MULTIVARIABLE PREDICTION MODELS FOR HOSPITALIZATION, VENTILATION, AND DEATH FROM COVID-19 IN IBD PATIENTS

ABSTRACT

Background: Risk calculators can be an important tool for enabling shared decision making between patients and their health care providers. Demographics, comorbidities, medication use, geographic region, and other factors may increase the risk for complications from COVID-patients with inflammatory bowel disease (IBD). We developed a prognostic risk prediction tool for estimating the probability of hospitalization, intensive care unit (ICU) admission, and death due to COVID-19 in patients with IBD. Methods: Based on reports to Surveillance Epidemiology of Coronavirus Under Research Exclusion for Inflammatory Bowel Disease (SECURE-IBD) from March to October 2020, we modeled the probability of Hospitalization+ (a composite outcome of hospitalization, ICU and/or death), ICU+ (a composite outcome of ICU admission, intubation, and/or death) and Death separately using the Least Absolute Shrinkage and Selection set consisting of a random sample of cases reported between March 2020 and a pre-set cutoff date. Model validation was conducted using a test data set consisting of the remaining cases not in the training data plus all additional cases from one month past the cutoff date. We assessed the resulting models' discrimination using the area under the curve (AUC) of the receiver-operator characteristic (ROC) curves and corresponding 95% confidence intervals. Results: Overall, 2709 cases from 59 countries were included (mean age 41.3 years (s.d. 633 (24%) were hospitalized, 137 (5%) were admitted to ICU or intubated, and 69 (3%) died. The models have excellent discrimination, with an AUC and associated 95% confidence interval estimated on the test data set of 0.79 (0.75, 0.83) for Hospitalization+, 0.88 (0.82, 0.95) for ICU+, and 0.95 (0.91, 0.99) for Death. Age, comorbidities, corticosteroid use, and male sex were associated with higher risk of death while use of biologic therapies was associated with a lower risk of death (Figure 1). The online risk calculator is free and publicly available at https//covidibd.org/covid-19-riskcalculator/ for health care providers to facilitate discussion of the risk from COVID-19 with their IBD patients (Figure 2). After the physician inputs patient information, the prediction tool numerically and visually summarizes the patient’s probabilities of adverse outcomes intervals. Conclusion: Prognostic models can effectively predict who is at higher risk for COVID-19-related adverse outcomes in a population of IBD patients. The risk calculator could provide a basis for distinguishing between high and low-risk patients to aid in personalizing clinical guidance.