Rising diabetes diagnosis in long COVID.

A few reports have raised the possibility that COVID-19 could increase the risk of type 2 diabetes.1, 2 In The Lancet Diabetes & Endocrinology, Xie and Al-Aly offer further evidence for the increased risk of diabetes beyond the first 30 days of infection (post-acute phase) by analysing the US Department of Veterans Affairs national health-care records of those who survived the first 30 days after a positive COVID-19 test. These data raise important questions about the relationship between COVID-19 and diabetes concerning causality, biological mechanisms, and implications for clinical care and public health.

Incidence of diabetes and antihyperglycaemic medication use was compared between 181 280 COVID-19 survivors, 4 118 441 contemporary controls without COVID-19 infection from the same year, and 4 286 911 historical controls from 2017. Diabetes was defined by International Classification of Diseases-10 diagnosis codes or an HbA1c of more than 6·4% (46 mmol/mol). The authors used inverse probability weighted survival analyses, including predefined and algorithmically selected high dimensional variables and reported two measures of risk; hazard ratio (HR) and burden per 1000 people at 12 months.

People who survived the first 30 days of SARS-CoV-2 infection exhibited an increased risk (hazard ratio [HR] 1·40, 95% CI 1·36–1·44) and excess burden (13·46, 95% CI 12·11–14·84) of incident diabetes per 1000 people at 12 months, compared with contemporary controls. COVID-19 survivors also had an increased risk (HR 1·85, 95% CI 1·78–1·92) and excess burden (12·35, 11·36–13·38) of incident antihyperglycaemic use. Results were consistent when historical controls were used. The association between COVID-19 and diabetes was consistent across strata of several potential confounders, namely, age, sex, race, BMI, diabetes risk score, steroid use, and area deprivation scores. The authors also did a full range of sensitivity analyses, including accounting for missing data with multiple imputation. Furthermore, an exposure–response relationship between severity of the acute infection, as shown by the care setting (non-hospitalised, hospitalised, or admitted to intensive care), and diabetes incidence was found. Notably, even those with mild, non-hospitalised COVID-19 had an excess 1-year burden of 8·28 per 1000 persons (95% CI 6·97–9·62). However, a major weaknesses of the study is the inclusion of a study population that was predominantly White and male. Furthermore, socioeconomic factors imposed by COVID-19 were not taken into account and there was no assessment for differential diabetes testing between cases and controls. Nevertheless, these data warrant further multipronged investigations.

Carefully designed multi-ethnic prospective epidemiological studies of long-term COVID-19 consequences (eg, incidence of diabetes types and associated conditions) across demographic and socioeconomic factors are needed to confirm or refute the findings from Xie and Al-Aly. Studies should use standardised approaches to accurately measure exposure to SARS-CoV-2 infection (eg, antigen and antibody tests) and diabetes (eg, using oral glucose tolerance tests). Factors postulated to influence long-term complications of COVID-19 should also be measured, including severity of infection, viral load, and the presence of antibodies signalling auto-immune attack. Variables that might help explain potential mechanisms for any association between COVID and diabetes should also be collected (eg, multiple point insulins, C-peptides, pancreatic and hepatic ectopic fat, and body composition), as well as co-occurring conditions, such as cardiovascular events, that could share disease mechanisms with diabetes pathogenesis. Studies should examine the extent that biological effects of SARS-CoV-2 infection are specific to diabetes or generalised by similar mechanisms across multiple systems.

Mechanistic studies will also be needed to explore the SARS-CoV-2 and diabetes connection. Among the most intriguing of hypotheses to explain the COVID-19 and diabetes connection is that SARS-CoV-2 virus might infect and replicate in the pancreas, injuring exocrine and endocrine cells.6, 7 Analysis of autopsy samples suggests that COVID-19 infection could lead to beta cell transdifferentiation mediated by the eIF2 signalling pathway. Other potential mechanisms, which the authors mention, are also worthy of investigation, including autonomic dysfunction, immune response or induced autoimmunity, and low grade inflammation. All of these factors should be considered within a real-world context of preadmission diabetes, steroid-induced diabetes, and sources of stress hyperglycaemia.

The data presented by Xie and Al-Aly have major implications for clinical policy and public health. If COVID-19 is indeed a risk factor for diabetes in the post-acute phase of infection, screening and management of dysglycaemia should be an integral part of clinical guidelines for COVID-19 diagnosis and follow-up. The long-term implications of SARS-CoV-2 infection increasing diabetes risk are profound. The rates of type 2 diabetes, and associated non-communicable diseases, are already growing, as shown on the diabetes atlas by the International Diabetes Federation. With large and growing numbers of people worldwide infected with SARS-CoV-2 (434 154 739 cumulative cases by Feb 28, 2022), any COVID-19-related increases in diabetes incidence could lead to unprecedented cases of diabetes worldwide—wreaking havoc on already over-stretched and under-resourced clinical and public health systems globally, with devastating tolls in terms of deaths and suffering.

The potential connection between COVID-19 and diabetes highlights that infectious diseases (eg, SARS-CoV-2) and chronic diseases (eg, diabetes) cannot be viewed in siloes. When we emerge out of the pandemic, the much-neglected non-communicable diseases, such as type 2 diabetes, will continue their relentless trajectory, possibly in an accelerated manner, as the leading burdens of global health.

We declare no competing interests. KMVN and LRS were partly supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health (award number P30DK111024) and Rapid Acceleration of Diagnostics (award number P30KD11024–05S1). LRS was also partly supported by Georgia Clinical and Translational Science Awards UL1 (UL1TR002378) and KL2 (KL2TR002381). The content is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health.