CORONADO is the first study specifically dedicated to people with diabetes infected with SARS-CoV-2 and admitted to hospital. CORONADO was designed to address three main goals: (1) assess the phenotypic characteristics of patients with diabetes hospitalised for COVID-19; (2) estimate the prevalence of the primary outcome, which combines death and tracheal intubation for mechanical ventilation within the first 7 days following admission; (3) identify in this specific population certain prognostic factors associated with early severity of COVID-19. When considering variables prior to admission, our results support no independent association between a severe course of COVID-19 and age, sex, long-term glucose control, chronic complications, hypertension or usual medications, including RAAS blockers and DPP-4 inhibitors. Only BMI turned out to be independently associated with the primary outcome. When considering variables on admission, dyspnoea, lymphopaenia, and increased AST and CRP levels were independent prognostic factors for severe course of COVID-19.
To our knowledge, CORONADO is the first study that provides precise information regarding the characteristics of diabetes in the severe forms of COVID-19. The study population roughly resembles the French population of people living with diabetes, except for HbA1c, which was clearly higher in our study (65 mmol/mol [8.1%]) compared with the nationwide ENTRED-2 survey participants older than 65 years (54 mmol/mol [7.1%]) . Of note, there was no overrepresentation of declared type 1 diabetes (only 3.0% of participants) in people with diabetes hospitalised for COVID-19.
The primary outcome occurred in 29.0% of CORONADO participants. While the design of the present study did not enable comparison of the severity of COVID-19 in people with or without diabetes, 20.3% of the study population required tracheal intubation for mechanical ventilation with a mortality rate of 10.6% as early as 7 days after admission. The severity of the prognosis of COVID-19 observed in people with diabetes in the present study is in accordance with previous epidemiological studies [10,11,12,13, 18, 19], and meta-analyses [14, 20]. An important issue is the choice of our primary endpoint, which combines death (an unequivocal outcome) with tracheal intubation for mechanical ventilation. It should be emphasised that the latter outcome can result from different factors, which were impossible to standardise in all centres, such as (1) clinical deterioration, (2) refusal to be intubated, or (3) futility (i.e. a medical decision not to intubate), leading to potentially fewer patients actually intubated compared with those meeting intubation criteria.
Regarding the clinical characteristics of COVID-19 in CORONADO participants, there was a high prevalence of fever and respiratory symptoms (cough, dyspnoea) and, to a lesser extent, digestive disorders. In addition to symptoms directly related to COVID-19, people with diabetes can also require management of acute metabolic disorders. In particular, physicians should be warned not only of the risk of ketoacidosis but also of hypoglycaemia, probably favoured by COVID-19-induced anorexia without concomitant adaptation of glucose-lowering drugs.
With the aim of providing clinicians with criteria to evaluate the risk of severe COVID-19 on an individual level in people with diabetes, we performed multivariable analyses to identify pre-admission and on-admission prognostic factors. Since some preclinical studies previously highlighted potential mechanistic links between glucose control, immune response and MERS-CoV infection , we were particularly interested in studying the relationship between long-term glucose control and COVID-19 prognosis. In fact, we failed to find any association between HbA1c (even with the highest values, >75 mmol/mol [9.0%]) and either the primary outcome or death on day 7. On the basis of this result and in order to increase the sample size for our analyses, we decided not to force HbA1c in the multivariate models.
An interesting finding is the association of BMI with study outcomes. Indeed, in our study, BMI was positively and independently associated with the primary outcome, which is largely driven by tracheal intubation. Interestingly, a recent report on COVID-19 patients in ICU showed an association between BMI and the requirement for mechanical ventilation, irrespective of diabetic status . However, such an association with BMI was no longer statistically significant when considering death on day 7. It should also be noted that the increased risk for the primary outcome appears to be less pronounced in patients with morbid obesity (grade 3, BMI ≥40 kg/m2) compared with those who were overweight or with grade 1–2 obesity, a situation previously described as the ‘obesity paradox’ in ICUs . Additional studies are clearly warranted to decipher the link between obesity, metabolic complications and COVID-19 severity with specific attention to fat mass distribution, insulin resistance and inflammatory/immune profiles.
