P2.C046: Using the N3C Data Enclave and machine learning tools to identify risk factors for major adverse cardiac events in children and young adults following SARS-COV2 infection

Background: SARS-CoV2 infection has been linked to a wide range of multiorgan system conditions. The cardiovascular effects of SARS-CoV2 infection most notably include myocarditis and a delayed multisystem inflammatory syndrome in children (MIS-C), a potentially serious condition affecting multiple organs including cardiac and vascular tissues. Myocarditis secondary to SARS-CoV2 infection may often go unrecognized and can present with late and nonspecific symptoms. Predicting individuals at risk of cardiovascular involvement enables prompt treatment, potentially averting life-threatening consequences.

Methods: The National Cohort COVID Collaborative (N3C) Data Enclave was used to identify patients aged 0-30 years, infected with SARS-CoV2 between 1/1/2020-3/31/2022 and whose health system provided data at least six months beyond diagnosis. Outcomes included cardiac events within six months of the index COVID-19 visit, such as arrhythmia, cardiac arrest, congestive heart failure, cardiomyopathy, heart transplant, Kawasaki disease, myocarditis, pericardial effusion, pericarditis, or undergoing cardiac testing(more than 1 EKG, echocardiogram and more than 3 troponin tests). Predictors included COVID-19 severity, known cardiac and non-cardiac comorbidities and Area Deprivation Index (ADI). Data were stratified by age group (0-4, 5-17, 18-30). In a second step patients with known cardiac comorbidities were excluded and the data re-analyzed to identify de-novo cardiac events. Patients with and without cardiac events were matched on data partner and on length of observation period before COVID infection. Random forest was used to build the models and optimized with best parameters for performance, and SHAP's beeswarm plot was applied to visually interpret features of importance.

Results: COVID-19 severity and prior known arrythmia were identified as major risk factors for cardiac events across all age groups when the data included children with known cardiac comorbidities (Fig 1). Congenital heart disease was a major risk factor in the 0-4 age group but ranked lower at older ages. As ADI increases (and therefore socioeconomic status decreases), patients -were more likely to develop a cardiac event, with this association being more pronounced in individuals under 5 years of age. Vaccinations in individuals aged 5 years and older were associated with a decreased risk of a cardiovascular event (Fig 1). When excluding children with cardiac comorbidities, COVID-19 severity remained the primary associated risk factor, followed by ADI, race and gender (Fig 2).

These analyses were conducted using the NCATS N3C Data Enclave supported by NCATS U24 TR002306.

Conclusion: COVID-19 severity emerged as the most critical risk factor for the cardiac event within six months of infection, both in the overall population and in patients without known cardiac comorbidities. Congenital heart disease posed a risk in the 0-4 age group, while arrythmia was an important risk factor in children with cardiac comorbidity across all age groups. ADI, race, and gender were identified as other important though more minor risk factors.