Real‐world risk factors of confirmed or probable COVID‐19 in Americans with diabetes: A prospective, community‐based study (iNPHORM)

Abstract Introduction Americans with diabetes are clinically vulnerable to worse COVID‐19 outcomes; thus, insight into how to prevent infection is imperative. Using longitudinal, prospective data from the real‐world iNPHORM study, we identify the intrinsic and extrinsic risk factors of confirmed or probable COVID‐19 in people with type 1 or 2 diabetes. Methods The iNPHORM study recruited 1206 Americans (18–90 years) with insulin‐ and/or secretagogue‐treated type 1 or 2 diabetes from a probability‐based internet panel. Online questionnaires (screener, baseline and 12 monthly follow‐ups) assessed COVID‐19 incidence and various plausible intrinsic and extrinsic factors. Multivariable Cox regression was used to model the rate of COVID‐19 (confirmed or probable). Risk factors were selected using a repeated backwards‐selection ‘voting’ procedure. Results A sub‐sample of 817 iNPHORM participants (type 1 diabetes: 16.9%; age: 52.1 [SD: 14.2] years; female: 50.2%) was analysed between May 2020 and March 2021. During this period, 13.7% reported confirmed or probable COVID‐19. Age, body mass index, number of chronic comorbidities, most recent A1C, past severe hypoglycaemia, and employment status were selected in our final model. Body mass index ≥30 kg/m2 versus <30 kg/m2 (HR 1.63 [1.05; 2.52]95% CI), and increased number of comorbidities (HR 1.16 [1.05; 1.27]95% CI) independently predicted COVID‐19 incidence. Marginally significant effects were observed for overall A1C (p = .06) and employment status (p = .07). Conclusions This is the first US‐based epidemiologic investigation to characterize community‐based COVID‐19 susceptibility in diabetes. Our results reveal specific and promising avenues to prevent COVID‐19 in this at‐risk population. ClinicalTrials.gov Identifier: NCT04219514.


| INTRODUC TI ON
Despite extensive vaccine rollouts, the United States (US) continues to report the highest numbers of confirmed COVID-19 diagnoses and fatalities in the world. 1 Type 1 and 2 diabetes ranks as the second most common underlying health condition among US cases, 2 contributing to an elevated risk of severe outcomes. 3 Even for the majority reporting very mild infection, diabetes has shown to exacerbate post COVID-19 syndrome (i.e. 'long COVID'), 4 exposing an already highrisk population to further medical, social, and economic challenges. 4 Insight into the factors altering COVID-19 susceptibility in diabetes is needed to understand viral transmission and control over timeespecially as the virus becomes increasingly endemic. Hospital-based studies have done well to quantify the burden and predictors of severe COVID-19 morbidity and mortality in patients experiencing hyperglycaemia, 5 helping usher improved antiviral and treatment protocols. However, what remains unknown is the overall epidemiology of COVID-19 in the US public with diabetes, and strategies to prevent it.
Here, we analyse longitudinal, prospective data from the realworld iNPHORM (Investigating Novel Predictions of Hypoglycemia Occurrence using Real-world Models) study to ascertain the realworld, time-varying distribution and determinants of confirmed or probable COVID-19 in people with diabetes. The results of this study will be instructive for future clinical and public health strategies aimed at mitigating COVID-19 risk in one of America's most prevalent and vulnerable disease populations. 3

| Study design
The current evaluation describes a sub-analysis of the larger iN-PHORM panel survey: a 1-year prospective investigation of real-world hypoglycaemia risk stratification in the US. 6 Longitudinal self-assessed data were examined to determine the incidence proportion and related intrinsic and extrinsic factors of infection with the SARS-CoV-2 virus leading to COVID-19 between May 2020 and March 2021. Sub-panels A and B completed an online baseline questionnaire and up to 12 waves of follow-up (monthly questionnaires) that were emailed automatically by IIS; the follow-up schedule between panels was offset by 2 months (Sub-panel A: February 2020 to January 2021; Sub-panel B: April 2020 to March 2021).

| Participants and data collection
Participants were given 7 days to complete each follow-up questionnaire using various internet-equipped devices (e.g. computers, tablets and smartphones). All responses were synchronously stored on the IIS platform. Reminders and token incentives were distributed prospectively.
iNPHORM questionnaires (screener, baseline and follow-ups) were produced in English and pretested/piloted before fielding.

| COVID-19 status
Structured items were disseminated to classify respondents as confirmed or probable COVID-19 cases based on guidelines from the Centers for Disease Control and Prevention (April 2020). 7 Confirmed cases were those who reported having been medically diagnosed with COVID-19 by either ribonucleic acid or viral antigen assay. Probable cases were those who did not have a formal medical diagnosis but who reported (1) symptoms typical of COVID-19 (e.g. a cough, difficulty breathing, fever [over 100 degrees Fahrenheit], sore throat, headache, tiredness, or muscle aches and pains) and (2) ≥1 form of epidemiological exposure (close contact with confirmed/ suspected case or international travel). At each follow-up, participants were asked to report on their COVID-19 status since their last completed questionnaire.

