Knowing the typical length of hospital stay of COVID-19 patientsand which factors affecting the stay time are important for hospital management.A factor can impact the stay time in conflicting ways: a condition may,by the same mechanism, increase the stay time for a severe patient while decrease the time for an even more severe and deceased patient.Material and Method: 3184 COVID-19 patients from the Tokat State Hospital were examined on arrival to determine if they are required further treatment in hospital or not. Among them,404 are inpatients whose hospital stay time is available. By a simple and conditional linear regressions, we examine which factors statistically impact the hospital stay time,both overall and after taking into account the patient's age or eventual survival status.We also use causespecific Cox proportional-hazard regression for competing risk to examine contribution from a factor to the rate of event of mortality, and to the event of discharge.Results: Surviving ICU patients have longer hospital stay time than nonsurviving ICU patients, which in turn longer than non-ICU patients. Older age within acertain range is always correlated with a longer hospital stay. The factors with strongest signal in our dataset are C-reactive protein (CRP), hemoglobin (HGB)and calcium level: increased CRP level, decreased HGB level and calcium level are associated with longer hospital stay, independent from the contribution from surviving status. Other potentially staytime-impacting factors include d-dimer, urea, glucose, white blood cell (WBC) count, neutrophil, but these signals may not be robust against logtransformation of the stay time, or confounding with age. We also observed that glucose is more important than HbA1C or diabetes status in its influence on hospital stay time.
Conclusion:Almost all factors we collected contribute to a faster/slower mortality or discharge rate, in particular C-reactive protein and hemoglobin. Measurement of the associated factors in COVID-19 patients could be used for a better hospital bed management.