An integrated approach-demographic, clinical, laboratory and radiological features in predicting mortality of sars-cov-2 patients in critical care medicine

Abstract

Background: With the rapidly evolving new variants of SARS-Cov-2, the scientific community is still learning to identify patients with higher risks for effective triaging and better resource allocation as there is no effective specific therapeutics for COVID-19 patients.Aim: To analyse the demographic, laboratory, clinical and radiological features in COVID -19 patients admitted in critical care medicine and to study their association with survivors and non survivors and to propose a model to predict mortality rate in critically ill COVID -19 patients. Methods: The data of RT-PCR confirmed COVID-19 patients (age, gender, RR, PR, BP, SpO2, DM, HTN, WBC, Hb, Platelet, CRP, LDH, D-dimer, Creatinine, Urea, CT Score, lung involvement pattern and distribution) was retrospectively evaluated and compared between survivors and non-survivors. Results: Among the 91 enrolled patients, 65(71.42%) survived and 26 (28.58%) succumbed to death. In the non-survivors mean age was 61.42±13.24, male 18(69.23%), female 8(30.76%). Backward stepwise logistic regression is used to identify the significant predictors of mortality. These parameters were significant in our Backward logistic regression model: RR(p:0.008, OR1.164), spO2(p:0.05, OR:0.928), WBC(p:0.001, OR:1.170), D-dimer (p: 0.005, OR:0.999), Urea (p:0.001, OR:0.916) and CT(p:0.000, OR:1.259). The sensitivity of the model is 80.00% (95% confidence interval is [59.30% 93.17%]), specificity is 92.68%. (95% CI is [80.08% 98.46%]). The overall accuracy is 87.88%. (95% CI is [77.51% 94.62%]). The positive predictive value is 86.96%. (95% CI is [68.79% 95.28%]). The negative predictive value is 88.37%. (95% CI is [77.55% 94.36%]). Conclusion: Involving clinical, laboratory and radiological features has shown to be a good approach in mortality prediction of critically ill COVID-19 patients.