JMIR Publications

ABSTRACT

Background:

As the coronavirus disease 2019 (COVID-19) pandemic continues, initial risk-adapted allocation is crucial for managing medical resources and providing intensive care.

Objective:

This study aimed to identify factors that predict COVID-19 mortality and develop a COVID-19 prognosis score (COPS) based on these factors.

Methods:

This study retrospectively analyzed nationwide cohort of laboratory-confirmed COVID-19 cases between January and April 2020 in Korea. This study identified factors associated with mortality, subsequently constructing a multivariable model to develop the scoring system. Each variable’s score in the COPS system was a log-scaled converted value of the adjusted odds ratio.

Results:

Among the 5,594 patients included in this analysis, 234 died after COVID-19 diagnosis. Survivors stayed in a hospital significantly longer than non-survivors (26.1±10.7 vs. 15.6±13.3 days); moreover, length of hospital stay and disease severity were directly associated in survivors (P<.001). Twelve parameters were significantly related to mortality: advanced age; male sex; heart rate ≥110/min; dyspnea, and mental disturbance; diabetes; chronic renal failure; cancer in active treatment; dementia; and cytopenia (hemoglobin <12.5 g/dL, absolute lymphocyte count <1,000/mm3, and platelet count <100,000/mm3). Accordingly, the COPS system was developed and risk groups were created: very low-risk (score ≤3.72), low-risk (3.73–4.96), intermediate-risk (4.97–6.20), high-risk (6.21–8.68), and very high-risk (>8.68), presenting mortality probabilities of 0.1%, 4.5%, 17.8%, 41.3%, and 83.3%, respectively (P<.001). Receiver operating characteristics curve analysis achieved an area under the curve of 0.959.

Conclusions:

The newly developed predictive COPS system may assist in risk-adapted decisions for medical resources allocation, including intensive care, during the COVID-19 pandemic.