To the best of our knowledge, this study provides an analysis of the largest cohort yet assembled of people with type 1 diabetes admitted to hospital with COVID-19 infection for whom detailed clinical information is available. We report distributions of age, sex, ethnicity, diabetic complications, comorbidities, treatment on admission and laboratory test results, stratified by outcome.
The results from the present study gives reassurance to the younger type 1 diabetes population without diabetes complications with regards to their risk of severe COVID-19. However, five out of 67 (7%) patients aged <55 years old died. Of these five people, three were admitted to hospital primarily for COVID-19 and two for DKA and all of them had one or both of microvascular and macrovascular complications. There were no deaths in people with type 1 diabetes under the age of 55 years old without diabetic complications.
Overall, 49 patients were admitted to hospital with DKA (29% of 171 patients for whom data were available). This compares with 7.3% of patients who were admitted to hospital specifically for the management of diabetes in the 2019 UK National Diabetes Inpatient Audit [11]. This fourfold increase in DKA requiring hospital admission in people with type 1 diabetes is a significant concern both for the population as well as the healthcare system. Our findings emphasise the importance of supporting people with type 1 diabetes in the community during the pandemic, to minimise the incidence of avoidable admission. Health policy development may require population-specific attention to ethnicity as a risk factor for DKA, because DKA incidence varied fourfold between ethnic groups in a recently published US study [12]. We performed a further analysis to re-examine the association between blood glucose levels and the primary outcome after adjusting for both age and DKA. The results show a non-significant association with the primary outcome for blood glucose ≥20 mmol/l vs <10 mmol/l (OR 0.37 [95% CI 0.14, 1.03], p = 0.06) but the association did reach statistical significance when comparing blood glucose 10–19.9 mmol/l vs <10 mmol/l (OR 0.35 [95% CI 0.14, 0.89], p = 0.03). We excluded four patients who were admitted to hospital with blood glucose <4 mmol/l in this analysis. This unexpected finding of reduction in primary outcome (rate of death and AICU admission) in people with modestly elevated admission glucose compared with those with normal admission glucose needs further confirmation from a larger dataset.
We compared the clinical characteristics of this inpatient cohort with those of the general type 1 diabetes population included in the retrospective registry-based population study recently reported by the National Diabetes Audit (NDA) for England & Wales [13]. We found similar sex and ethnicity distribution (male, 55% vs 56%; white ethnicity, 70% vs 80%) but our cohort was of older age and relatively poor glycaemic control (proportion of age > 64 years, 45% vs 15%; HbA1c ≤ 75 mmol/mol [9.0%], 40% vs 71%) in COVID-19 inpatients compared with the general type 1 diabetes population. The crude (i.e. not adjusted for age) prevalence of diabetes-related complications was higher in COVID-19 inpatients than in the general type 1 diabetes population as reported in the 2017–2018 data report [14]: diabetic foot ulcer, 21% vs 5%; ischaemic heart disease, 24% vs 1.9%; cerebrovascular disease, 15% vs 0.5%. The complications data in the NDA report are extracted from primary care data which may be up to 15 months old, whereas our findings are based on contemporaneous admission records.
