SHM Converge Scientific Abstract ePoster Gallery-Development and Validation of Risk Prediction Model for Progression to Organ Failure or Death in Hospitalized Patients With Covid-19

Pdf Summary

This study developed and internally validated a risk prediction model for hospitalized adults with laboratory-confirmed COVID-19 to estimate the chance of progressing to respiratory failure, circulatory failure, or death within 28 days of admission.<br /><br />Researchers analyzed 3,844 consecutive patients across 16 Mayo Clinic hospitals in Arizona, Florida, and Minnesota from March 2020 to February 2021. The cohort was randomly split into a derivation set (70%) and a validation set (30%). The model used routinely available admission data, including demographics, social factors, comorbidities, vital signs, laboratory values, and medications.<br /><br />Of the total patients, 389 (10%) experienced the composite outcome. Fifteen admission characteristics remained in the final logistic regression model, including age, sex, non-White race, current smoking, heart disease, COPD, dementia, immunocompromised state, transplant status, liver disease, low systolic blood pressure, high respiratory rate, and elevated temperature.<br /><br />The model showed good performance, with an AUROC of 0.79 in the development cohort and 0.76 in the validation cohort, indicating strong discrimination. Calibration testing also suggested the model’s predicted risks matched observed outcomes well.<br /><br />The authors conclude that this model may help clinicians stratify risk early during hospitalization, guiding decisions such as ward versus ICU placement and prioritization of aggressive treatment. It may also support resource allocation in strained hospital settings.<br /><br />Strengths of the study include its large, multicenter real-world cohort, inclusion of both academic and rural hospitals, detailed patient data, and adherence to TRIPOD reporting standards. Limitations include the integrated nature of the health system, limited racial/ethnic diversity, and the fact that the model was not externally validated. The cohort also predates later variants such as Delta and Omicron, so applicability may differ in later phases of the pandemic.

Asset Subtitle

Mohammed Yousufuddin

Meta Tag

Author List Eimad Ahmmad, Francis Owusu, Kanika Khandelwal, Manal Bakhiet, Mohammad Murad, Mohammed Yousufuddin, Sadia Munawar, Sarah Gerard, Tahir Mehmood, Zhen Wang

Category Research

Concept Predictive Model

Concept Organ Failure

Concept Death

Concept Hospitalized COVID-19 Patients

Concept Discrimination

Distinguished Non-Finalist

Presenter Organization Mayo Clinic

Presenting Author Mohammed Yousufuddin

Track Outcomes Research

Keywords

COVID-19

risk prediction model

hospitalized adults

respiratory failure

circulatory failure

mortality

logistic regression

AUROC

clinical risk stratification

Mayo Clinic hospitals

Predictive Model

Organ Failure

Death

Hospitalized COVID-19 Patients

Discrimination