Back to Journals » Diabetes, Metabolic Syndrome and Obesity » Volume 14

Newly-Diagnosed Diabetes and Sustained Hyperglycemia are Associated with Poorer Outcomes in COVID-19 Inpatients Without Pre-Existing Diabetes

Authors Lin L, Chen Z, Ding T, Liu H, Zhou F, Huang X, Zhang X, Liu W, Zhang BH, Yuan Y, Zhang P, Zhang XJ, She ZG , Cai J, Chen W, Li H

Received 5 August 2021

Accepted for publication 9 October 2021

Published 9 November 2021 Volume 2021:14 Pages 4469—4482

DOI https://doi.org/10.2147/DMSO.S332819

Checked for plagiarism Yes

Review by Single anonymous peer review

Peer reviewer comments 3

Editor who approved publication: Prof. Dr. Antonio Brunetti



Lijin Lin,1– 4,* Ze Chen,4,5,* Ting Ding,1,2,* Hui Liu,4,6,* Feng Zhou,4,7 Xuewei Huang,3,4 Xingyuan Zhang,4,8 Weifang Liu,4,8 Bing-Hong Zhang,9 Yufeng Yuan,10 Peng Zhang,4,8 Xiao-Jing Zhang,4,8 Zhi-Gang She,3,4 Jingjing Cai,4,11 Wenping Chen,1,2 Hongliang Li2– 4,7

1Department of Endocrinology, Huanggang Central Hospital, Huanggang, People’s Republic of China; 2Huanggang Institute of Translational Medicine, Huanggang, People’s Republic of China; 3Department of Cardiology, Renmin Hospital of Wuhan University, Wuhan, People’s Republic of China; 4Institute of Model Animal of Wuhan University, Wuhan, People’s Republic of China; 5Department of Cardiology, Zhongnan Hospital of Wuhan University, Wuhan, People’s Republic of China; 6Neonatology of Gastroenterology, Tongren Hospital of Wuhan University & Wuhan Third Hospital, Wuhan, People’s Republic of China; 7Medical Science Research Center, Zhongnan Hospital of Wuhan University, Wuhan, People’s Republic of China; 8Basic Medical School of Wuhan University, Wuhan, People’s Republic of China; 9Neonatology Department, Renmin Hospital of Wuhan University, Wuhan, People’s Republic of China; 10Department of Hepatobiliary and Pancreatic Surgery, Zhongnan Hospital of Wuhan University, Wuhan, People’s Republic of China; 11Department of Cardiology, The Third Xiangya Hospital, Central South University, Changsha, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Hongliang Li; Wenping Chen Email [email protected]; [email protected]

Purpose: To analyze the impact of hyperglycemia on the clinical outcome of COVID-19 in patients with newly diagnosed diabetes (NDD).
Patients and Methods: We performed a retrospective study of 3114 cases of COVID-19 without pre-existing diabetes, 351 of which had NDD, in Hubei Province, China. The Cox regression model was used to calculate the risk of adverse clinical outcomes comparing the NDD vs non-NDD group before and after propensity score-matched (PSM) analysis. Patients with NDD were further divided into a sustained hyperglycemia group, a fluctuating group, and a remitted group based on their blood glucose levels during hospitalization as well as into hypoglycemic agent users and nonusers.
Results: Compared to the non-NDD individuals, individuals with NDD had a significantly increased risk of all-cause mortality (adjusted HR after PSM, 2.65; 95% CI, 1.49– 4.72; P = 0.001) and secondary outcomes involving organ damage during the 28-day follow-up period. Subgroup analyses indicated that among individuals with NDD, the individuals with remitted hyperglycemia had the lowest 28-day mortality, whereas those with sustained hyperglycemia had the highest (IRR 24.27; 95% CI, 3.21– 183.36; P < 0.001). Moreover, individuals treated with hypoglycemic agents had significantly lower all-cause mortality than those not treated with hypoglycemic agents (IRR 0.08; 95% CI, 0.01– 0.56; P < 0.001).
Conclusion: Our study reinforces the clinical message that NDD is strongly associated with poor outcomes in COVID-19 patients. Furthermore, resolved hyperglycemia in the later phase of the disease and the use of hypoglycemic agents were associated with improved prognosis in patients with NDD.

Keywords: COVID-19, newly diagnosed diabetes, hyperglycemia, prognosis, blood glucose

Introduction

The outbreak of coronavirus disease 2019 (COVID-19) continues to escalate with particular intensity around the world. As of 10 September 2021, there have been more than 223 million confirmed cases of COVID-19 globally, including 4,602,882 deaths. Diabetes mellitus (DM) is one of the most common comorbidities in patients with COVID-19.1,2 A large number of studies almost unanimously showed that COVID-19 patients with DM are more likely to develop severe illness and have a higher risk of mortality than non-DM patients.3,4 Intriguingly, several recent studies suggest a bidirectional relationship between DM and COVID-19.4,5 Indeed, new-onset hyperglycemia is observed in COVID-19 patients without pre-existing diabetes.6,7 More importantly, currently available evidence hints that COVID-19 patients with newly diagnosed diabetes (NDD) have poorer outcomes than those with normoglycemia and those with pre-existing diabetes.8,9

Despite the significance of these findings, most of the current evidence implicating NDD in worse COVID-19 prognosis has come from relatively limited sample cohorts.9–11 In addition, most of these studies used baseline blood glucose for diagnosis rather than repeated measurements obtained during hospitalization.9,11–13 Another unsettled issue is that an independent association is difficult to determine due to the existence of substantial confounding factors. For instance, concerns have been raised that in-hospital use of glucocorticoids may mediate the detrimental effects of NDD on COVID-19 severity.14 Additionally, it is unclear whether the alteration of glucose metabolism that occurs with a sudden onset in severe COVID-19 persists or remits during the later phase of the disease. Moreover, our previous study demonstrated that well-controlled blood glucose is associated with a significant reduction in mortality in COVID-19 patients with pre-existing diabetes.15 This indicates that glycemic management can be crucial in improving COVID-19 outcomes in diabetic patients. However, whether unsatisfactory glucose control mediates the impact of NDD on COVID-19 outcome has yet to be investigated.

