Clinical Risk Prediction Scores in Coronavirus Disease 2019: Beware of Low Validity and Clinical Utility

Area of Science:

• Medical research
• Epidemiology
• Clinical informatics

Background:

• Several risk stratification tools have been developed for coronavirus disease 2019 (COVID-19) to predict disease progression.
• However, external validation of these tools is lacking, limiting their clinical applicability.

Purpose of the Study:

• To externally validate three previously published risk stratification tools for predicting severe COVID-19 outcomes.
• To assess the predictive performance of the COVID-GRAM score, CALL score, and a nomogram in a multicenter cohort.

Main Methods:

• A multicenter study involving 169 patients diagnosed with COVID-19.
• Data collected included patient demographics, comorbidities, and outcomes.
• Performance of risk stratification tools was evaluated using the area under the receiver operating characteristic curve (AUC).

Main Results:

• The COVID-GRAM score achieved an AUC of 0.636 (95% CI: 0.550-0.722).
• The CALL score demonstrated an AUC of 0.500 (95% CI: 0.411-0.589).
• The nomogram yielded an AUC of 0.628 (95% CI: 0.543-0.714).

Conclusions:

• The validated risk stratification tools exhibited suboptimal predictive performance for severe COVID-19 progression in this external cohort.
• The findings suggest a need for the development and validation of more accurate tools for COVID-19 risk stratification.