Objective:

Intensive care units (ICUs) collapsed under the global wave of coronavirus disease 2019 (COVID-19). Thus, we designed a clinical decision-making model that can help predict at hospital admission what patients with COVID-19 are at higher risk of requiring critical care.

Methods:

This was a cross-sectional study in 119 patients that met hospitalization criteria for COVID-19 including less than 30 breaths per minute, peripheral oxygen saturation < 93%, and/or ≥ 50% lung involvement on imaging. Depending on the need for critical care, patients were retrospectively assigned to ICU and non-ICU groups. Demographic, clinical, and laboratory parameters were collected at admission and analyzed by classification and regression tree (CRT).

Results:

Forty-five patients were admitted to ICU and 80% of them were men older than 57.13 ± 12.80 years on average. The leading comorbidity in ICU patients was hypertension. The CRT revealed that direct bilirubin (DB) > 0.315 mg/dl together with the neutrophil-to-monocyte ratio (NMR) > 15.90 predicted up to correctly in 92% of the patients the requirement of intensive care management, with sensitivity of 93.2%. Preexisting comorbidities did not influence on the tree growing.

Conclusions:

At hospital admission, DB and NMR can help identify nine in 10 patients with COVID-19 at higher risk of ICU admission.