Impaired glucose metabolism in patients with diabetes, prediabetes, and obesity is associated with severe COVID‐19

Abstract Background Identification of risk factors of severe coronavirus disease 2019 (COVID‐19) is critical for improving therapies and understanding severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) pathogenesis. We analyzed 184 patients hospitalized for COVID‐19 in Livingston, New Jersey for clinical characteristics associated with severe disease. The majority of patients with COVID‐19 had diabetes mellitus (DM) (62.0%), Pre‐DM (23.9%) with elevated fasting blood glucose (FBG), or a body mass index >30 with normal hemoglobin A1c (HbA1C) (4.3%). SARS‐CoV‐2 infection was associated with new and persistent hyperglycemia in 29 patients, including several with normal HbA1C levels. Forty‐four patients required intubation, which occurred significantly more often in patients with DM as compared with non‐diabetics. Severe COVID‐19 occurs in the presence of impaired glucose metabolism in patients, including those with DM, preDM, and obesity. COVID‐19 is associated with elevated FBG and several patients presented with new onset DM or in DKA. The association of dysregulated glucose metabolism and severe COVID‐19 suggests that SARS‐CoV‐2 pathogenesis involves a novel interplay with glucose metabolism. Exploration of pathways by which SARS‐CoV‐2 interacts glucose metabolism is critical for understanding disease pathogenesis and developing therapies.

obesity (body mass index [BMI] > 40), chronic kidney disease and a history of heart failure were most associated with hospitalization, while critical illness was linked to low oxygen saturation (<88%) at admission, first d-dimer (>2500), first ferritin (>2500), and first C-reactive protein (>200) indicating hypoxia and inflammation in patients with clinically progressive disease. 6 A number of studies have identified an increased risk of severe disease in COVID-19 patients with underlying health conditions. Data compiled by the COVID-19 Associated Hospitalization Surveillance Network identified hypertension (49.7%), obesity (48.3%), chronic lung disease (34.6%), diabetes mellitus (DM) (28.3%), and cardiovascular disease (27.8%) as the most commonly found co-morbidities among hospitalized patients with COVID-19 in the United States. 3 A recent study of Lopinavir-Ritonavir in adults hospitalized with severe COVID-19 found 13% of patients had DM, reinforcing early observations that diabetes is a risk factor for more severe disease. 7 This is supported by data from a study of 24 patients hospitalized for COVID-19 in nine Seattle-area hospitals in which 58% of critically ill patients had DM and an average BMI of 33. 8 Interestingly, in the 2003 SARS-CoV outbreak in China, hyperglycemia and DM were also noted as risk factors for mortality and morbidity. 9 These observations and several in-depth reviews [10][11][12] have raised concerns that diabetics with elevated fasting blood glucose are at increased risk of developing severe COVID-19.
We report here our experience of 184 patients admitted for COVID-19 to a teaching hospital in Livingston, New Jersey within the epicenter of the SARS-Cov-2 pandemic in the United States. Extending early observations, we find the vast majority of our patients with COVID-19 are diabetic, prediabetic, or obese. Moreover, we identify COVID-19 patients with prediabetes (preDM) and others with normal hemoglobin A1c (HbA1C) levels who developed new onset DM, similar in presentation to type 1 DM, coincident with recent acquisition of SARS-CoV-2 infection. Our data establish that impaired glucose metabolism, due to either DM or obesity, is significantly associated with severe COVID-19 in this high-risk population.

| Diabetes status
A high percentage of patients testing positive for SARS-CoV-2 and referred to our practice were already known diabetics and receiving treatment for DM at the time of admission. We used the American Diabetes Association definitions to diagnose DM, new onset DM and preDM. 13 A new diagnosis of DM was made in patients previously unaware of their condition based on an HbA1C > 6.4%. New onset DM was defined by persistently elevated fasting blood glucose (FBG) > 125 mg/dL and requiring insulin therapy. PreDM was defined by an HbA1C of 5.7% to 6.4%. Nondiabetic patients were defined as having an HbA1C < 5.7% and FBG ≤ 125 mg/dL. Fever was defined as Tmax ≥100°F during the first 6 hours after admission. Hypoxia was defined as room oxygen saturation <94%.

| Outcomes
The primary indicator of severe COVID-19 was intubation. The need for intubation was determined on the basis of clinical presentation in patients receiving full care throughout their hospitalization. Death during hospitalization included patients put on comfort care at any time during or after admission. Comfort care measures were determined by the primary attending physician and included but were not limited to morphine drips or intensive care without further escalation of care. We developed a simple, scoring system for outcomes, based up a patient's diabetes status, BMI, A1C, Age, and initial blood glucose level. For details, please see Supplemental Information.

