Characterization of Multisystem Inflammatory Syndrome in Children (MIS-C) and Its Relationship to Kawasaki Disease

Summary

Participation Requirements

Description

Clinical characterization (Burns, Tremoulet, Sivilay, Roberts, Jain): Over the 8 months of enrollment funded by this supplement, 30 sites will collect data on KD and MIS-C patients using the detailed case report form in our REDCap database. This includes data on patient demographics, clinical presen...

Clinical characterization (Burns, Tremoulet, Sivilay, Roberts, Jain): Over the 8 months of enrollment funded by this supplement, 30 sites will collect data on KD and MIS-C patients using the detailed case report form in our REDCap database. This includes data on patient demographics, clinical presentation, laboratory data, treatments, clinical outcomes, and cardiovascular outcomes on all KD and MIS-C patients so that the investigators can develop a systematic picture of the different patient groups (see details below). Parent observations (Kim): To learn about signs and symptoms in the patient and family members leading up to acute presentation a parent questionnaire will be devised and analyzed by Dr. Katherine Kim with the Patient and Parent Advisory Board. The investigators will collect known signs and symptoms as well as those that may not have been previously reported in the literature and record presence/absence, location in the body, and severity level by day. The questionnaire will be available as a mobile/web application and via paper. With these data, the investigators will conduct an exploratory analysis to characterize symptom phenotypes and the relationships of these profiles with demographic features and clinical characteristics. In addition, the investigators will assess whether we can detect differences between the symptom phenotypes of KD and MIS-C. The investigators will conduct cluster analysis to identify symptom phenotypes from aggregate symptom observations blinded as to presumed or verified diagnosis (n=100 with each sign/symptom on each day as a distinct data point). Phenotypes may include characteristics such as symptoms that co-occur or are independent, and symptom burden index (e.g., number symptoms present). The investigators will use Ward's hierarchical cluster analysis to estimate the number of likely clusters.3 The investigators will apply K-means nonhierarchical cluster analysis repeated 100 times in a leave-one out validation model to assure repeatability and stability within the model. The investigators will create score indices analyzed with logistic regression, principal component analysis, factor analysis and correlation analysis. The investigators will then assess whether clusters are related to verified diagnosis and/or sociodemographic characteristics such as age, race/ethnicity, geography. The investigators will use discriminant analysis techniques and more recent classification techniques such as CART to examine if symptom phenotypes of KD and MIS-C are dissimilar. If information in terms of symptom clusters at a point in time or a trajectory of a single symptom over time are insufficient to distinguish between the two conditions, this would provide support for the argument that the recorded symptoms are insufficient for discrimination or that the two disease entities are not symptomatically different. This exploratory analysis will provide important information that can be further developed for clinical guidance and parent education. Photography (Kim, Tremoulet): Patient photographs of the eye, mouth/tongue, and rash will be collected as a novel addition to the usual clinical data. These photographs obtained before treatment will document the presence or absence of conjunctival injection and perilimbal sparing, mucocutaneous changes in the oropharynx including changes in the vermillion border (erythema, fissuring) and the tongue (strawberry tongue), and the nature of the rash. The photographs will be subjected to analysis using facial recognition software and artificial intelligence approaches used by our collaborators at the University of Southern California Center for Artificial Intelligence in Society led by Hayden Shively and Lucas Hu to evaluate whether a computer algorithm can be created that can differentiate the clinical characteristics of MIS-C from photographs taken of children with acute KD and those with other pediatric febrile illnesses. Dr. Katherine Kim at UC Davis will compare the photos with the questionnaire responses to supplement the parent descriptions of signs and symptoms and validate the observations. This comparison can result in enhanced descriptions using the words of parents themselves. Photography is needed to document these physical findings as the investigators have learned over the years that physician description of these features is woefully inaccurate. The finding, for example, of a strawberry tongue is a specific injury pattern that involves sloughing of the cornified tips of the filiform papillae and has historically been associated with only 3 conditions: staphylococcal and streptococcal toxin-mediated disease and KD.

Tracking Information