Table 1.

Overview of selected models for predicting short term mortality in patients admitted to hospital with SARS-CoV-2 infection

Model	Country of development	Development population	Predicted outcome	Predictors	Model type	Estimation method
Bello-Chavolla et al14	Mexico	All reported confirmed cases of covid-19, including hospital admission, ICU admission, and outpatient treatment.	30 day mortality	Age, diabetes (type 2), obesity (clinician- defined), pneumonia, chronic kidney disease, chronic obstructive pulmonary disease, immunosuppression	Score	Rounding of Cox regression coefficients (unpenalised)
Xie et al33	China	Adults (≥18 years) with confirmed covid-19, admitted in officially designated covid-19 treatment centres	In-hospital mortality	Age, lactate dehydrogenase, lymphocyte count, oxygen saturation	Prediction model	Logistic regression (unpenalised)
Hu et al34	China	Patients with severe covid-19 in Tongji Hospital, which specifically accommodated for people with covid-19. Patients directly admitted to intensive care unit were excluded. Patients with certain comorbidities (including cancer, uraemia, aplastic anaemia) were also excluded. Patients with a short hospital stay (<7 days) were excluded	In-hospital mortality	Age, high sensitivity C reactive protein, D-dimer, lymphocyte count	Prediction model	Logistic regression (unpenalised)
Zhang et al DCS and DCSL models35	China	Adults (≥18 years) admitted to two hospitals	In-hospital mortality	DCS model: Age, sex, diabetes (unspecified), immunocompromised, malignancy, hypertension, heart disease, chronic kidney disease, cough, dyspnoea DCSL model: Age, sex, chronic lung disease, diabetes (unspecified), malignancy, cough, dyspnoea, neutrophil count, lymphocyte count, platelet count, C reactive protein, creatinine	Prediction model	Logistic regression (lasso penalty)
Knight et al 4C Mortality Score36	UK	Adults (≥18 years) admitted across 260 hospitals	In-hospital mortality	Age, sex, number of comorbidities (chronic cardiac disease, respiratory disease, renal disease, liver disease, neurological conditions; dementia; connective tissue disease; diabetes (type 1 and 2); AIDS/HIV; malignancy, obesity), respiratory rate, oxygen saturation (room air), Glasgow coma scale score, urea, C reactive protein	Score	Rounding of logistic regression coefficients (lasso penalty)
Wang et al clinical and laboratory models37	China	Adults (≥18 years) admitted to hospital. Pregnant women were excluded	In-hospital mortality	Clinical model: Age, history of hypertension, history of heart disease Laboratory model: Age, oxygen saturation, neutrophil count, lymphocyte count, high sensitivity C reactive protein, D-dimer, aspartate aminotransferase, glomerular filtration rate	Prediction model	Logistic regression (unpenalised). Intercept from nomogram