Materials and Methods

Study Design and Participants

This retrospective, multicenter study involved 7871 original participants diagnosed with COVID-19; each of the participants had at least two fasting plasma glucose (FPG) records and was admitted to one of 17 hospitals in Hubei Province, China between December 30th, 2019 and April 12th, 2020 (Figure 1). COVID-19 was diagnosed based on chest computed tomography (CT) and/or reverse transcription-polymerase chain reaction (RT-PCR) following WHO interim guidance16 and the criteria of the New Coronavirus Pneumonia Prevention and Control Program (5th edition) published by the National Health Commission of China.17 This study was performed in compliance with the ethical principles of the Declaration of Helsinki. The study protocol was approved by the central ethics committee of Renmin Hospital of Wuhan University and Zhongnan Hospital of Wuhan University. The study design was also individually approved by each collaborating hospital or by the hospital’s institutional ethics board. The requirement to obtain informed consent from the study participants was waived by the ethics committees of the individual hospitals due to the urgency of the COVID-19 pandemic. Personal identification information (eg, name and ID) of the study subjects were anonymized and replaced with a coding system before data extraction. Among the initially included participants, those for whom complete electronic medical records were not available (eg, transfer to any other hospital), individuals <18 or >75 years of age, pregnant women, and individuals with acute lethal organ injury (ie, acute coronary syndrome, acute stroke, and severe acute pancreatitis) were excluded. Subjects with a previous history of diabetes or with a previous history of using glucose-lowering medication and those who received glucocorticoid treatment during hospitalization were also excluded from our final analysis (Figure 1).

Figure 1 The flow chart of patient inclusion and analysis procedures in the study. aThe participants with at least three times fasting blood glucose records. bThe particients with newly-diagnosed diabetes.

Abbreviations: FBG, fasting blood glucose; NDD, newly-diagnosed diabetes; HA, hypoglycemic agents.

Data Collection

The medical records of the patients were analyzed by an integrated research group that included physicians, data scientists, and statisticians. Basic information, clinical characteristics, laboratory findings, radiographic manifestations from CT, therapeutic intervention, and outcomes during hospitalization were extracted from electronic medical records. Major clinical symptoms (ie, fever, cough, fatigue, dyspnea, diarrhea, and comorbidities) were collected. Laboratory findings, including a routine blood test, FPG, 2-h postprandial blood glucose (2hPG) and random blood glucose, C-reactive protein (CRP), procalcitonin, D-dimer, and serum biochemical markers of liver injury, kidney injury and cardiac dysfunction, were also recorded during hospitalization. In-hospital medication and life support intervention included the types of drugs administered, oxygen inhalation treatment and use of mechanical ventilation. The primary and secondary outcomes were evaluated by physicians. To guarantee accuracy and consistency, the participants’ medical records were reviewed, confirmed, and double-checked by experienced physicians.

Definitions and Outcomes

The diagnosis of NDD was confirmed by at least two FPG readings ≥7 mmol/L according to the Chinese Guidelines for the Prevention and Treatment of Type 2 Diabetes (2017 edition) and the American Diabetes Association (ADA) guidelines criteria.18,19 Among patients with NDD and ≥3 FPG records during hospitalization, we conducted subgroup analyses in which we divided these patients into a sustained hyperglycemia group (FPG ≥7 mmol/L in all records), a remitted group (FPG ≥7 mmol/L at the first two records but <7 mmol/L at later tests), and a fluctuating group (the remaining participants). We also analyzed NDD patients in subgroups based on whether or not they received hypoglycemic agents (HA, either insulin or oral hypoglycemic agents).

The primary endpoint was 28-day all-cause death. The secondary endpoints were the occurrence of acute respiratory distress syndrome (ARDS), acute liver injury, acute kidney injury, acute cardiac injury, or heart failure. ARDS was defined according to the WHO interim guideline “Clinical management of severe acute respiratory infection when novel coronavirus (2019-nCoV) infection is suspected”. Acute kidney injury was diagnosed by an elevation in serum creatinine level ≥26.5 µmol/L within 48 hours.20 Acute cardiac injury was defined as a serum level of cardiac troponin I/T (cTnI/T) above the upper limit of normal (ULN).21 Acute liver injury was determined based on serum alanine transaminase (ALT) or alkaline phosphatase levels more than 3-fold the ULN.22

Propensity Score-Matched Analysis

We used propensity score-matched analysis (PSM) to match the patients with and without NDD.23 Baseline matching variables with a standardized difference (SD) greater than 0.10 were selected; they are as follows: age, sex, oxygen saturation, respiratory rate, pre-existing comorbidities (coronary heart disease, hypertension, chronic kidney disease, and chronic liver disease), and biomarkers indicative of disease severity and organ injury (ie, neutrophil count increase, CRP increase, lymphocyte count decrease, alanine aminotransferase (ALT) increase, creatinine increase, high-sensitivity C-reactive protein (hs-CRP) increase, procalcitonin increase, HDL-C decrease, triglyceride (TG) increase, and bilateral lesions on chest CT).

