Clinical risk scores for the early prediction of severe outcomes in patients hospitalized for COVID-19

Abstract

Coronavirus disease of 2019 (COVID-19) is associated with severe acute respiratory failure. Early identification of high-risk COVID-19 patients is crucial. We aimed to derive and validate a simple score for the prediction of severe outcomes. A retrospective cohort study of patients hospitalized for COVID-19 was carried out by the Italian Society of Internal Medicine. Epidemiological, clinical, laboratory, and treatment variables were collected at hospital admission at five hospitals. Three algorithm selection models were used to construct a predictive risk score: backward Selection, Least Absolute Shrinkage and Selection Operator (LASSO), and Random Forest. Severe outcome was defined as the composite of need for non-invasive ventilation, need for orotracheal intubation, or death. A total of 610 patients were included in the analysis, 313 had a severe outcome. The subset for the derivation analysis included 335 patients, the subset for the validation analysis 275 patients. The LASSO selection identified 6 variables (age, history of coronary heart disease, CRP, AST, D-dimer, and neutrophil/lymphocyte ratio) and resulted in the best performing score with an area under the curve of 0.79 in the derivation cohort and 0.80 in the validation cohort. Using a cut-off of 7 out of 13 points, sensitivity was 0.93, specificity 0.34, positive predictive value 0.59, and negative predictive value 0.82. The proposed score can identify patients at low risk for severe outcome who can be safely managed in a low-intensity setting after hospital admission for COVID-19.

Clinical risk scores for the early prediction of severe outocomes in patients hospitalized for COVID-19: comment

“…Moreover, they did not compare survivors with non-survivors, but patients with severe with those with non-severe COVID-19. Notwithstanding these differences their baseline results reported in their Table 1 [ 1 ] are broadly similar to those obtained in the present registry. Our patients who died had higher WBC values, with a very high neutrophil to lymphocyte ratio owing to a decrease in lymphocytes.…”

“…These laboratory results obtained at hospital admission in patients with COVID-19 can be discussed with the results obtained by Ageno et al [ 1 ] in the context of both their derivation and validation cohorts, even though they did not use laboratory markers to specifically evaluate the condition of body organs and systems, but to build up the prediction score. Moreover, they did not compare survivors with non-survivors, but patients with severe with those with non-severe COVID-19.…”

“…On behalf of the Italian Society of Internal Medicine Ageno et al [ 1 ] chose to use the database of a registry including 610 patients admitted with COVID-19 to five hospitals in Northern Italy in order to develop and validate a score that at the time of hospital admission would be able to predict a severe disease outcome, defined as the need of non-invasive ventilation, orotracheal intubation or death [ 1 ]. In the frame of their analysis meant to build the score and its accompanying validation the authors ultimately produced a score that, made of six clinical and laboratory variables, was able to identify patients at relatively low risk of a severe outcome who could be handled in the context of the low intensity of care offered by internal medicine wards [ 1 ].…”

“…On behalf of the Italian Society of Internal Medicine Ageno et al [ 1 ] chose to use the database of a registry including 610 patients admitted with COVID-19 to five hospitals in Northern Italy in order to develop and validate a score that at the time of hospital admission would be able to predict a severe disease outcome, defined as the need of non-invasive ventilation, orotracheal intubation or death [ 1 ]. In the frame of their analysis meant to build the score and its accompanying validation the authors ultimately produced a score that, made of six clinical and laboratory variables, was able to identify patients at relatively low risk of a severe outcome who could be handled in the context of the low intensity of care offered by internal medicine wards [ 1 ]. During the selection and development of the best predictive score they chose to evaluate for inclusion data on demographic variables (age, sex, and time elapsed from disease onset), a few comorbidities (hypertension, diabetes, coronary artery disease, chronic obstructive pulmonary disease) but also an array of laboratory values obtained at admission, such as the white blood cells counts (WBC) (split by neutrophil and lymphocyte absolute counts), C-reactive protein (CRP), alanine amino transferase (ALT), aspartate amino transferase (AST), albumin, creatinine, D-dimer, and the neutrophil/lymphocyte ratio.…”

“…During the selection and development of the best predictive score they chose to evaluate for inclusion data on demographic variables (age, sex, and time elapsed from disease onset), a few comorbidities (hypertension, diabetes, coronary artery disease, chronic obstructive pulmonary disease) but also an array of laboratory values obtained at admission, such as the white blood cells counts (WBC) (split by neutrophil and lymphocyte absolute counts), C-reactive protein (CRP), alanine amino transferase (ALT), aspartate amino transferase (AST), albumin, creatinine, D-dimer, and the neutrophil/lymphocyte ratio. Of these laboratory values CRP, AST, D-dimer, and the neutrophil/lymphocyte ratio were ultimately included in the actual score, with added clinical and demographic variables such as patient age and history of coronary artery disease [ 1 ].…”