This was a retrospective cohort study. We included all consecutive patients aged ≥18 years admitted to two hospitals within one hospital Trust (King’s College Hospital, London, UK and Princess Royal University Hospital, Orpington, UK), which has a dedicated myocarditis service, between 12th February 2009 and 4th October 2021 with a diagnosis of AM.

The study was conducted under London South-East Research Ethics Committee approval (reference 18/LO/2048) granted to the King’s Electronic Records Research Interface (KERRI) (21). We used an open-source retrieval system for unstructured clinical data (CogStack) to identify patients. The CogStack engine, developed at King’s College London (22), was used with Elasticsearch to process structured and unstructured textual clinical data from various hospital databases. Additional data cleansing was performed using Python in JupyterLab and returned as CSV spreadsheets.

Further unique patients were identified from our local Hospital Episode Statistics (HES) data if they had a discharge diagnosis of AM in the first diagnostic position according to appropriate ICD codes (B33.2 viral carditis, I01.2 acute rheumatic myocarditis, I09.0 rheumatic myocarditis, I40 AM, I41 myocarditis in diseases classified elsewhere, I51.4 myocarditis, unspecified). We also searched the hospital and Intensive Care Unit (ICU) discharge summaries for inpatients discharged alive containing the keywords “myocarditis” or “myopericarditis.” Finally, we searched for patients that died during the study period where the keywords “myocarditis” or “myopericarditis” were given as a cause of death on the death notification. Patients and public were not involved in the design of this study.

Diagnostic criteria for AM were defined according to 2013 ESC-position statement on myocarditis and 2018 Lake Louise Criteria for AM (4, 11, 23, 24), which were independently applied to retrieved data by a minimum of two authors. Patients were included if they presented to the hospital with a consistent clinical presentation and one or more diagnostic criteria based on ECG, cardiac enzymes or cardiac imaging structure and function, in the absence of significant coronary artery disease (CAD) on invasive or non-invasive coronary imaging or low likelihood of CAD (patients <40 years with low clinical suspicion). In addition, all patients had CMR confirmation of AM according to consensus recommendations (24). CMR was considered acceptable if it included cine imaging, late-gadolinium enhancement imaging, and parametric mapping of myocardial T1 and T2. Patients with suspected/confirmed COVID-19 or vaccine-related AM were excluded. Patients with cardiac sarcoidosis were also excluded. All patients underwent a history and physical examination. Laboratory parameters, echocardiography and CMR findings were recorded.

We categorised patients into three groups based on their main clinical presentation: (1) chest pain; (2) breathlessness; and (3) arrhythmia. Chest pain presentations were defined as acute chest pain (either ischaemic or pericarditic-sounding) in the absence of CAD and significant left ventricular systolic dysfunction (LVSD). Breathlessness presentations were defined as new onset or progressive HF syndrome. Life-threatening arrhythmias were defined as advanced atrioventricular block, sustained ventricular arrhythmias, or aborted sudden cardiac death. We also categorised patents according to their self-reported ethnicity, and sex. Main clinical presentation was defined by interrogation of the medical record by two authors (DB and AC). Disagreements were resolved by a third author (PS). For each patient, data on baseline co-morbidities and cardiovascular risk factors were obtained using a mix of CogStack retrieval, which was manually validated, and manual searches of the electronic patient record. The primary outcome was a composite of all-cause mortality, resuscitated cardiac arrest, and appropriate implantable cardioverter defibrillator (ICD) therapy following hospital discharge. Appropriate ICD therapy was defined as an appropriate shock for life-threatening arrhythmias.

The results were reported in line with the “Strengthening the Reporting of Observational Studies in Epidemiology” (STROBE) guidelines (Supplementary Table 1) (25). Continuous variables were expressed as mean and standard deviation, or median and interquartile range (IQR), where appropriate. Categorical variables were expressed as counts and percentage. Comparisons between groups were made by the analysis of variance (ANOVA) test on continuous variables or the Student’s t-test, or by the non-parametric Mann–Whitney test when appropriate. The Chi-square test or the Fisher’s exact tests were calculated for discrete variables.

Survival curves for the primary outcome were estimated and compared between groups by means of the log-rank test. Univariable Cox regression models were performed to obtain hazard ratios for adverse events in the study population. Given the low number of events in the population, multivariable analysis was not performed to avoid overfitting. p < 0.05 was considered significant. All analyses were performed with IBM SPSS Statistics Version 28.0 (IBM Corp., Armonk, NY, USA) and R (R Foundation for Statistical Computing, Vienna, Austria).