Mixed-Effects Cox Model

The Cox proportional hazards model was used to calculate the risk of primary and secondary endpoints and the corresponding hazard ratio (HR) comparing the NDD group with the non-NDD group before and after PSM. Baseline covariates that changed the HRs by at least 10% when added to the Cox model and that had an SD greater than 0.10 between the groups were adjusted.24,25 In the mixed-effects Cox model, we modeled the site as a random effect and used correlation testing based on the Schoenfeld residuals to verify proportional hazard assumptions.25 In the Cox analysis, discharged patients were treated as having no competing risk, but their data were not censored. The reasons for this are as follows: one, individuals with COVID-19 would not be discharged unless continuous viral PCR was negative twice in succession and their symptoms were relieved; two, because it was necessary to quarantine individuals for two weeks after their discharge from the hospital, any deaths that occurred among these patients would be documented. Discharged individuals were less likely to die from COVID-19 than patients who remained hospitalized, and information on their survival after discharge was still available.25

Missing Data and Imputation

A complete set of variables for each patient was used for matching in PSM analysis and for adjustment in Cox analysis. We used nonparametric missing value imputation based on the missForest procedure in R to impute the missing data on the noninvasive test.26 A random forest model based on the remaining variables in the dataset was constructed and used to predict the missing values with an estimation of the internally cross-validated errors.23

Statistical Analysis

Continuous variables are expressed as median and interquartile range (IQR), and categorical variables are presented as frequency and percentage (%). For continuous variables, Student’s t-tests (for normally distributed data) and Mann–Whitney U-tests (for nonnormally distributed data) were used to analyze comparisons between groups. For categorical variables, Fisher’s exact test and the chi-square test were used to analyze comparisons. Dynamic changes in the levels of laboratory indicators in different groups were presented using locally weighted scatterplot smoothing (LOESS). The risk for endpoint outcomes and corresponding HR were analyzed using the mixed-effects Cox model. E-value analysis was performed to address potential underlying confounding effects and to assess the robustness of the association between NDD and all-cause mortality in the Cox model.27 Person-time data (incidence) of two groups with different exposures (ie, the sustained hyperglycemia group or the fluctuating group vs the remitted group, or NDD patients with HA vs those without) are expressed as the difference between incidence rates (IRs) or as incidence rates ratios (IRRs). The IRRs of endpoint outcomes were calculated to estimate the difference in absolute change in the incidences of two comparison groups.25 The cumulative rates of death were compared by applying the Kaplan–Meier method.28 A difference with a two-sided α less than 0.05 was considered statistically significant. All statistical analyses were performed using R-3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Clinical Characteristics of COVID-19 Patients with and without NDD Upon Admission

A total of 3114 of 7871 patients with COVID-19 admitted to 17 hospitals in Hubei Province, China were included in our final analyses; among them, 351 (11.27%) were classified as NDD cases (192 males, 54.70%), and 2763 (88.73%) were designated as non-NDD cases (1198 males, 43.36%) (Figure 1). Patients with NDD were older (61 years vs 56 years, P < 0.001), more likely to be male (54.70% vs 43.36%, P < 0.001) and had more symptoms of dyspnea (30.48% vs 18.02%, P < 0.001) than non-NDD patients. In addition. NDD patients had a higher burden of pre-existing comorbidities, including hypertension (37.89% vs 25.73%, P < 0.001), chronic renal disease (5.70% vs 1.63%, P < 0.001), chronic liver disease (3.13% vs 1.70%, P = 0.097), coronary heart disease (7.69% vs 5.57%, P = 0.140), and cerebrovascular disease (3.70% vs 1.95%, P = 0.053) than non-NDD patients. The median interval from symptom onset to admission in patients with or without NDD was 11 (IQR, 7–17) and 13 (IQR, 7–21) days, respectively (Table 1).

Table 1 Characteristics of COVID-19 Patients in the Newly-Diagnosed Diabetes (NDD) and Non-NDD Groups Before and After Propensity Score Matching

Compared to patients without NDD, higher proportions of patients with NDD had lymphopenia (55.04% vs 27.80%), increased leukocyte counts (19.60% vs 4.34%), or increased neutrophil counts (27.38% vs 7.22%). Blood glucose levels in NDD cases were higher than those in the non-NDD group (7.60 [6.50–9.49] mmol/L vs 5.18 [4.70–5.83] mmol/L). Elevated serum markers indicating inflammation (CRP [59.60% vs 32.87%] and procalcitonin [51.03% vs 28.02%]), liver injury (AST [28.39% vs 15.83%]), abnormal kidney function (creatinine [10.40% vs 4.00%] and urea nitrogen [12.17% vs 2.55%]), and coagulation disorder (D-dimer [59.28% vs 38.95%]) were found more frequently in the NDD group than in the non-NDD group. Additionally, SpO2 ≤ 93% occurred more frequently in patients with NDD than in those without NDD (28.83% vs 8.96%) (Table 1).

Dynamic Changes in Inflammatory Markers During Hospitalization

To determine the changes in parameters that are indicative of inflammation in the patients, multiple measurements of inflammation indicators were performed during hospitalization, and the results were recorded. LOESS models were used to illustrate the dynamic changes in FPG, lymphocytes, neutrophil count, lactate dehydrogenase (LDH), TNFα, and IL-6 in the NDD and non-NDD groups during hospitalization (Figure 2). All these parameters except lymphocyte counts were elevated more significantly upon admission and maintained at a higher level during the later phase of the disease in patients with NDD than in patients without NDD.

Figure 2 Dynamic profiles of BG (A), neutrophils (B), lymphocytes (C), LDH (D), TNFα (E), and IL-6 (F) in NDD and Non-NDD groups during hospitalization. aThe relative levels of LDH, TNFα and IL-6 were applied and normalized according to their upper limits of the normal range in each hospital.

Abbreviations: BG, blood glucose; NDD, newly-diagnosed diabetes; LHD, lactate dehydrogenase; TNFα, tumor necrosis factor-α; IL-6, Inter Leukin 6.

