Study population and setting
This study included all patients with non-severe, laboratory-confirmed COVID-19 (n=208) admitted to two hospitals in China between January 20, 2020 and February 22, 2020. One hundred sixty-eight patients were classified as having clinically stable disease, and 40 experienced progression to severe illness. Cox proportional hazards regression was used to develop a nomogram for predicting disease progression. A nomogram is a diagram that represents the relationship between three or more variables of interest such that the value of one variable (e.g. disease progression) can be found by typically drawing a straight line which intersects the other variable (predictors) scales at appropriate values.
Summary of Main Findings
At hospital presentation, older age, low lymphocyte count, high lactate dehydrogenase levels, and the presence of medical comorbidities were independently associated with the likelihood of progression to severe disease. A risk score (nomogram) constructed using these factors had good negative predictive value, but only moderate positive predictive value for predicting clinical progression to severe illness.
Risk factors were identified through multivariable analysis. The predictive score has few components and is easily applied.
Model validity was not assessed on an independent population. Because the study population only includes those without severe disease presentation, the resulting risk factors are only appropriate for risk-stratifying those with mild/moderate disease. The generalizability of the factors and the predictive accuracy of the resulting score is unclear. Predictive tools using results from laboratory tests are less useful in health care settings requiring rapid sorting of patients into risk groups.
This study is one of the first to offer a predictive tool for COVID-19 disease progression.