Demographic and Clinical Factors in COVID-19 Testing

We included the following independent demographic variables: age, sex, and ethnicity, using an ontology to maximise case identification;²² (link) practice-level deprivation using the English Index of Multiple Deprivation quintiles (we combined the two most deprived quintiles as there was a low frequency of testing, leading to sparse data, in the most deprived quintile);²³ household size based on pseudonymised patient address; and rural–urban classification. We included the latest recording of the following clinical variables, which are similar to those associated with increased susceptibility to influenza: body-mass index (BMI); smoking status; pregnancy; hypertension; chronic kidney disease; coronary heart disease; chronic respiratory disease, including asthma and chronic obstructive pulmonary disease; and type 1 and type 2 diabetes. We created a variable combining patients with malignancy and immunocompromise because there were small numbers in each group. Malignancy was identified using most recently recorded disease codes, and we used records of prescriptions for prednisolone and prescriptions for disease-modifying anti-rheumatic drugs as surrogates for immunosuppression. The outcome variable was testing positive for SARS-CoV-2.