| Potential risk factors
A broad range of plausible intrinsic and extrinsic COVID-19 risk factors were determined in consultation with the literature and diabetes experts. Information pertaining to these variables, including time-varying characteristics, were collated between the screener, baseline and follow-up questionnaires (  puted, bootstrapped backwards-selection models were retained in our final analysis. Coefficients were combined using Rubin's rules to correct for uncertainty due to missingness. 14 Two-sided significance tests (α = .05) were conducted using the median p-value of coefficients estimated for each multiply imputed dataset. 15 Analyses were performed in Stata 15 11 and R 4.1. 16

| Ethical considerations
Ethics approval was obtained from Western University's Research Ethics Board (#112986, December 9, 2020) prior to recruitment and updated upon addition of the COVID-19 sub-questionnaire. iN-PHORM was registered with www.Clini calTr ials.gov (NCT04219514, January 7, 2020). 17 Ipsos Interactive Services encrypted participant data to ensure confidentiality. Only de-identified data were transferred to the Western University research team. All participants provided informed consent before enrolment; they could withdraw at any time.   (Table S2).

| Risk factors of COVID-19 incidence
Age, BMI, number of chronic comorbidities, most recent A1C, previous severe hypoglycaemia and employment status were identified in ≥50% of backwards-selected models and retained for analysis ( Table 2). Table 3 summarizes the estimated cause-specific hazard ratios of our final multivariable model. Estimated causespecific hazard ratios for COVID-19 vaccination are reported in

| DISCUSS ION
Americans with type 1 and 2 diabetes with COVID-19 exhibit higher rates of morbidity, 3 mortality, 3

TA B L E 1 (Continued)
been identified as the most prevalent and predictive comorbidity of severe outcomes among hospitalized COVID-19 cases with diabetes. 18 We now provide new evidence that it may also augment SARS-CoV-2 initiation. 19 In people with obesity, low-level chronic inflammation can aggravate viral susceptibility by reducing macrophage activation, proinflammatory cytokine production, and B-and T-cell responses. 20 Adipose tissue can also act as a major source of inflammatory molecules, including interleukin-6 (IL-6), a gene implicated in the expression of angiotensin converting enzyme 2 (ACE2) and SARS-CoV-2. 19 With >85% of Americans with diabetes diagnosed as overweight/obese, our analysis signals a key opportunity for effective pandemic surveillance and risk mitigation. 21 But even more generally, we found that chronic diabetes comor- The role of glycaemia on COVID-19 susceptibility has been widely debated. In our study, haemoglobin A1C-a well-established marker of glycaemic control-was identified as a marginally signifi- Recent data out of Scotland demonstrated greater COVID-19 morbidity and mortality risk in people exposed to frequent low blood glucose. 27 As hypoglycaemia is known to induce inflammation, including IL-6 expression, and decrease immune responsiveness, it is not surprising that it may also exacerbate biologic predisposition to infection. 25 Amid challenges to sustain routine diabetes care during the pandemic, 28 it is essential that efforts to optimize glucose management not wane.
Interestingly, most socio-demographic factors (e.g. race) were not retained in our final model-perhaps due to insufficient power, or proximate clinical variables nullifying upstream effects.
Employment status emerged as the only marginally significant extrinsic risk factor in our study (p = .07). Relative to full-time workers, COVID-19 rates were 39% higher in part-time workers, TA B L E 2 Potential risk factors for contracting COVID-19 considered in the backwards-selection model .57 yet 32% lower in unemployed participants, retirees and students.
Employment-related environmental and behavioural dynamics could conceivably influence the probability of SARS-CoV-2 contact. 24 For example, part-time work (often on-site and involving close physical proximity), 29 in contrast to at-home (un)employment, may not only exacerbate environmental exposure, but also inhibit optimal prevention behaviour. 29 Clinical and public health strategies should account for the possible interplay between employment status and risk for COVID- 19. 24 Several additional variables were included in the analysis but failed to reach statistical significance. This could be due to a lack of association between these variables and susceptibility to COVID-19, or because our analysis was inadequately powered to identify these associations due to its smaller sample size.

| Strengths and limitations
This iNPHORM sub-analysis examines a large general cohort of American adults with type 1 or 2 diabetes recruited from a probability-based, real-world internet panel with high participation rates and minimal loss-to-follow-up. Our primary epidemiologic in-  ismpp.org/gpp3). grants.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.