Development and external validation of a prognostic tool for COVID-19 critical disease

Area of Science:

• Critical care medicine
• Epidemiology
• Biostatistics

Background:

• The COVID-19 pandemic highlighted limitations in critical care capacity.
• Predicting disease severity is crucial for managing resources and improving patient outcomes.
• Early identification of patients at high risk for critical illness is essential.

Purpose of the Study:

• To develop and externally validate a prognostic model for predicting critical COVID-19 disease.
• To create a clinical tool to assess the likelihood of severe COVID-19 upon presentation.
• To aid in critical care capacity management and targeted interventions.

Main Methods:

• Retrospective study using logistic regression to develop a prognostic model.
• Derivation cohort from University of California Irvine Medical Center; external validation at Emory Healthcare.
• Model performance assessed using concordance statistics.

Main Results:

• The model identified key predictors of critical COVID-19: comorbidities, BMI, respiratory rate, WBC, lymphocytes, creatinine, LDH, troponin I, ferritin, procalcitonin, and CRP.
• High model discrimination observed in the external validation cohort (concordance statistic: 0.94).
• A web-based tool was developed for clinical use.

Conclusions:

• A validated prognostic model accurately predicts critical COVID-19 risk.
• The model utilizes readily available clinical data (comorbidities, vitals, labs).
• This tool can assist in prognostication and critical care resource allocation.