OBJECTIVES: This study aims to evaluate COVID-19 in-hospital costs and identify predictors at a patient-level in Brazil.

METHODS: This is multicenter, prospective cohort study that applied time-driven activity-based costing (TDABC) method in five Brazilian reference centers for COVID-19 treatment. Patients hospitalized between March and August 2020 (first wave of the disease) and had their COVID-19 status confirmed by reverse transcription-polymerase chain reaction (RT–PCR) at arrival were included in our sample. The cost information was calculated at the patient level and multivariable analyses were applied to identify clinical predictors of cost variability, considering ICU admissions and patient’s comorbidities. RESULTS: 830 patients were included into the analysis. The median cost per patient was I$4,428 (IQR 2,019; 11,464), and patients hospitalized in ICU demonstrated significative higher costs (p<0.001). Patients hospitalized in ICU the median was I$11,596 (IQR 6,016; 23,374), while for those who were hospitalized in ward was 1,895 (IQR 1,050; 3,317). Median cost per day was I$ 455 (IQR 308; 711) for the total sample, I$690 (IQR I,528; 1,046) for ICU patients and I$350 (IQR 255; 449) for non-ICU. Gender (p<0.001), Obesity (p = 0.005) and Chronic pulmonary diseases (p = 0.044) were identified as clinical predictors for hospital costs.

CONCLUSIONS: By developing a multicenter microcosting study for COVID-19 this study allowed to measure the variability in resource consumption per patients’ according their clinical characteristics. These findings can sustain the development of financially sustainable health policies in middle-income countries such as Brazil.