Screening hospital and ICU discharge summaries for keywords “myocarditis” or “myopericarditis” revealed 683 unique patients. Three additional patients were identified by screening death notifications for the same keywords. A further 152 unique patients were identified by searching for admissions with ICD-10 codes corresponding to myocarditis. Discharge summaries and patient records were screened for eligibility and 207 cases were excluded due to not being a discharge diagnosis or death notification of AM. A further eight patients were excluded as they were younger than 18 years old at time of admission. Next, data on diagnostic criteria were extracted. 196 patients had insufficient evidence of AM, CAD was not sufficiently excluded in 18, and an alternative diagnosis was evident in 161. A further 49 patients had clinically suspected AM that was not proven by CMR or EMB. Finally, 199 patients with CMR-proven AM were included in our cohort. Only one patient underwent EMB.

A total of 199 patients with CMR-proven AM were included in the study (Table 1). The majority (n = 130, 65%) were male and the average age was 39 ± 16 years. The study population was ethnically diverse. While most patients were White (n = 99, 52%), Black, African, Caribbean, or Black British (n = 61, 32%), Asian (n = 8, 4%), and “other” (n = 24, 12%) ethnicities were also represented. The most common clinical presentation was with chest pain (n = 149, 75%), with smaller numbers presenting with breathlessness (n = 32, 16%) and arrhythmias (n = 18, 9%). Fewer than half of patients (n = 80, 41%) experienced prodromal symptoms, which included flu-like symptoms in 26% (n = 51). Only a small number of patients had a pre-existing autoimmune disorder (n = 24, 12%) and the most common cardiac co-morbidity was hypertension (n = 26, 13%). Among those with arrhythmic presentation, 14 patients had sustained ventricular tachycardia or ventricular fibrillation.

At admission, all patients underwent ECG, blood testing and CMR as a minimum. Most patients had an abnormal ECG at presentation (n = 149, 74%), predominantly consisting of repolarisation abnormalities (n = 72, 39%) with a smaller proportion demonstrating ST elevation (n = 48, 26%). Overall, renal function was mildly impaired (eGFR 77 ± 22 ml/min/m²) with elevated inflammatory markers (CRP 65 ± 85 U/L). Peak troponin was elevated at 771 times the upper limit of normal (×ULN). Of patients that had an echo at baseline (n = 173, 87%), two thirds presented with normal left ventricular ejection fraction (LVEF).

Overall, most patients (n = 132, 68%) were managed in a cardiology ward. Only a small minority were treated with advanced therapies, including inotropes (n = 18, 9%), and renal replacement therapy (RRT) (n = 12, 6%). Immunosuppression was started in 16 patients (8%). Specifically, 2 patients (1%) received intravenous immunoglobulin (IVIG) and 14 patients received steroids (7%). At discharge, approximately half of patients were on renin-angiotensin-aldosterone system (RAAS) inhibitors. Cardiac devices were present in 21 patients (8%).

Baseline characteristics according to clinical presentation are summarised in Table 2. Patients presenting with chest pain were younger and more frequently male compared to other presentations, while ethnicity was equally distributed. Co-morbidities and baseline medications were broadly similar between patients with different presentations, although fewer patients with a chest pain presentation had hypertension (p = 0.018) and more patients in the breathlessness group were on immunosuppression (p = 0.002). Admission observations were similar between groups, though patients with a breathlessness presentation had a significantly higher heart rate than chest pain or arrhythmia presentations (p < 0.001).

Patients with a chest pain presentation were more likely to have ST elevation than other presentations (31% compared to 14% with breathlessness and none with arrhythmia, p = 0.004), while patients in the arrhythmia group were more likely to have LBBB (p = 0.006). Patients with breathlessness had worse renal function, lower Hb and higher inflammatory markers than other presentations (p < 0.001 for all). Peak troponin was highest in the chest pain group and lowest in those presenting with breathlessness (p = 0.02). Mean LVEF was higher in the chest pain group compared to breathlessness and arrhythmia patients (53 ± 9 vs. 40 ± 13 vs. 45 ± 12%, respectively, p < 0.001). A total of 78% of patients presenting with breathlessness had LVSD, and more often had moderate or severe LVSD compared to those with chest pain or arrhythmias at presentation (65 vs. 17 vs. 38%, respectively, p < 0.001).

In-hospital management was significantly different between groups. Less than one fifth of patients admitted with breathlessness were managed on a cardiology ward compared to three quarters for other presentations (17% for breathlessness vs. 76% for chest pain vs. 71% for arrhythmia, p < 0.001). With the exception of IVIG, advanced therapies were more commonly used in breathlessness presentations (inotropes p < 0.001, steroids p < 0.001, RRT p = 0.01). Furthermore, patients with a breathlessness presentation were more likely to be treated with RAAS inhibitors (p = 0.042) as well as mineralocorticoid receptor antagonists (MRA, p < 0.001), diuretics (p < 0.001) and/or immunosuppression (p < 0.001). Arrhythmia patients were more likely to be treated with beta blockers (p < 0.001) or statins (p = 0.003), while patients presenting with chest pain were more likely to be treated with colchicine (p < 0.001).

