Family history of diabetes and risk of SARS‐COV‐2 in UK Biobank: A prospective cohort study

Abstract Introduction The aim of this study was to determine risk of being SARS‐CoV‐2 positive and severe infection (associated with hospitalization/mortality) in those with family history of diabetes. Methods We used UK Biobank, an observational cohort recruited between 2006 and 2010. We compared the risk of being SARS‐CoV‐2 positive and severe infection for those with family history of diabetes (mother/father/sibling) against those without. Results Of 401,268 participants in total, 13,331 tested positive for SARS‐CoV‐2 and 2282 had severe infection by end of January 2021. In unadjusted models, participants with ≥2 family members with diabetes were more likely to be SARS‐CoV‐2 positive (risk ratio‐RR 1.35; 95% confidence interval‐CI 1.24–1.47) and severe infection (RR 1.30; 95% CI 1.04–1.59), compared to those without. The excess risk of being tested positive for SARS‐CoV‐2 was attenuated but significant after adjusting for demographics, lifestyle factors, multimorbidity and presence of cardiometabolic conditions. The excess risk for severe infection was no longer significant after adjusting for demographics, lifestyle factors, multimorbidity and presence of cardiometabolic conditions, and was absent when excluding incident diabetes. Conclusion The totality of the results suggests that good lifestyle and not developing incident diabetes may lessen risks of severe infections in people with a strong family of diabetes.


| INTRODUC TI ON
People with existing health conditions are at higher risk of SARS-CoV-2 infection 1 and related mortality, including people with diabetes. 2 A meta-analysis from 30 different studies found more than twice higher risk of SARS-CoV-2 disease and mortality among patients with diabetes. 3 Family history (parental or sibling) of diabetes is a strong risk factor for type 2 diabetes, with stronger familiality (ie >one family member affected) being associated with higher risk. 4,5 Previous research has suggested that in addition to genetic risk, lifestyle-related factors such as adiposity and dietary habits may explain some of this observed association. 4,5 The risk for SARS-CoV-2 infection and poor outcomes among those with family history of diabetes has not been previously investigated. Given the current uncertainty about the course and duration of the pandemic continues, and the prevalence of diabetes in many populations worldwide, the role of whether family history of diabetes in predisposition to damaging SARS-CoV-2 outcomes is important to know as diabetes can be prevented. We investigated the association of family history of diabetes with the risk of developing SARS-CoV-2 and severe SARS-CoV-2, and the potential role of demographic and lifestyle factors in such a link.

| Ethical approval
This study was conducted as part of UK Biobank project number 14151 and is covered by the generic ethics approval for UK Biobank studies from the NHS National Research Ethics Service (16/ NW/0274). All participants gave written informed consent before enrolment in the study, which was conducted in accordance with the principles of the Declaration of Helsinki.

| Exposure variable
Family history of diabetes (mother/father/sibling) was reported at the time of recruitment and categorized into none, one, or two or more, according to the number of family members with a history of diabetes.

| Outcome variables
Confirmed SARS-CoV-2 infection (≥one positive result) and severe SARS-CoV-2 infection (defined as testing positive during an inpatient hospital episode).

| Sociodemographics, lifestyle and underlying health conditions
Age at recruitment was used as a continuous variable. Sex and ethnicity (categorized into 'Asian or Asian British', 'Black or Black British', 'Chinese', 'Mixed', 'Other ethnic group' or 'White' groups) were used as categorical variables. Townsend score, a measure of socioeconomic status based on participant postcode, was used as a continuous score. 6 Smoking status self-reported at the time of UKB study recruitment was categorized as never, previous or current.
Body mass index (BMI) was measured by a study nurse at the time of UKB study recruitment and used as a continuous variable. Level of physical activity was defined as 'none', 'low', 'medium' or 'high'

| Statistical analysis
We used Poisson regression models with robust standard errors to test for cross-sectional associations between family history of diabetes and SARS-CoV-2 infection or severe SARS-CoV-2. Unadjusted estimates were presented followed by stepwise adjustments. The models were first adjusted for sociodemographic variables (age, sex, Townsend score and ethnicity), then for lifestyle variables (smoking status, BMI and physical activity levels), and finally for presence of cardiovascular conditions at the time of study recruitment (hypertension, ischaemic heart disease and stroke), incident admissions for cardiometabolic conditions (hypertension, diabetes, ischaemic heart disease and stroke) and multimorbidity count.

