Characteristics Associated With Multisystem Inflammatory Syndrome Among Adults With SARS-CoV-2 Infection

This cohort study describes the clinical characteristics and outcomes of patients with multisystem inflammatory syndrome among adults with laboratory-confirmed SARS-CoV-2 infection at a single US tertiary care medical center.

The overview of patient selection is depicted in eFigure 1. Adults ≥ 21 years-old at risk of multisystem inflammatory syndrome associated with SARS-CoV-2 infection (MIS-A) were identified as those hospitalized surrounding SARS-CoV-2 testing based on diagnostic method as follows: (eFigure 2) 1) RT-PCR group: a. Positive SARS-CoV-2 RT-PCR nasopharyngeal swab followed. b. Hospital admission ≥ 14 days and ≤ 84 days (12 weeks) from initial positive PCR test. 2) Serology group: a. Positive SARS-CoV-2 IgG. b. Hospital admission within 15 days before or after positive IgG test.
In the RT-PCR group, a 14-day "blanking period" was implemented to avoid contamination from admissions driven by acute COVID-19 rather than MIS-A. We selected 14 days, because it has been previously suggested that a substantial proportion of MIS-C patients were infected with SARS-CoV-2 at least 1-2 weeks before MIS-C onset. 3 Patients admitted >84 days following positive SARS-CoV-2 RT-PCR were excluded. All others admitted were included in the Acute COVID-19 admission group used for comparison. Although the CDC working definition of MIS-A, defines evidence of prior SARS-CoV-2 infection as any positive test within 12 weeks of symptoms, and does not stipulate a 14-day blanking period from SARS-CoV-2 RT-PCR, we modified this criterium due to the limitations of our retrospective study and potential overlap with acute COVID-19.
For serology, a 15-day window was selected as what the authors thought represented reasonable period for a serology result to be obtained in the onset of MIS-A or to return following potential discharge following MIS-A admission and not be attributable to another systemic illness.
Then, charts of individuals at risk of MIS-A were manually reviewed by one of three physicians in the team (D.E.C., G.E.D., E.H.) to identify charts to discuss in an adjudication group. Patients were excluded during screening if recorded admissions were: 1) Procedural: Delivery, c-section, and other elective procedures.
2) Well-supported alternative diagnosis: For example, sepsis with positive bacterial cultures. 3) Not meeting at least 4 of the 5 MIS-A criteria from the CDC definition. 4 4) Lack of recovery and progression from COVID-19 initial presentation.
All remaining cases at risk for MIS-A were then presented in a three-physician decision group (comprised by D.E.C., G.E.D., E.H.) and adjudicated as case or excluded. Disagreements were resolved by consensus.

Case classification and definition of organ system involvement:
Cases reviewed in full were adjudicated as MIS-A if the review team agreed that the cases met all 5 MIS-A criteria from CDC's case definition and no alternative explanation of the presentation was plausible.
Organ-system involvement was adjudicated on the basis of symptoms, clinical findings, and laboratory measurements, similarly to previously described 3 and are presented in eTable 2.

Data Collection of cases:
Demographics, BMI, and chronic comorbidities for all participants were extracted from the EHR. Definitions used to identify comorbidities are provided in eTable 1.
Additional data collected of MIS-A cases included clinical presentation, vital signs on admission, laboratory parameters and microbiologic data obtained during admission, organ system involvement as described above, findings on cardiovascular diagnostic studies, therapies, and survival. Data were obtained from medical records and collected in a Vanderbilt's Research Electronic Data Capture (REDCap) by a member of the research team (D.E.C., G.E.D, E.H., A.Y.).

Statistics:
Data are presented as median and interquartile range for continuous variables, and counts and percentages for categorical variables unless otherwise specified. Missing data were not imputed. Continuous variables were compared using Mann-Whitney U test, and categorical variables compared with Fisher's exact test. Tests were twotailed. A p-value <0.05 was considered significant. Data analyses were performed and graphs created using R software, version 4.0.2 (R foundation for Statistical Computing, Vienna, Austria).