| Statistical analyses
A one-sample proportion Z-test was used to determine the prevalence of DM, preDM, and nonDM in patients with COVID-19 as compared with the US population. The sample size used for this analysis was 184 with at least 10 patients in each DM status. Onesided hypothesis tests were used to determine if the proportions of COVID-19 patients with DM and preDM were larger than the U.S. population proportions, and if the proportion of NonDM patients was smaller than the U.S. population proportion. A χ 2 test was used to determine significance between intubation and diabetes status within each patient group. 95% confidence intervals were calculated using standard errors. Statistical significance was defined as a P < .05.

| Increased prevalence of diabetes, pre-diabetes, and obesity
The majority of patients with COVID-19 had DM (62.0%), preDM (23.9%) or BMI > 30 with normal HbA1C (4.3%). The prevalence of DM was 4.7-fold higher in this patient group as compared with the general US population, while the prevalence of preDM was 1.3-fold higher. 14 A significant number of patients were clinically obese. The mean BMI of the study patients was 29.8 (17.5-61.4), including 20 patients with BMIs > 40. HbA1C levels measured at admission in 171 patients also showed significant elevation with 64 patients (37.4%) having values between 5.7% and 6.4% and 82 (48.0%) having values ≥6.5%.

| Age relationship to BMI, HbA1c, and initial blood glucose level
To determine whether patient age was associated with differences in clinical presentation, data on BMI, HbA1C, and initial FBG were stratified by age at admission. The rates of DM and preDM were similar in patients ≤60 years as compared with those >60 years ( Table 2) as were mean initial FBG levels (200.5 vs 165.4 mg/dL). However, patients ≤60 years of age were significantly more likely to be clinically obese. As compared with patients >60 years, the frequency of obesity and the mean BMI in those ≤60 years were significantly higher (26.6% vs 65.3% and 27.2 vs 33.4, respectively; P < .0001) ( Table 3). Patients ≤60 years were also significantly more likely to be severely obsese with a BMI > 40 (20.0% vs 3.7%; P = .0013). Similarly, patients ≤60 years had a significantly higher mean HbA1C level than older patients (8.0% vs 6.9%; P = .003) suggesting more pronounced metabolic dysregulation in younger patients. Taken together, these data indicate that younger patients may be more likely to present with abnormalities in glucose metabolism due to obesity, which may put them at increased risk of developing severe COVID-19. These findings are consistent with a recent report of 265 patients with COVID-19 demonstrating a significant  Our data in patients with severe COVID-19 and DM are consistent with a recent report by Bhatraju et al. 8 In both studies, 58% to 62% of patients severely ill with COVID-19 were diabetic with mean BMIs > 30 and the majority had elevated blood glucose. Additionally, we found 24% of patients with moderate-severe COVID-19 in our study were prediabetic. Taken together these data suggest that insulin resistance and uncontrolled glycemia play a significant role in worsening COVID-19. In all critically ill COVID-19 patients, blood glucose levels were elevated and tight glycemic control may therefore be an important consideration for improving clinical outcomes.
Several studies on patients with COVID-19 have reported on diabetes as a pre-existing diagnosis. In two recent observational studies, approximately 36% of patients with COVID-19 were diabetic. 16,17 These studies relied on passive surveillance at the time of admission. Binding of ACE2 by SARS-CoV-2 in COVID-19 also suggests that prolonged uncontrolled hyperglycemia, and not just a history of DM, may be important in the pathogenesis of the disease. 18 A known history of DM and ambient hyperglycemia were found to be independent risk factors for morbidity and mortality in SARS. 9 In a follow-up analysis of 135 patients, high fasting plasma glucose (FPG) was an independent predictor of SARS mortality. 19 Diabetes was found in 7.4% of a cohort of hospitalized patients with COVID-19 and appeared to be a risk factor for severity of disease. 20  Clinically, SARS-CoV-2 appears to cause new or worsening hyperglycemia, which may lead to more severe pneumonia. In our experience, a tipping point is reached in patients with COVID-19 who have symptoms lasting anywhere from 2 days to over 3 weeks and the disease then "takes off." Hospitalization before this acceleration can reduce the rate of critical illness.
It is important to note that our study has several limitations.
Patients were seen at a single clinical site and cared for by one group of clinicians. While it is possible our study population is disproportionately weighted towards patients with poor underlying health, the patients with COVID-19 in this study were consecutive referrals to our service over the course of 7 weeks in a suburban hospital. It is, therefore, unlikely that a selection bias exists, except safely and efficiently.