited sensitivity 4,13 . Autopsy studies have shown estimates of the prevalence of myocarditis to be 2-12% in those who died suddenly [14][15][16] . Other studies reported myocarditis as a cause of initially unexplained DCM in 9% 17 and 10% in unexplained HF 9 .
Our data showed a slight increment of incidence of clinical suspected myocarditis during a period of 15 years. Despite that endomyocardial biopsy is the diagnostic gold standard; it is seldom used in daily clinical practice, partly because of its low sensitivity 1 , but also because myocarditis is often a selflimiting and mostly benign condition for which an invasive procedure is not warranted. Some guidelines recommend endomyocardial biopsy only in a limited number of clinical scenarios, including hemodynamic compromise, patients with life-threatening arrhythmia and those whose condition does not respond to conventional therapy 13 . Therefore, noninvasive examinations and biomarkers are essential for the recognition of myocarditis in clinical practice. During the past few decades, the application of biomarkers has notably improved. Cardiac troponins are more sensitive to myocyte injury than creatinine kinase in patients with clinically suspected myocarditis 18, 19 . In Sweden, troponin T assay was introduced in 1997, which certainly contributed to better detection of myocarditis. However, the increased incidence of myocarditis cannot be sole explained by improvements in biomarker sensitivity, as the increasing trend could be seen already from when coverage for the NPR became nationwide in 1987 (data not shown). The increase might be attributable to other reasons, such as greater awareness among physicians of the myocarditis diagnosis, but also a possible real increase in myocarditis due to a rise in viral infections. Thus, our data may well indicate a true increase in myocarditis over the past decade.
Little information is available about the development of HF/DCM in patients with myocarditis. In our previous studies 20, 21 we hypothesized that an increased incidence of myocarditis might at least partly account for the rise in the incidence of HF in the younger sector of the population. Moreover, some studies reported persistent cardiac dysfunction in about 25% of cases and either death or end-stage DCM in 12-25% [1][2][3][4][5][6]22,23 . In the current study, we found that 6.4% of the patients with myocarditis developed incident HF/DCM within 1 year after the index hospitalization and 9.3% throughout the study period. The incident rates of HF/DCM were higher from 2000 to 2004, with 7.1% developing these conditions within 1 year after index hospitalization, compared with 5.8% in the last period. For patients ≥ 50 years of age, the incidence of HF/DCM was much higher, with 12% during the rst year after the index hospitalization and approximately 19% during long-term follow-up, which is close to previously reported rates 1-6, 22,23 . Of note, we observed a steadily decreased incidence of HF/DCM over time, which might be attributable to improved diagnosis and medical care of myocarditis, as well as of hypertension and ischemic heart disease which also causes HF. The highest risk of both HF/DCM and death occurred in the immediate post-discharge period after index hospitalization, in particular in those ≥ 50 years, suggesting that there is great potential for further improvement of acute care of myocarditis and subsequent follow-up.
Despite higher mortality in patients with myocarditis than in the reference population, mortality continued to decrease in the past decade (2000-2014). The declining trend in 1-year mortality in patients with myocarditis is probably multifactorial, partly attributable to earlier recognition of myocarditis and improved acute management, including HF treatment. Even though we noted a similar trend in mortality in the reference population, the fall was steeper in those diagnosed with myocarditis. The higher 1-year mortality in the myocarditis group, as compared with the reference population, indicates that myocarditis is not an entirely benign condition. Identifying the risk factors for cardiovascular death in patients with myocarditis is imperative and will be the focus of our future research.

Limitations and strengths
The main strengths of this study are access to data from practically all persons in Sweden and that the study covers an extended period. In Sweden, patients with suspected myocarditis based on symptoms and objective signs of cardiac dysfunction are routinely hospitalized. Almost all known myocarditis in Sweden during the study period are thus included in our database.
It is not possible to validate the individual diagnosis in nationwide register studies, which is a limitation. Still, in this study we performed diagnostic validation at the Sahlgrenska University Hospital in Gothenburg -a conglomerate of three hospitals at different locations in the city -every fth year and records for all patients with a diagnosis of myocarditis. In this diagnosis-validated subpopulation, we showed a similar trend with an increasing incidence as in the overall cohort.
In conclusion, by having access to the Swedish national databases (for hospitalization, cause of death and the general population), we could show an increasing trend in the incidence of myocarditis over time but also a declining trend in both development of HF/DCM and mortality in these patients MF designed the study, interpreted the data and drafted the manuscript. SK, ET, CK, EB, MS and AR participated in the design of the study, interpretation of data, and critical revision of the manuscript for important intellectual content. Besides, ET and CK collected the data. TZS analyzed and interpreted the data and critically revised the manuscript. Data Availability Statement: The data underlying this article will be shared on reasonable request to the corresponding author. Figure 1