COVID-19 Patients with NDD Required More Intensive in-Hospital Treatment

As shown in Table 1, participants with NDD received more intensive interventions than did the non-NDD cases; this was manifested by the higher proportions of NDD patients who required treatment with vasoactive drugs (3.70% vs 0.29%), immunoglobin (21.94% vs 12.96%), oxygen inhalation (88.03% vs 75.14%), invasive mechanical ventilation (IMV) (19.66% vs 0.51%), noninvasive ventilation (24.22% vs 3.62%), renal replacement therapy (4.56% vs 0.14%), and intensive care unit (ICU) treatment (22.79% vs 4.31%) (Table 1).

NDD Was Strongly Associated with All-Cause Mortality and Multiorgan Damage in Patients with COVID-19

We observed that in-hospital death was significantly higher (18.52% vs 1.16%) among patients with NDD than among non-NDD individuals during the 28-day follow-up period (Table S1). In the mixed-effects Cox model using the hospital site as a random effect, the crude HR for all-cause mortality between the two groups was 15.17 (95% CI, 9.89–23.29; P < 0.001) (Table 2). After adjustment for confounding factors, including age, sex, comorbidities, and indicators of COVID-19 severity, the aHR for all-cause mortality between the two groups was 3.63 (95% CI, 2.24–5.88; P < 0.001) (Table 2). Considering the effects of unmeasured potential confounders, we conducted E-value analysis and found that the E-value (6.72 with CI 3.91 in the fully adjusted model) was substantially greater than the accepted risk factors for COVID-19 mortality. The Kaplan–Meier survival curves also illustrated that the NDD group had a significantly higher mortality than the non-NDD group (Figure S1). In addition, compared with the non-NDD subjects, patients with NDD had higher occurrence of multiorgan damage, including ARDS (35.90% vs 7.42%), acute kidney injury (13.11% vs 0.25%), acute liver injury (15.10% vs 5.68%), acute heart injury (23.93% vs 2.57%), and heart failure (33.05% vs 8.54%) (Table S1). After applying a mixed-effects Cox model and adjusting for age, sex, comorbidities, and severity of COVID-19, the respective HRs for ARDS, acute kidney injury, acute liver injury, acute heart injury, and heart failure were 2.53 (95% CI, 1.96–3.26; P < 0.001), 15.70 (95% CI, 6.67–36.98; P < 0.001), 2.17 (95% CI, 1.51–3.13; P < 0.001), 3.72 (95% CI, 2.61–5.29; P < 0.001), and 1.93 (95% CI, 1.50–2.49; P < 0.001), respectively, between the two groups (Table 2).

Table 2 Hazard Ratios for Outcomes in NDD and Non-NDD Groups Under Cox Adjusted Model and Propensity Score Matching Model

To determine the robustness and reliability of the association between NDD and clinical outcomes, we further conducted a PSM analysis in which 307 individuals with NDD and 614 non-NDD cases were matched at a ratio of 1:2 (Figure 1). Using a mixed-effects Cox model to further adjust imbalanced variables after matching (ie, total bilirubin and dyspnea), the association between patients with NDD and poor outcomes remained consistent and statistically significant, indicating a higher risk of all-cause mortality for patients with NDD than for non-NDD subjects (aHR, 2.65; 95% CI, 1.49–4.72; P = 0.001) (Tables 2 and S2). The Kaplan–Meier survival curves also showed that the NDD group had a significantly higher mortality after PSM (Figure 3). Moreover, after PSM, NDD was associated with a significantly increased risk of secondary endpoints, including ARDS, acute kidney injury, acute liver injury, acute heart injury, and heart failure, compared with non-NDD, with adjusted HRs of 2.36 (95% CI, 1.73–3.23; P < 0.001), 14.65 (95% CI, 4.27–50.27; P < 0.001), 2.21 (95% CI, 1.43–3.40; P < 0.001), 2.80 (95% CI, 1.82–4.31; P < 0.001), and 1.67 (95% CI, 1.24–2.25; P = 0.001), respectively (Tables 2 and S2).

Figure 3 Kaplan-Meier curves for cumulative probability of COVID-19 mortality during hospitalization in NDD and Non-NDD groups in the propensity score-matched model.

Abbreviation: NDD, newly-diagnosed diabetes.

Sustained Hyperglycemia Was Intensively Associated with Poor Outcomes in COVID-19 Patients with NDD

Subgroup analyses indicated that among patients with NDD who had ≥ 3 FPG records, 13.93(39/280) had sustained hyperglycemia throughout hospitalization; hyperglycemia was resolved in 18.21%(51/280) of patients, and the remaining 67.86%(190/280) showed a fluctuating pattern (Table S3). The dynamic trajectory of the patients’ blood glucose levels is depicted in Figure S2. The IR of death during the 28-day follow-up was 1.79 cases per 100 person-days in the subgroup with sustained hyperglycemia, 0.76 cases per 100 person-days in the fluctuating group, and 0.07 cases per 100 person-days in the remitted group (Table 3). Compared to the individuals with remitted hyperglycemia, those with sustained hyperglycemia had the highest 28-day mortality (IRR 24.27; 95% CI, 3.21–183.36; P < 0.001), followed by the fluctuating group (IRR 10.31; 95% CI, 1.41–75.13; P = 0.004) (Table 3). The Kaplan–Meier survival curves also illustrated that patients with sustained hyperglycemia had the highest mortality risk, followed by patients with a fluctuating pattern, while patients with resolved hyperglycemia had the lowest risk (Figure 4). Similar trends in secondary outcomes were observed, as shown in Table 3.

Table 3 Association of Dynamic Blood Glucose Patterns with 28-Day Poor Outcomes in Patients with NDD

Figure 4 Survival curves of newly-diagnosed diabetes patients with sustained, fluctuant, and remitted blood glucose levels.

