Algorithm for Individual Prediction of COVID-19 Hospitalization from Symptoms: Development and Implementation Study (Preprint)

BACKGROUND The coronavirus disease 2019 (COVID-19) pandemic has generated a huge strain on the health care system worldwide. The metropolitan area of Milan, Italy was one of the most hit area in the world. OBJECTIVE Risk prediction models are needed to stratify individual patient risk for public health purposes. METHODS A predictive algorithms was developed in 36.834 COVID-19 patients in order to foresee the risk of being hospitalized. Exposures considered were age, sex, comorbidities and symptoms associated with COVID-19 (vomiting, cough, fever, diarrhoea, myalgia, asthenia, headache, anosmia, ageusia, and dyspnoea). Discrimination and calibration of the model were assessed. RESULTS The predictive model showed a good fit for predicting COVID-19 hospitalization (C-index 0.79), a good overall prediction accuracy (Brier score 0.14) and was well calibrated (intercept -0.0028, slope 0.9970). Using these results, 118,804 patients with COVID-19 from October 25 to December 11, 2020 were stratified into low, medium and high risk for COVID-19 severity. Among the overall population, 67.030 (56%) were classified as low-risk, 43.886 (37%) medium-risk, and 7.888 (7%) high-risk, with 89% of the overall population being assisted at home, 9% hospitalized, and 2% dead. Among those assisted at home, most people (60%) were classified as low risk, whereas only 4% were classified at high risk. According to ordinal logistic regression, the OR of being hospitalised or dead was 5.0 (95% CI 4.6-5.4) in high-risk patients and 2.7 (95% CI 2.6-2.9) in medium-risk patients, as compared to low-risk patients. CONCLUSIONS A simple monitoring system, based on primary care datasets with linkage to COVID-19 testing results, hospital admissions data and death records may assist in proper planning and allocation of patients and resources during the ongoing COVID-19 pandemic. CLINICALTRIAL