While hypertension was previously reported as the most prevalent comorbidity in the general population with severe COVID-19 [2, 9, 12], it was not independently associated with the severity of the disease in the study. In addition, RAAS blockers (ACE inhibitors, ARBs and MRAs) were not independently associated with the main outcome, supporting the recent recommendation not to discontinue RAAS blockade . Moreover, we found no association between glucose-lowering drugs, including DPP-4 inhibitors, that have been suggested to potentially interfere with coronavirus infection and COVID-19 prognosis [21, 25].
Our complementary multivariable approach was suitable for the identification of characteristics on admission associated with COVID-19 prognosis, of particular relevance for the management of people with diabetes in the setting of an emergency room. Notably, we found an age- and sex-independent association between increased admission plasma glucose levels and the severity of COVID-19, as previously reported in critically ill patients . However, we speculate that this observation is rather the consequence of the severity of the infection than a causal primary factor.
Another important result concerns the identification of the prognostic factors of early death in people with diabetes and COVID-19. Compared with the primary outcome, which reflects aggressive management in ICUs with tracheal intubation, death on day 7 was more prevalent in elderly participants with an OR >14 for people older than 75 years, compared with younger individuals. In addition, these individuals also very frequently exhibited complications of diabetes (microvascular and macrovascular complications, mainly coronary heart disease) as well as pulmonary diseases (such as OSA). As expected, they were also more frequently on insulin therapy and taking multiple drugs (such as diuretics). Conversely, metformin use was associated with a reduced risk of early death, probably reflecting a less advanced stage of diabetes with fewer comorbidities (such as severe chronic kidney disease) that contraindicate its use. In multivariable analyses, age, diabetic complications and treated OSA remained significantly and independently associated with death on day 7. In addition, dyspnoea, reduced eGFR and platelet count, and increased AST and CRP on admission were independent markers of early death.
The discrepancy between the primary combined outcome (mainly driven by tracheal intubation) and death on day 7 could be explained by the fact that there were medical decisions not to pursue aggressive therapy in this frail population. In contrast, our data can be considered reassuring for the majority of people living with type 1 diabetes. Indeed, there was no death in participants with type 1 diabetes younger than 65 years. Additional data collection is currently ongoing to provide a precise picture of the rare individuals with type 1 diabetes hospitalised for COVID-19.
Some limitations must be acknowledged in the current analysis. We focused on hospitalised COVID-19 cases and our results cannot be generalised to all people with COVID-19 and diabetes, especially those with a less severe form of the disease. A secondary limitation is the size of our study population and the large proportion (i.e. 35.7%) of patients without available HbA1c. This is in accordance with the observation that only 55% of the people with diabetes had had three or more HbA1c determinations in the previous year according to French national registry data . Finally, the present report focuses only on short-term prognosis (i.e. 7 days after admission) and one cannot exclude the possibility that diabetes characteristics prior to admission could be associated with severe COVID-19 outcomes in the longer term. However, strengths must be acknowledged such as the originality of the medical question leading to the CORONADO initiative and the inclusion of participants on a national basis. In addition, a large majority (>93%) of COVID-19 cases were confirmed with a positive PCR test, with few cases diagnosed from medical and/or radiological observations only. We also structured data collection in order to obtain a precise and standardised recording of phenotypic characteristics of the diabetic study population.
In conclusion, the CORONADO study refined the phenotypes of COVID-19 individuals with diabetes admitted to hospital and showed that chronic glycaemic control did not impact the immediate severity of COVID-19. Elderly populations with long-term diabetes with advanced diabetic complications and/or treated OSA were particularly at risk of early death, and might require specific management to avoid contamination with SARS-CoV-2. BMI also appears as an independent prognostic factor for COVID-19 severity in the population living with diabetes, requiring hospital admission.