Using HA Was Correlated with a Reduced Risk of Adverse Outcomes in COVID-19 Patients with NDD

To further explore whether using HA was correlated with a reduced risk of primary and secondary outcomes in patients with NDD, we conducted another subgroup analysis of NDD patients in which we compared patients using HA with nonusers. The baseline characteristics of the two groups are shown in Table S4. Among the 55 patients using HA, 16 (29.09%) used insulin, 21 (38.18%) used oral hypoglycemic agents (including metformin, glycosidase inhibitors, dpp4 inhibitors, nateglinide, and sulfonylureas) and 18 (32.73%) used both insulin and oral hypoglycemic agents (Tables S5 and S6). Only one of the 55 patients who used insulin treatment died (Table S7). As shown in Figure 5, the dynamic trajectory of FPG in nonsurvivors who did not use HA was distinct from that in survivors with/without HA. The FPG levels in nonsurvivors without HA were maintained at higher levels during hospitalization, whereas those in survivors gradually decreased. Similar trends were observed in the levels measured in all blood glucose tests (including FPG, 2 h-PG, and random blood glucose). The in-hospital death rate was markedly lower in NDD individuals taking HA than in nonusers (n = 1, 1.82% vs n = 64, 21.62%; P = 0.001). The IR of all-cause mortality was 0.07 cases per 100 person-days in HA users and 0.90 cases per 100 person-days in nonusers. Compared to nonusers, patients using HA had lower all-cause mortality (IRR 0.08; 95% CI, 0.01–0.56; P < 0.001). Those using HA also had a lower occurrence of secondary outcomes, including ARDS (IRR 0.32; 95% CI, 0.16–0.63; P < 0.001), acute kidney injury (IRR 0.35; 95% CI, 0.11–1.14; P = 0.069), acute liver injury (IRR 0.78; 95% CI 0.35–1.72; P = 0.533), acute heart injury (IRR 0.22; 95% CI 0.08–0.61; P = 0.001), and heart failure (IRR 0.60; 95% CI 0.34–1.07; P = 0.078), than nonusers (Table 4).

Table 4 Association of Hypoglycemic Agents (HA) Use with 28-Day Poor Outcomes in Patients with NDD

Figure 5 Dynamic profiles of fasting blood glucose (A) and blood glucose (B) in survivors with HA, survivors without HA, and non-survivors without HA from patients with newly-diagnosed diabetes during hospitalization.

Abbreviation: HA, hypoglycemic agents.

Discussion

In our analysis of 3114 COVID-19 patients without pre-existing diabetes who did not receive corticosteroid treatment during hospitalization, NDD was associated with a significantly higher risk of in-hospital death and of secondary endpoints such as ARDS and acute organ injuries. Furthermore, among patients with NDD, those who displayed sustained hyperglycemia throughout the hospitalization period were at the highest risk of poor outcomes. Patients taking HA had a significantly lower occurrence of adverse outcomes than nonusers. Our study is the largest to date to investigate the relationship between NDD and COVID-19 outcomes and the first to indicate that sustained hyperglycemia is associated with poorer outcomes in COVID-19 patients with NDD.

Several previous studies have shown a link between NDD and COVID-19 outcomes, as summarized in recent reviews and meta-analyses.6,29–31 It is important to note that these studies used very different criteria to define NDD. In a single-center retrospective analysis of 453 Chinese COVID-19 patients, Li et al found NDD (defined as admission FBG ≥ 7 mmol/L and/or HbA1c ≥ 6.5%) in 21% of the cases (94/453) and reported a significant increase in all-cause mortality (HR 9.42; 95% CI, 2.18–40.7) in those patients compared to patients with normoglycemia (FBG < 5.6 mmol/L and HbA1c < 5.7%).9 In addition, patients with NDD had a higher percentage of admissions to the ICU (11.7%), and more of them required invasive mechanical ventilation (11.7%) than did patients with normal glucose levels (1.5% and 2.3%, respectively). Similarly, Wang et al conducted a retrospective study of COVID-19 patients at two hospitals in China and reported that 29% (176/605) of such patients had NDD (defined by admission FBG ≥ 7 mmol/L). Patients with NDD had a higher risk of in-hospital complications (HR 3.99; 95% CI, 2.71–5.88) and all-cause death (HR 2.30; 95% CI, 1.49–3.55) than individuals with baseline FBG <6.1 mmol/l.11 Moreover, in a retrospective study of 413 COVID-19 patients in an Italian hospital, Fadini et al reported NDD (defined as HbA1c ≥ 6.5% or a random glucose level ≥ 11.1 mmol/L) in 5% of the cases (21/413) and found a significantly higher risk of severe COVID-19 (ICU admission or death; RR 3.06; 95% CI, 2.04–4.57) in those patients compared to patients with normoglycemia.10 The prevalence of NDD in our cohort was 11.27%, lower than that reported in the studies by Li et al9 and Wang et al11 but higher than that reported in the study by Fadini et al.10 A recent meta-analysis of 3711 COVID-19 cases in eight studies showed a pooled proportion of patients with NDD of 14.4% (13.4% in China).6 The differences in NDD prevalence among studies result in part from differences in the diagnostic criteria used. In the present study, NDD was defined as at least two FPG measurements ≥ 7 mmol/L, in accordance with the ADA criteria.19 However, due to the urgency of the circumstances during the COVID-19 outbreak, HbA1c and symptoms of hyperglycemia were not measured or recorded in our study. Despite differences in the criteria used to define NDD, the present study and earlier studies have consistently indicated that NDD is significantly associated with adverse outcomes in COVID-19 patients.9–11,30