| Sensitivity analysis
The main analysis was repeated for the subset of participants who were tested positive for SARS-CoV-2, with the risk of severe SARS-CoV-2 as the outcome variable. The above analysis was also repeated after excluding participants who were diagnosed with diabetes (based on hospitalization records) since the study recruitment to the onset of COVID pandemic.
All analyses were carried out using R version 3.6.1, using the 'glm' function and 'quasipoisson' formula.

| RE SULTS
After exclusions noted above, 401,268 were included for analyses (please see Figure 1). On recruitment, 70,427 participants (17.5%) reported having one family member with diabetes, and 12,443 (3.1%) reported having two or more family members with diabetes (Table 1).
In total, 13,331 (3.32%) tested positive for SARS-CoV-2 and 2282 (0.6%) participants had severe SARS-CoV-2 (ie infection detected in hospital or through death). The proportion of study sample testing positive of SARS-CoV-2 infection was 3.2%, 3.65% and 4.33% respectively for those with no, one and two or more family members with diabetes. Participants with at least one family member with diabetes had higher average HbA1c than participants with no family history (see Table 1). In the unadjusted model, participants with two or more family members with diabetes were more likely to test positive for SARS-CoV-2 infection (risk ratio (RR) 1 Table 2 shows the results of association between family history of diabetes and risk of severe SARS-CoV-2 only for the subset of participants who were tested positive for SARS-CoV-2. The results

| Sensitivity analysis
show that there was no significant association between family history of diabetes and severe SARS-CoV-2 in the subset who tested positive. Table 3 shows the results of sensitivity analysis after excluding participants who were diagnosed with diabetes (based on hospitalization records) since the study recruitment to the onset of COVID pandemic. In the sensitivity analysis, there was significant association between family history of diabetes and testing positive for SARS-CoV-2 infection; however, risk was attenuated after adjusting for demographics, lifestyle factors and underlying health. In this group, there was no association between family history of diabetes of severe COVID-19 infections.

| DISCUSS ION
Given diabetes is a strong prognostic factor for SARS-CoV-2 severity and mortality, 9 relatives of such patients may be rightly concerned that they are also at higher risk. The present analyses have shown this to be the case with a 35% higher chance of a positive SARS-CoV-2 test and a 30% higher chance of severe SARS-CoV-2 infection in those with two first-degree relatives with diabetes.
However, it appears as if this risk may be mitigated by better lifestyles which also lessen the risk of comorbidity. Lifestyle factors such as smoking, BMI and physical activity are likely to influence the risk of COVID-19 and/or associated complications as evidenced by previous studies. A separate study using UK Biobank data found higher risk of severe SARS-CoV-2 with physical inactivity, smoking and obesity but not heavy alcohol consumption. 10 In addition, relationship between unhealthy lifestyle and severe disease risk was dose dependent. Another study by the COVID-19 host genetics initiative found similar association between high BMI, smoking and physical activity and higher risk severe COVID-19 illness using mendelian randomization approach. 11 Hopkinson et al. found that people who smoke were at higher risk of developing symptomatic COIVD-19, based on data from 2.4 million UK users of the Zoe COVID-19 symptom study. 12 A study in United States found that patients with COVID-19 who were physically inactive had a higher risk of hospitalization, ICU admission and mortality, with some evidence of dose-response relationship. 13 We accept some limitations. UK Biobank participants are relatively healthier and more affluent than general UK population although there is some evidence that risk factor associations with health outcomes are generalizable. 14 The work did not differentiate Also, we could not fully adjust for all incident diabetes; only admissions for diabetes were adjusted for. In addition, information on lifestyle and underlying health was based on self-report and these data were collected more than 10 years ago. During the early study period of the 'first wave' of the outbreak in England, SARS-CoV-2 testing was highly selective and limited to only those participants who presented with severe symptoms. There was no association found between family history of diabetes and severe SARS-CoV-2 risk when analysis was limited to those who tested positive to reduce the bias due to selective testing. That said, this sensitivity analysis by no means excludes bias as those initially tested for SARS-CoV-2 in the UK had more severe symptoms and it is impossible to ascer-   Severe SARS-CoV-2 defined as participants who tested positive during an inpatient hospital episode.
Results in bold suggest statistical significance (p < 0.05).