Early in the course of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic it became evident that adults with preexisting type 1 or type 2 diabetes had a high risk of coronavirus disease 2019 (COVID-19)–related morbidity and mortality (1). Reports soon emerged of increased diagnoses of diabetes in hyperglycemic crisis, often copresenting with COVID-19 or soon after infection (2,3). In some cases, these were likely individuals in whom the viral infection had unmasked undiagnosed type 2 diabetes, yet in other cases, the presentation and subsequent clinical course was more in keeping with incident type 1 diabetes.
Type 1 diabetes pathogenesis, although autoimmune in nature, appears to require environmental triggers that likely include certain infections (4,5). A central and recurring question, therefore, has been whether the SARS-CoV-2 virus could be increasing type 1 diabetes presentations (6). Unsurprisingly, studies attempting to analyze this association have provided conflicting results that arise from broader challenges impeding rapid, clear answers (Fig. 1), including accounting for variation in COVID-19 symptomatology and testing, such that a negative test cannot exclude prior mild or asymptomatic infection; the unknown and likely heterogeneous time frame between infection and putative development of diabetes; intrinsic biases in the data sets studied; the challenge of type 1 diabetes diagnosis based purely on clinical presentation; and finally, and perhaps most importantly, difficulties segregating the effect of pandemic measures (such as changes to daily lifestyle during and after lockdown that underpin a rise in obesity [7] and changes in exposure to other endemic viral infections [5,8]) from direct exposure of an individual to the SARS-CoV-2 virus.
The study by Pietropaolo et al. (9) is a welcome addition to the growing literature, as it specifically examines incident cases of insulin-requiring diabetes in those COVID-19 positive versus negative, with a focus on children and young adults (aged 0–30 years), an age-group where diabetes presentations are likely to be underpinned by insulin deficiency. The authors hypothesized that, if SARS-CoV-2 is inducing insulin-requiring diabetes, disease incidence should be higher in those testing positive. Adopting a big-data approach and using a global electronic health database of >80 million deidentified patients across multiple, international health care organizations, they examined type 1 diabetes incidence in 293,358 individuals aged 0–30 years with a diagnosis of COVID-19 and 1,275,854 age-matched individuals without evidence of COVID-19 infection between January 2020 and June 2021. They adjusted for age, sex, race, ethnicity, and other comorbidities and based the diagnosis of type 1 diabetes on the presence of a diagnostic code or persistent insulin requirement at least 1 day after a diagnosis of COVID-19.
The study revealed a similar incidence of insulin-requiring diabetes in COVID-19–positive versus COVID-19–negative children between the ages of 0 and 18 years and a modest increase in the incidence between the ages of 19 and 30 years in those COVID-19–positive versus COVID-19–negative. No differences were observed in the incidence between COVID-19–positive and COVID-19–negative individuals for type 2 diabetes.
This study has several strengths: it is large and multicenter and reflects an international population. Other analyses of diabetes incidence have not specifically compared COVID-19–negative and COVID-19–positive groups (10), or if they have, the groups studied were not reflective of a broader population sample (11,12), with some focusing entirely on adults (13). It also examined data from a breadth of health care settings, including both hospitalized and primary care locations, incorporating patients who may not have been sick enough to seek hospital-based medical care, a limitation of other studies (11). The study also attempted to segregate type 1 from type 2 diabetes to mitigate the mixed signals observed when grouping all diabetes cases together, a limitation of other similarly designed studies (12).
As acknowledged by the authors, there are also limitations. First and foremost, it is probable that a proportion of those in the “negative” control population had COVID-19 but were not tested or diagnosed. This is especially true in pediatric populations, as children are often asymptomatic or mildly affected. Second, case numbers in the COVID-19–positive group were small, which may affect the power to detect differences. Third, there are also the usual but important issues around verification of coding for the diagnosis of type 1 diabetes: the authors attempt to mitigate this by studying young people who are more likely to have type 1 diabetes than type 2 diabetes and specifically analyzed those who were “insulin requiring”; however, the use of “insulin requiring” to determine type 1 diabetes would potentially also incorporate individuals with more severe presentations of type 2 diabetes or, indeed, stress hyperglycemia, and the authors were unable to verify coding with autoantibody data.
Together with the most current literature (10–12), this study investigates the recent suggestion that cases of insulin-requiring/type 1 diabetes have risen acutely in some populations. However, whether any observed changes are related to COVID-19 infection per se or to other factors remains unclear. The suggestion that COVID positivity has resulted in an increase of insulin-requiring diabetes in those aged 19–30 years but not in younger individuals is a tantalizing finding that, if genuine, could underscore differences in diabetes etiology between children and young adults or differences in clinical characteristics and diabetes risk factors. Of course, any putative effects of COVID-19 on type 1 diabetes incidence may also differ by age in ways that would not initially change the overall disease incidence and, ultimately, prevalence. Meanwhile, it is perhaps of little surprise that the lack of a difference in incidence (between COVID-19–positive and COVID-19–negative individuals) in this study between the ages of 0 and 18 years contradicts the recent Centers for Disease Control and Prevention analysis (12), given the differences in outcome definitions, data sets, study design, denominators, and methods of data collection.
Putatively, there are at least three mechanisms whereby direct effects of the SARS-CoV-2 virus could increase type 1 diabetes: 1) SARS-CoV-2 infection could be an environmental trigger precipitating type 1 diabetes in those individuals predisposed, 2) SARS-CoV-2 virus may directly infect and injure pancreatic β-cells (which would occur equally in those predisposed to autoimmune diabetes and those not) (6,14), and 3) COVID-19 might unmask those already developing type 1 diabetes earlier in the disease trajectory due to patients being acutely unwell, insulin resistant, and catabolic.
Given the interplay of multiple factors influencing diabetes diagnosis and incidence during the pandemic, although the data remain of interest and are critical to follow, it is unlikely that we will have crystal clear answers soon. Critical to this mission are analyses of population-level data sets, with attention paid to the limitations posed by intrinsic biases within data collection, carefully phenotyped diabetes to ascertain subtype, and perhaps most importantly the requirement for long-term follow-up of infected individuals (noting that analysis of the German–Austrian DPV data set at two time points during the pandemic showed contradictory findings [10,15]). These studies will take time. Until then, the putative induction of type 1 diabetes through direct and/or indirect effects of SARS-CoV-2 infections remains unproven. There will be continued speculation in the field, and we will witness more contradictory findings as this story crystallizes.
See accompanying article, p. 2656.
Article Information
Funding. S.M. is supported by the National Institute for Health Research Biomedical Research Centre at Imperial College Healthcare NHS Trust.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.