The existence of confounding factors may bias the association between NDD and COVID-19 outcomes to a large extent.32 For instance, in-hospital use of glucocorticoids may represent an important confounder that was not properly adjusted for or matched in the aforementioned studies, in which a greater proportion of individuals with NDD received glucocorticoid treatment compared to individuals with known diabetes and those with normal blood glucose.10,14,33 To avoid this confounding effect, we excluded participants who received in-hospital glucocorticoid therapy. In addition, inflammatory activation may mediate the association between NDD and COVID-19 severity.9,32,34 In our study and prior studies,9–11 inflammatory markers such as neutrophil count, CRP, hs-CRP, and procalcitonin levels were significantly higher in patients with NDD than in those without NDD at admission and in the later phase of hospitalization. As has been well studied in patients with severe pneumonia, overactivated inflammatory responses could drive stress hyperglycemia and a severe disease course.35,36 Thus, there is a possibility that a high blood glucose level might simply represent a biomarker of more severe disease. Nevertheless, in our analyses, the aHRs for the outcomes remained consistent and statistically significant after rigorous adjustment and matching, indicating that the association was independent of baseline confounders, including markers of inflammation. It is also noteworthy that prior studies consistently found that COVID-19 patients with NDD had poorer prognoses than did patients with pre-existing diabetes.9–11,30 This could be partially explained by the fact that diabetes is often associated with manifest organ impairment that can be accounted for clinically and statistically,10,37 In contrast, individuals who are unaware of their diabetes status may have occult organ impairment that is likely to be ignored by their treating physicians, and this impairment may not be accounted for in the statistical adjustment.10 Nonetheless, the association between NDD and COVID-19 outcomes is unlikely to be due only to possible occult multiorgan damage, given the relatively large adjusted effect sizes in our study and previous studies.10

Previous studies have reported that hyperglycemia is usually transient in patients hospitalized with severe acute respiratory syndrome (SARS).38 However, whether the alterations in glucose metabolism that occur with sudden onset of COVID-19 persist or remit during the later phase of the disease and what the impact of these changes are on the prognosis are unclear.39 Montefusco et al applied continuous glucose monitoring in a cohort of patients hospitalized for COVID-19 in Italy and reported glycemic alterations not only in the acute phase of COVID-19 but also long after remission of the disease. Indeed, glycemic abnormalities could be detected for at least 2 months in patients who recovered from COVID-19.40 Consistent with that study, our subgroup analyses showed that only 18.21% of patients with NDD displayed a remitted blood glucose pattern and that 13.93% had sustained hyperglycemia during the 28-day hospitalization. Moreover, among patients with NDD, those with sustained hyperglycemia were at the highest risk of poor outcomes, while the remitted group had the lowest risk, further suggesting that hyperglycemia may influence disease progression. The possible mechanisms by which NDD might contribute to poor COVID-19 outcomes include metabolic inflammation, an impaired innate immune response, possibly an altered level of angiotensin-converting enzyme 2 (ACE2), vascular dysfunction and the existence of a prothrombotic state due to hyperglycemia.1,2 Nevertheless, the pathophysiology of COVID-19-related diabetes is complex and remains unclear.7 The possible role of hyperglycemia as a driving force in COVID-19 progression needs to be confirmed in future studies.

A number of studies have provided clinical evidence linking improved blood glucose control to better outcomes in COVID-19 patients with pre-existing diabetes.15,41–44 For example, the study by Zhu et al, which included 7337 individuals with COVID-19, indicated that patients with well-controlled blood glucose (glycemic variability within the range 3.9 to 10.0 mmol/L) experienced significantly lower mortality than patients with poorly controlled blood glucose (upper limit of glycemic variability exceeding 10.0 mmol/L) during hospitalization (adjusted HR, 0.14).15 However, whether unsatisfactory glucose control mediates the impact of NDD on COVID-19 outcome has not yet been investigated. Our subgroup analyses demonstrated that patients with NDD whose hyperglycemia later resolved had the best prognosis among the three subgroups studied. In addition, NDD patients using HA had significantly lower in-hospital death and fewer secondary outcomes than nonusers. These results provide further evidence linking blood glucose control and COVID-19 prognosis in patients with NDD. However, the effects of specific types of HA on COVID-19 prognosis remains largely unknown. Evidence from Yu et al showed that insulin treatment was associated with enhanced systemic inflammation and aggravated injuries of vital organs,45 whereas metformin has shown benefits against COVID-19 through mechanisms besides lowering blood glucose such as attenuating inflammation and heart injury.46 Thus, the independent effect of HA on clinical outcomes warrants to be further studied.

Limitations

Our study has notable strengths, including large sample size, rigorous control of confounders, and the utilization of repeated measured blood glucose data. Nonetheless, it has several limitations. First, due to the retrospective nature of the study, a causal relationship between NDD and COVID-19 outcomes cannot be established. Second, some clinical variables were unavailable for all patients due to the urgent circumstances caused by the COVID-19 outbreak. For example, HbA1c was not measured in most patients, and this might have biased the detection of pre-existing diabetes. Third, blood glucose level determinations were conducted at different time intervals for each patient. Bias might occur due to more frequent testing of patients with severe illness. Fourth, a sample of 3114 subjects may allow for reasonable power in the study, but the subgroup analysis may be underpowered due to the small sample size. Moreover, due to the limited number of events, we were unable to further analyze the independent relationship between the use of specific HA and clinical outcomes or to investigate whether the beneficial effects of HA on clinical outcome are independent of glycemic conditions. Fifth, all data were obtained from hospitalized patients in 17 hospitals in Hubei Province, China. Thus, the results from our study cannot be extrapolated to the outpatient setting or to ethnically or geographically diverse populations without careful validation.

Conclusion

In conclusion, our study reinforces the clinical message that NDD is strongly associated with poor outcomes in COVID-19 patients without pre-existing diabetes. Furthermore, our study provides preliminary evidence linking unsatisfactory blood glucose control to poor COVID-19 outcomes in NDD patients. Therefore, it seems reasonable that COVID-19 patients with NDD should be under intensive blood glucose surveillance and should be managed carefully to achieve tight glucose control, similar to patients with known diabetes, although further evidence from clinical trials is urgently needed.