To our knowledge, few studies have been published on characteristics or outcomes in people with type 1 diabetes and COVID-19. The largest study to date has been the population-based study in England (population 61 million people) describing in-hospital mortality rate among 364 people with type 1 diabetes (1.45% of all hospital admissions due to COVID-19) [3]. The study showed that mortality was mainly confined to an older population, with no deaths occurring in those aged <50 years [3]. Another UK population study linked data between the NDA and the Office of National Statistics and compared the mortality rate between the year 2020 and the three previous years [15]. The study found that weekly death registrations in the first 19 weeks of 2020 exceeded the corresponding 3 year weekly averages for 2017–2019 by 50.9% in people with type 1 diabetes and that the increased COVID-19-related mortality rate was associated with cardiovascular/renal complications of diabetes and with glycaemic control and BMI [15]. A recent Scottish population study [16] included 51 (0.1%) of 34,383 people with type 1 diabetes who developed fatal or critical care unit-treated COVID-19 between 1 March 2020 and 31 July 2020. Overall, the risk of fatal or critical care unit-treated COVID-19 was increased by 2.4 times in those with type 1 diabetes compared with those without diabetes. Previous hospital admissions with hypoglycaemia or DKA were strongly associated with fatal or critical care unit-treated COVD-19. Compared with the general population of people with type 1 diabetes, those in the fatal/AICU admission group were older (mean age 71.4 vs 44.5 years), had higher mean BMI (27 vs 26 kg/m2), higher mean HbA1c (69 vs 67 mmol/mol), lower mean eGFR (72 vs 100 ml min−1 [1.73 m]−2), higher prevalence of heart disease (61% vs 14%) and lower insulin pump use (2% vs 14%). The patients included in this current study have not been excluded from other UK-based COVID registry studies. The present study included hospital admissions with type 1 diabetes and COVID from 1 March until 31 October 2020.
The Coronavirus-SARS-CoV-2 and Diabetes Outcomes (CORONADO) observational study from 68 French hospitals reported data for 56 people with type 1 diabetes [8]. In this cohort 55.4% were men, mean age was 56.0 years and mean BMI of 25.8 kg/m2. At 7 days, 11 patients (19.6%) had required tracheal intubation for mechanical ventilation, three had died (5.4%) and nine (16.1%) had been discharged. Overall, those with severe disease or who died were older (65.3 vs 53.2 years) and more likely to have hypertension (OR 5.21 [95% CI 1.24, 21.9]) than people who had not developed these outcomes within 7 days. Poor outcomes from COVID-19 in people with type 1 diabetes in this cohort were most strongly related to age, with no deaths occurring in those aged <65 years [8], whereas in a larger cohort with follow-up until discharge and a different healthcare system, we found a 7% mortality rate in people aged <55 years. The T1D exchange quality improvement collaborative (TIDX-QI), involving 64 sites in the USA, reported data on 33 patients with type 1 diabetes and laboratory-confirmed COVID-19 [7]. Among this study population, 54.5% were female and their mean age was 24.8 years and mean HbA1c was 69 mmol/mol (8.5%) [7]. The most prevalent comorbidities reported in this study were obesity (39.4%) and hypertension or CVD (12.1%). DKA occurred in 45.5% of patients [7].
Another US study conducted a retrospective chart review in 35 people with type 1 diabetes admitted to the Beth Israel Deaconess Medical Centre in Boston, MA, of which seven had COVID-19 [9]. Although the cohort was small, the study indicated that, compared with those who were COVID-19 negative, a higher proportion of SARS-CoV-2-positive patients were of non-Hispanic Black American ethnicity, with no significant differences in sex, body weight, glucose or HbA1c on admission. Only one patient in the COVID-19-positive group and two people in the COVID-19-negative group had DKA. The study showed similar glycaemic control prior to admission among people with type 1 diabetes, regardless of COVID-19 test results [9].
Our study has several strengths. The data were collected from a large number of centres across the UK, using a structured proforma with variables prespecified based on previous studies. This is the largest published cohort of people with type 1 diabetes for whom contemporaneous admission data are available. The data include ethnicity and outcomes to discharge or death, up to 70 days from admission, reflecting the prolonged hospital course of many patients with severe COVID-19. Limitations to our study include the retrospective nature of data collection and the absence of data from contemporaneous hospital admissions of people with type 1 diabetes without COVID-19 infection, and from people with type 1 diabetes and COVID-19 who were not admitted to hospital. We recognise that the patient population admitted to hospital are significantly different from the wider type 1 diabetes population and so the conclusions should be restricted to this subgroup of people. A number of centres had missing data for some of the variables (we report the completeness of variables in Table 1).
In conclusion, we report the largest study of people with type 1 diabetes admitted to hospital with COVID-19. In this population, higher BMI, worse renal function and the presence of microvascular complications were associated with higher risk of death and/or admission to AICU. However, no people with type 1 diabetes <55 years of age without microvascular or macrovascular complications died or were admitted to AICU.