The disproportionate excess mortality risk of COVID-19 in younger people with diabetes warrants vaccination prioritisation

To the Editor: Diabetes has consistently been shown to independently increase the risk of poor coronavirus disease-2019 (COVID-19) outcomes [1–3]. Rather than a simple additive effect of diabetes and age-related risk, recent large studies suggest a more complex relationship, with a disproportionately higher excess relative mortality risk in younger people with diabetes compared with older people with diabetes [1, 3, 4]. Better understanding of the interaction between age and diabetes in the context of COVID-19 will further inform the complex prioritisation decisions around COVID-19 vaccination [5]. To explore this relationship, we triangulated the evidence on heterogeneity of diabetes effect by age on COVID-19 mortality from large population-based and critical care-based studies. Two UK population-based studies (OpenSAFELY [n = 17,278,392; 8.8% with diabetes] and QCOVID [n = 6,083,102; 7.0% with diabetes]) have previously reported adjusted age-specific hazard ratios for COVID-19-related mortality risk associated with diabetes [1, 4]. OpenSAFELY reported these stratified by recent HbA1c measurements as recorded in primary care (</≥58 mmol/mol [</≥7.5%] or not available), but not by diabetes type. QCOVID reported agespecific hazard ratios for type 2 diabetes by sex, but did not report age-specific values for type 1 diabetes (see electronic supplementary material [ESM] Methods). The overall 90-day COVID-19-re la ted mor ta l i ty ra te was 0.06% in OpenSAFELY (study period: 1 February 2020–6 May 2020) and the 97-day COVID-19-related mortality rate was 0.07% in QCOVID (derivation cohort study period: 24 January 2020–30 April 2020). Building on our previous analysis in the critical care setting [3], we also examined adjusted age-specific hazard ratios associated with type 2 diabetes in people with severe COVID-19 using the COVID-19 Hospitalisation in England Surveillance System (CHESS) cohort (n = 19,256 individuals admitted to critical care in England; 18.3% with type 2 diabetes; see ESM Methods). In this cohort, the 30-day in-hospital mortality rate was 26.4% (see ESM Table 1 for age-stratified mortality rates). To aid interpretability of our findings, based on the work of Spiegelhalter [6, 7], we translated hazard ratio estimates into ‘COVID-age’, which represents the additional years of COVID-19 mortality risk added to an individual’s chronological age if diabetes is present. Full details of our approach are provided in the ESM Methods. The study was reviewed and approved by the Warwick BSREC (BSREC 119/19-20). The additional COVID-19 mortality risk associated with diabetes is, in terms of COVID-age, markedly higher in younger than older people (Fig. 1). This reflects the higher relative risk of COVID-19-related mortality associated with diabetes in younger age groups (hazard ratios for diabetes >5 in adults under 50 years of age in population-based studies). Population-based and critical care-based estimates are similar, despite differences in setting, time period and adjustments for * John M. Dennis j.dennis@exeter.ac.uk