Acknowledgments

This work was supported by grants from the National Key R&D Program of China (grant number 2016YFF0101504, 2020YFC2004702), the Special Foundation for Emergency Research on Prevention and Control of COVID-19 of Guangdong Province (grant number 2020B1111330003), the National Science Foundation of China (grant number 81630011, 81970364, 81970070, 81970011, 81770053, 81870171), the Hubei Science and Technology Support Project (grant number 2019BFC582, 2018BEC473), and Medical flight plan of Wuhan University (grant number TFJH2018006).

Disclosure

The authors report no conflicts of interest in this work.

References

1. Drucker DJ. Diabetes, obesity, metabolism, and SARS-CoV-2 infection: the end of the beginning. Cell Metab. 2021;33(3):479–498. doi:10.1016/j.cmet.2021.01.016

2. Lim S, Bae JH, Kwon HS, Nauck MA. COVID-19 and diabetes mellitus: from pathophysiology to clinical management. Nat Rev Endocrinol. 2021;17(1):11–30.

3. Bornstein SR, Rubino F, Khunti K, et al. Practical recommendations for the management of diabetes in patients with COVID-19. Lancet Diabetes Endocrinol. 2020;8(6):546–550. doi:10.1016/S2213-8587(20)30152-2

4. Apicella M, Campopiano MC, Mantuano M, et al. COVID-19 in people with diabetes: understanding the reasons for worse outcomes. Lancet Diabetes Endocrinol. 2020;8(9):782–792. doi:10.1016/S2213-8587(20)30238-2

5. Papachristou S, Stamatiou I, Stoian AP, et al. New-onset diabetes in covid-19: time to frame its fearful symmetry. Diabetes Ther. 2021;12(2):461–464. doi:10.1007/s13300-020-00988-7

6. Sathish T, Kapoor N, Cao Y, et al. Proportion of newly diagnosed diabetes in COVID-19 patients: a systematic review and meta-analysis. Diabetes Obes Metab. 2021;23(3):870–874. doi:10.1111/dom.14269

7. Li G, Chen Z, Lv Z, et al. Diabetes mellitus and COVID-19: associations and possible mechanisms. Int J Endocrinol. 2021;2021:7394378. doi:10.1155/2021/7394378

8. Sathish T, de Mello GT, Cao Y. Is newly diagnosed diabetes a stronger risk factor than pre-existing diabetes for COVID-19 severity? J Diabetes. 2021;13(2):177–178. doi:10.1111/1753-0407.13125

9. Li H, Tian S, Chen T, et al. Newly diagnosed diabetes is associated with a higher risk of mortality than known diabetes in hospitalized patients with COVID-19. Diabetes Obes Metab. 2020;22(10):1897–1906. doi:10.1111/dom.14099

10. Fadini GP, Morieri ML, Boscari F, et al. Newly-diagnosed diabetes and admission hyperglycemia predict COVID-19 severity by aggravating respiratory deterioration. Diabetes Res Clin Pract. 2020;168:108374. doi:10.1016/j.diabres.2020.108374

11. Wang S, Ma P, Zhang S, et al. Fasting blood glucose at admission is an independent predictor for 28-day mortality in patients with COVID-19 without previous diagnosis of diabetes: a multi-centre retrospective study. Diabetologia. 2020;63(10):2102–2111. doi:10.1007/s00125-020-05209-1

12. Sardu C, D’Onofrio N, Balestrieri ML, et al. Hyperglycaemia on admission to hospital and COVID-19. Diabetologia. 2020;63(11):2486–2487. doi:10.1007/s00125-020-05216-2

13. Coppelli A, Giannarelli R, Aragona M, et al. Hyperglycemia at hospital admission is associated with severity of the prognosis in patients hospitalized for COVID-19: the Pisa COVID-19 study. Diabetes Care. 2020;43(10):2345–2348. doi:10.2337/dc20-1380

14. Li G. Inpatient use of glucocorticoids may mediate the detrimental effect of new-onset hyperglycemia on COVID-19 severity. Diabetes Res Clin Pract. 2020;168:108441. doi:10.1016/j.diabres.2020.108441

15. Zhu L, She Z-G, Cheng X, et al. Association of blood glucose control and outcomes in patients with COVID-19 and pre-existing Type 2 diabetes. Cell Metab. 2020;31(6):1068–1077e3. doi:10.1016/j.cmet.2020.04.021

16. World Health Organization. Laboratory Testing for 2019 Novel Coronavirus (2019-nCoV) in Suspected Human Cases: Interim Guidance. World Health Organization; 2020.

17. National Health Commission of China. New Coronavirus Pneumonia Prevention and Control. National Health Commission of China Program; 2020. Available from: http://www.nhc.gov.cn. Accessed October 16, 2021.

18. Chinese Diabetes Society. Guidelines for the prevention and treatment of Type 2 diabetes in China (2020 Edition). Chin J Endocrinol Metab. 2021;37(04):311–398.

19. American Diabetes Association. 2. Classification and diagnosis of diabetes: standards of medical care in diabetes-2020. Diabetes Care. 2020;43(Suppl 1):S14–S31.

20. Kellum JA, Aspelin LN, Barsoum RS, et al. Kidney disease: improving global outcomes (KDIGO) acute kidney injury work group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl. 2012;2:91–138.

21. Yancy CW, Jessup M, Bozkurt B, et al. 2017 ACC/AHA/HFSA focused update of the 2013 ACCF/AHA guideline for the Management of Heart Failure: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Failure Society of America. Circulation. 2017;136(6):e137–e161.

22. Marrone G, Vaccaro FG, Biolato M, et al. Drug-induced liver injury 2017: the diagnosis is not easy but always to keep in mind. Eur Rev Med Pharmacol Sci. 2017;21(1 Suppl):122–134.

