Background Multisystem inflammatory syndrome in adults (MIS-A) was reported in association with the COVID-19 pandemic. MIS-A was included in the list of adverse events to be monitored as part of the emergency use authorizations issued for COVID-19 vaccines. Methods Reports of MIS-A patients received by the Centers for Disease Control and Prevention (CDC) after COVID-19 vaccines became available were assessed. Data collected on the patients included clinical and demographic characteristics and their vaccine status. The Vaccine Adverse Events Reporting System (VAERS) was also reviewed for possible cases of MIS-A. Results From December 14, 2020 to April 30, 2021, 20 patients who met the case definition for MIS-A were reported to CDC. Their median age was 35 years (range, 21-66 years), and 13 (65%) were male. Overall, 16 (80%) patients had a preceding COVID-19-like illness a median of 26 days (range 11-78 days) before MIS-A onset. All 20 patients had laboratory evidence of SARS-CoV-2 infection. Seven MIS-A patients (35%) received COVID-19 vaccine a median of 10 days (range, 6-45 days) before MIS-A onset; 3 patients received a second dose of COVID-19 vaccine 4, 17, and 22 days before MIS-A onset. Patients with MIS-A predominantly had gastrointestinal and cardiac manifestations and hypotension or shock. Conclusions Although 7 patients were reported to have received COVID-19 vaccine, all had evidence of prior SARS-CoV-2 infection. Given the widespread use of COVID-19 vaccines, the lack of reporting of MIS-A associated with vaccination alone, without evidence of underlying SARS-CoV-2 infection, is reassuring.

Background Classification of MIS-C, COVID-19, and other pediatric inflammatory conditions is challenged by phenotypic overlap and absence of diagnostic laboratory evidence. Due to public health need and based on limited data from early cases, CDC developed a necessarily broad MIS-C surveillance case definition in May 2020. Studies have since shown that some criteria do not distinguish between MIS-C and other conditions and may contribute to misclassification. To inform planned revision to the CDC definition, we evaluated the impact of narrowing these criteria on case inclusion in national MIS-C surveillance. Methods Of state and local health-department reported cases meeting the current MIS-C case definition as of 04/14/2022, we describe the proportion that met revised criteria under consideration including fever duration, C-reactive protein (CRP) elevation using a defined cutoff, and organ involvement represented by specific criteria. We also evaluated cases identified using potential combinations of revised criteria. Results Of 8,096 MIS-C cases fulfilling the original case definition, 6,332 (78%) had sufficient data for evaluation of criteria. Of these, 96% had fever for ≥2 days and 94% had a CRP ≥ 3.0 mg/dL (Table 1). Cardiac involvement defined by key features of MIS-C was present in 84% of cases (62% if BNP/proBNP elevation was excluded); 43% had shock. Dermatologic, gastrointestinal (GI) and hematologic involvement were present in 75%, 89% and 37% of cases, respectively. Neurologic (excluding headache), renal, and respiratory involvement were present in 16%, 20%, and 63% of cases, respectively. The number of cases with ≥ 2 of cardiac (without BNP/proBNP elevation), shock, dermatologic, GI, or hematologic involvement was 5,733 (91%). SARS-CoV-2 testing results are shown in Table 2. Conclusion The CDC MIS-C case definition is intentionally broad. Using national surveillance data, we evaluated case inclusion under narrower criteria, prioritizing features of MIS-C that distinguish it from similar pediatric inflammatory conditions. A surveillance case definition may not capture all cases and is not intended to replace clinical judgment. We plan to assess additional criteria combinations, describe potentially excluded cases, and incorporate findings into a revised definition. Disclosures All Authors: No reported disclosures.

Objective: When novel diseases such as COVID-19 emerge, predictors of clinical outcomes might be unknown. Using data from electronic medical records (EMR) allows evaluation of potential predictors without selecting specific features a priori for a model. We evaluated different machine learning models for predicting outcomes among COVID-19 inpatients using raw EMR data. Materials and Methods: In Premier Healthcare Data Special Release: COVID-19 Edition (PHD-SR COVID-19, release date March, 24 2021), we included patients admitted with COVID-19 during February 2020 through April 2021 and built time-ordered medical histories. Setting the prediction horizon at 24 hours into the first COVID-19 inpatient visit, we aimed to predict intensive care unit (ICU) admission , hyperinflammatory syndrome (HS), and death. We evaluated the following models: L2-penalized logistic regression, random forest, gradient boosting classifier, deep averaging network, and recurrent neural network with a long short-term memory cell. Results: There were 57,355 COVID-19 patients identified in PHD-SR COVID-19. ICU admission was the easiest outcome to predict (best AUC=79%), and HS was the hardest to predict (best AUC=70%). Models performed similarly within each outcome. Discussion: Although the models learned to attend to meaningful clinical information, they performed similarly, suggesting performance limitations are inherent to the data. Conclusion: Predictive models using raw EMR data are promising because they can use many observations and encompass a large feature space; however, traditional and deep learning models may perform similarly when few features are available at the individual patient level.