Mortality And Intensive-Care Unit Admission Prediction Models And Risk Score For Hospitalized Covid-19 Patients

Abstract

BACKGROUND: The global spread of covid-19 disease has placed enormous burden on health care systems worldwide. Many patients quickly deteriorate, highlighting the importance of early risk stratification. Risk stratification is critical during the acute phase of COVID-19 as it may assist guide decisions making such as the severity of the initial treatment and the duration of COVID-19 treatment.

Objective: The objective of this study is to develop prediction models and risk scores that estimates intensive-care-unit admission and mortality of hospitalised COVID-19 patients from hospital records.

STUDY DESIGN: Retrospective review of medical records of demographics, clinical characteristics, laboratory findings and comorbidities of the admitted COVID-19 patients.

RESULTS: A total number of 693 hospitalised COVID-19 patients were included in the final analysis (mean age, 58.17 ± 15.77). Of which 320 were cases of general admission, 373 were admitted in ICU while 160 were those who died during hospital stay. Five significant predictor variables identified by logistic regression model for ICU group were age (OR: 1.01, 95% CI: 1.001 – 1.02), male gender (OR: 1.66, 95% CI: 1.16 – 2.37), diabetes (OR: 1.41, 95% CI: 1.001 – 2.01), pneumonia (OR: 6.62, 95% CI: 4.50 – 9.75) and septic shock (OR: 15.0, 95% CI: 1.95– 115.6). Six significant variables predicting mortality were age (OR: 1.05, 95% CI: 1.03 – 1.06), LDH (OR: 1.003, 95% CI: 1.002 – 1.004), pneumonia (OR: 1.62, 95% CI: 1.01 – 2.77), respiratory failure (OR: 3.82, 95% CI: 2.42 – 6.03), acute kidney infection (OR: 1.80, 95% CI: 1.01– 3.51) and septic shock (OR: 11.0, 95% CI: 3.02– 40.46,). The risk scores were developed from the final logistic regression model identified by the backward Wald-method. Each variable was assigned risk score based on significant odds ratio. Our model prediction performance yielded an AUC of 0.76 [95% CI, 0.72–0.79] for ICU admission and  0.86 [95% CI, 0.82–0.89] for mortality.

CONCLUSION: In conclusion, this simple tool developed from regression model may be useful for future decision support in prioritising patients for hospitalisation, particularly in situations where medical resources are limited. This risk score will assist clinicians in identifying severe COVID-19 patients at an early stage and will be helpful in determining the best treatment plan.