23. Waljee AK, Mukherjee A, Singal AG, et al. Comparison of imputation methods for missing laboratory data in medicine. BMJ Open. 2013;3:8. doi:10.1136/bmjopen-2013-002847

24. Jaddoe VW, de Jonge LL, Hofman A, et al. First trimester fetal growth restriction and cardiovascular risk factors in school age children: population based cohort study. BMJ. 2014;348:g14. doi:10.1136/bmj.g14

25. Zhang XJ, Qin J-J, Cheng X, et al. In-hospital use of statins is associated with a reduced risk of mortality among individuals with COVID-19. Cell Metab. 2020;32(2):176–187e4. doi:10.1016/j.cmet.2020.06.015

26. Stekhoven DJ, Buhlmann P. MissForest–non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012;28(1):112–118. doi:10.1093/bioinformatics/btr597

27. Haneuse S, VanderWeele TJ, Arterburn D. Using the E-value to assess the potential effect of unmeasured confounding in observational studies. JAMA. 2019;321(6):602–603. doi:10.1001/jama.2018.21554

28. Jager KJ, van Dijk PC, Zoccali C, et al. The analysis of survival data: the Kaplan-Meier method. Kidney Int. 2008;74(5):560–565. doi:10.1038/ki.2008.217

29. Sachdeva S, Desai R, Gupta U, et al. Admission hyperglycemia in non-diabetics predicts mortality and disease severity in covid-19: a pooled analysis and meta-summary of literature. SN Compr Clin Med. 2020;1–6:2161.

30. Singh AK, Singh R. Hyperglycemia without diabetes and new-onset diabetes are both associated with poorer outcomes in COVID-19. Diabetes Res Clin Pract. 2020;167:108382. doi:10.1016/j.diabres.2020.108382

31. Gentile S, Strollo F, Mambro A, Ceriello A. COVID-19, ketoacidosis and new-onset diabetes: are there possible cause and effect relationships among them? Diabetes Obes Metab. 2020;22(12):2507–2508.

32. Sathish T, Tapp RJ, Cooper ME, et al. Potential metabolic and inflammatory pathways between COVID-19 and new-onset diabetes. Diabetes Metab. 2021;47(2):101204. doi:10.1016/j.diabet.2020.10.002

33. Song S, Zhang S, Wang Z, et al. Association between longitudinal change in abnormal fasting blood glucose levels and outcome of COVID-19 patients without previous diagnosis of diabetes. Front Endocrinol (Lausanne). 2021;12:640529. doi:10.3389/fendo.2021.640529

34. Cai J, Li H, Zhang C, et al. The neutrophil-to-lymphocyte ratio determines clinical efficacy of corticosteroid therapy in patients with COVID-19. Cell Metab. 2021;33(2):258–269e3. doi:10.1016/j.cmet.2021.01.002

35. Schuetz P, Friedli N, Grolimund E, et al. Effect of hyperglycaemia on inflammatory and stress responses and clinical outcome of pneumonia in non-critical-care inpatients: results from an observational cohort study. Diabetologia. 2014;57(2):275–284. doi:10.1007/s00125-013-3112-9

36. MacIntyre EJ, Majumdar SR, Gamble J-M, et al. Stress hyperglycemia and newly diagnosed diabetes in 2124 patients hospitalized with pneumonia. Am J Med. 2012;125(10):1036e17–23. doi:10.1016/j.amjmed.2012.01.026

37. Davis TM, Coleman RL, Holman RR, et al. Prognostic significance of silent myocardial infarction in newly diagnosed type 2 diabetes mellitus: United Kingdom Prospective Diabetes Study (UKPDS) 79. Circulation. 2013;127(9):980–987. doi:10.1161/CIRCULATIONAHA.112.000908

38. Yang JK, Lin -S-S, Ji X-J, et al. Binding of SARS coronavirus to its receptor damages islets and causes acute diabetes. Acta Diabetol. 2010;47(3):193–199. doi:10.1007/s00592-009-0109-4

39. Rubino F, Amiel SA, Zimmet P, et al. New-onset diabetes in Covid-19. N Engl J Med. 2020;383(8):789–790. doi:10.1056/NEJMc2018688

40. Montefusco L, Ben Nasr M, D’Addio F, et al. Acute and long-term disruption of glycometabolic control after SARS-CoV-2 infection. Nat Metab. 2021;3(6):774–785. doi:10.1038/s42255-021-00407-6

41. Singh AK, Khunti K. Assessment of risk, severity, mortality, glycemic control and antidiabetic agents in patients with diabetes and COVID-19: a narrative review. Diabetes Res Clin Pract. 2020;165:108266. doi:10.1016/j.diabres.2020.108266

42. Zhou J-H, Wu B, Wang W-X, et al. No significant association between dipeptidyl peptidase-4 inhibitors and adverse outcomes of COVID-19. World J Clin Cases. 2020;8(22):5576–5588. doi:10.12998/wjcc.v8.i22.5576

43. Ling P, Luo S, Zheng X, et al. Elevated fasting blood glucose within the first week of hospitalization was associated with progression to severe illness of COVID −19 in patients with preexisting diabetes: a multicenter observational study. J Diabetes. 2021;13(1):89–93. doi:10.1111/1753-0407.13121

44. Alahmad B, Al-Shammari AA, Bennakhi A, et al. Fasting blood glucose and COVID-19 severity: nonlinearity matters. Diabetes Care. 2020;43(12):3113–3116. doi:10.2337/dc20-1941

45. Yu B, Li C, Sun Y, et al. Insulin treatment is associated with increased mortality in patients with COVID-19 and Type 2 diabetes. Cell Metab. 2021;33(1):65–77e2. doi:10.1016/j.cmet.2020.11.014

46. Cheng X, Liu Y-M, Li H, et al. Metformin is associated with higher incidence of acidosis, but not mortality, in individuals with COVID-19 and pre-existing Type 2 diabetes. Cell Metab. 2020;32(4):537–547e3. doi:10.1016/j.cmet.2020.08.013

Creative Commons License © 2021 The Author(s). This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution - Non Commercial (unported, v3.0) License. By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms.