Predictors of COVID-19 Hospital Treatment Outcome

BACKGROUND: There are more than 228,394,572 confirmed cases and 4,690,186 confirmed deaths caused by COVID-19 worldwide. The magnitude of the COOVID-19 pandemic has stimulated research on the treatment and diagnosis of COVID-19 patients. OBJECTIVE: In this report, a battery of specific parameters was used to develop a model that allows prediction of the outcome of the COVID-19 treatment. These parameters are C-reactive protein, procalcitonin, fibrinogen, D-dimers, immature granulocytes, and interleukin-6. METHODS: The study was carried out on a sample of N = 49 survivors (22 men, 27 women) and 83 deceased patients (62 men, 21 women). The distribution of means and differences in means of the parameters studied between survivors and deceased patients were evaluated using the bootstrap method. RESULTS: A mathematical model that allows for the prediction of hospitalization outcome was obtained using the Naive Bayes model. The results demonstrated a statistically significant difference between survivors and deceased patients in all parameters studied. A mathematical model employing a battery of parameters provided a 97% precision in predicting the outcome of hospitalization. CONCLUSION: This study showed that the cross-correlation of survivability with absolute levels of C-reactive protein, procalcitonin, fibrinogen, D-dimers, immature granulocytes, and interleukin-6 could be used successfully in the hospital setting as a diagnostic tool.