COVID-19 Disease in people with Diabetes in Scotland: Incidence, Severity and Risk Stratification using matched case-control and prospective cohort studies

Background: The risks of, and risk factors for, COVID-19 disease associated with diabetes are poorly quantified.

Methods: We identified as cases all those in Scotland with a positive SARS-CoV-2 nucleic acid test in the national laboratory database and anyone else with a death certificate mentioning COVID-19. Seven controls matched for age, sex and general practice were selected per case. Data were linked to the national diabetes register, hospitalisation and critical care unit (CCU) databases. Analyses focused on those with COVID-19 requiring critical care or dying. Analyses were by conditional and unconditional logistic regression.

Findings: 0.3% (n=845) of those with diabetes had developed severe or fatal COVID19, representing rate ratios of 3.86 (2.74, 5.45) in type 1 and 1.69 (1.56, 1.84) in type 2 diabetes. Rates were almost threefold in the most versus least socioeconomically deprived quintiles of the population. Most (77%) cases had another recognised comorbidity such as heart or lung disease (OR 2.6). Diabetes specific factors associated with increased risk included; HbA1c odds ratio (OR) >85 mmol/mol versus <53: 1.96(1.55,2.48);p-value <0.001, prior diabetic ketoacidosis: OR 2.44(1.41,4.20);p-value 0.001 and hypoglycaemia hospitalisations: OR 3.28(2.41,4.46);p-value <0.001. Chronic retinal and renal complications were also associated with increased risk. A crossvalidated predictive model of severe or fatal disease had a C-statistic of 0.83.

Interpretations: Relative risks of severe or fatal COVID-19 are substantially elevated in both types of diabetes. Risk scores based on prior clinical history should be useful for identifying those with diabetes needing tailored protective measures.