Ethnic differences among patients presenting with AM are presented in Table 3. No differences in age, sex, clinical presentation, comorbidities, or baseline treatment were evident. Compared to White patients, Black and Minority Ethnic (BAME) patients had more repolarisation abnormalities on ECG (47 vs. 32%, p = 0.04) and a more pronounced leucocytosis (neutrophils 10.3 ± 4.2 vs. 9.2 ± 4.1 × 10⁹/L, p = 0.025; lymphocytes 7.8 ± 4.1 vs. 6.7 ± 4.0 × 10⁹/L, p = 0.018). BAME and White patients had similar rates of moderate or severe left ventricular (LV) systolic dysfunction, measured with echo (23 vs. 27%, p = 0.47), and no differences on CMR. Inpatient and discharge medications were also indistinct between groups.

Baseline characteristics according to sex are summarised in Table 4. Compared to men, women were older (44.5 ± 18.1 vs. 36.4 ± 14.6, p = 0.003), while ethnicity was equally distributed. Women were also more likely to have a prior autoimmune disorder (p < 0.001), hypertension (p = 0.002) and renal insufficiency (eGFR < 60 mL/min/m², p = 0.023). More women were on aspirin and/or immunosuppression at the time of presentation (p = 0.035 and <0.001, respectively).

Women were more likely to present with breathlessness compared to men (p = 0.001), and were also more tachycardic at admission (89 ± 24 vs. 80 ± 22 bpm, p = 0.003). Men were more likely to have a chest pain presentation along with ST elevation (p < 0.001 for both).

Women with AM had worse renal function (p < 0.001 for eGFR), lower Hb (p < 0.001) and higher NT-proBNP (p = 0.004) than men. However, men had significantly higher peak troponin (p < 0.001). Inflammatory markers were similar between the groups. Mean LVEF was similar between the groups using both echo and CMR, but both modalities indicated more LV dilatation in men (p < 0.001 for echo).

In-hospital management was significantly different between groups with higher use of steroids, inotropes and RRT in women (p < 0.001 for all). Immunosuppression was also more commonly prescribed for women at discharge (p < 0.001), but other treatments were similar.

Over a median follow-up of 53 (IQR 34–76) months, 11 patients (6%) experienced the primary outcome. Of those, 10 died (5 in the chest pain group, 2 in the breathlessness group and 3 in the arrhythmia group) and 1 patient experienced a successful appropriate ICD shock for monomorphic VT. No patients had resuscitated cardiac arrest.

Overall, 3-year, event-free survival was 96%. Patients in the chest pain group had a more favourable prognosis compared to those presenting with breathlessness or arrhythmia (Figure 1, 3-year event-free survival 98% in the chest pain group, 96% in the breathlessness group, and 89% in the arrhythmia group, p = 0.003). Outcomes were similar between ethnicities (Figure 2, 3-year event-free survival 97% in White patients vs. 96% in BAME patients, p = 0.6) and sexes (Figure 3, 3-year event-free survival 96% in females vs. 97% in males, p = 0.82).

In univariable analyses, the variables most strongly associated with the primary outcome were increasing age (p = 0.003), arrhythmia clinical presentations (p = 0.004), comorbidities including hypertension, dyslipidaemia and diabetes (all p < 0.001), sodium (p = 0.003), and LV septal wall thickness (p = 0.045) (Table 5). Compared to chest pain presentations, arrhythmia presentations were associated with a nearly sevenfold risk of the primary outcome [hazard ratio (HR) 6.97; 95% confidence interval (CI) 1.87–26.00, p = 0.004], whereas breathlessness presentations were not significantly associated with outcome. Sex and ethnicity were not associated with the outcome.

As a retrospective study of prospectively collected data we recognise several important limitations. We attempted to capture all patients admitted with AM by performing keyword searches of electronic health records and death certificates. However, some patients with AM may have been missed, potentially leading to selection bias. We included patients admitted to two hospitals with AM. The patient populations served by these hospitals may not reflect the general population in the UK or other countries. Furthermore, clinical practice may differ from other national and international centres, which may limit the generalisability of our results. Although all cardiac follow-up that occurred in our institution was captured, it was possible that some endpoints were missed. Patients that moved abroad and subsequently died would have been missed, as would patients that had ICDs implanted at another centre. Finally, unstable patients who could not undergo CMR may have been missed. EMB was not routinely performed and post-mortem analysis was not available, so we were unable to assess aetiology. Despite these limitations, our findings are very similar to those of a comparable multicentre study in terms of clinical presentation and outcomes (15). Finally, although ours is a relatively large study by the standards of AM, our cohort was small with a low number of events. We could therefore not perform multivariable analysis and our study power was limited in identifying variables with a less strong association with adverse events. The use of surrogate endpoints in future AM studies (such as change in EF) may provide sufficient power to identify factors that are independently associated with outcome, especially in male and female subgroups. In addition, studies that examine inflammation in patients with AM would be welcomed.