Estimating risk of long COVID using a Bayesian network-based decision support tool

Background: Long COVID causes substantial health burden globally, affecting ~30.6% of adults who have ever had symptomatic COVID-19. Despite this, long COVID remains overlooked in public health decision-making. We built a model and easy-to-access online tool for exploring six-month long COVID risk factors. Methods: A Bayesian network model was developed to estimate long-term COVID-19 adverse outcome probability using data from published studies and government reports. The model calculates probabilities of hospitalization, ICU admission, and death, under different scenarios of vaccine coverage, sex, age, comorbidities, previous infection number, and drug treatments. The model also estimates six-month long COVID symptom risk including cardiovascular, gastrointestinal, musculoskeletal, pulmonary, or neurologic symptoms, kidney issues, metabolic problems, coagulation disorders, fatigue, and mental health problems. Results: Model estimates show incomplete vaccination, missed drug treatment during acute infection, and repeated infections to be the greatest controllable influences of increased long COVID risk. The model can be updated to include emerging best evidence, data pertinent to specific countries, vaccines, and outcomes. The interactive user-friendly web-based risk-assessment tool (part of the COVID-19 Risk Calculator (CoRiCal) suite), enables easy access to model outputs. Conclusions: This model and online tool can be used by individuals or in conjunction with clinicians for shared decision making on vaccination, pursuing early drug treatment during acute infection, and continuing protective behaviors such as masking and social distancing. It may also assist public health decision-makers to assess such effects at a population level, contributing to better-informed public health policies.