1. INTRODUCTION
Dexamethasone significantly reduces mortality 1 and is now standard treatment for patients with COVID‐19 who require supplemental oxygen and/or mechanical ventilation. However, supraphysiological doses of glucocorticoids may exacerbate dysglycaemia and precipitate hyperglycaemic complications, particularly in those with or at risk of type 2 diabetes. 2 The RECOVERY trial 1 reported a low incidence of hyperglycaemic complications (2/1996, 0.1%), although the real‐world incidence is likely to be much higher. 3 Type 2 diabetes itself increases the risk of severe COVID‐19, 4 and hyperglycaemia independently predicts poor outcomes. 5 We investigated the possibility that patients with diabetes may derive less survival benefit from steroid therapy in the setting of severe COVID‐19 infection.
2. METHODS
We performed a retrospective analysis of the characteristics of all nonpregnant adults hospitalized with COVID‐19 infection between 9 March 2020 and 22 April 2020 (UK COVID‐19 Wave 1) 6 and between 1 November 2020 and 31 January 2021 (UK COVID‐19 Wave 2) in three hospitals within Imperial College Healthcare NHS Trust (ICHNT), London. Clinical factors associated with a composite outcome of death and/or intensive care unit (ICU) admission, or mortality alone, within 30 days of COVID‐19 diagnosis were assessed in all patients in Wave 1 and Wave 2 combined.
In Wave 1, dexamethasone was not used as a treatment for COVID‐19 infection in ICHNT. By Wave 2, ICHNT inpatient COVID‐19 treatment guidelines included 6 mg/d dexamethasone as standard of care for all patients admitted with COVID‐19 infection that required supplemental oxygen, regardless of their diabetes status. We also recorded dexamethasone‐induced hyperglycaemic complications in Wave 2, as well as length of stay in hospital in Wave 1 and Wave 2.
GraphPad Prism 9.0 statistical software was used to perform Mann‐Whitney tests on nonparametric data, and Fisher's exact tests were used for univariate analyses. Matlab code and STATA software were used to perform unbiased multivariate logistic regression analysis to determine factors associated with death and/or ICU admission or 30‐day mortality alone. The multivariate analysis was performed on all 2261 patients studied (Wave 16 plus Wave 2) and we report both the logit coefficients and associated odds ratios. In brief, this analysis aims to identify and quantify the contribution of both preexisting conditions and presenting clinical features that are independently (eg, all other measured features being equal) associated with death from COVID‐19.
One of the measures that were chosen a priori was FiO2, that is, the maximum oxygen requirement during admission—a direct measure of disease severity. Rather than arbitrarily remove this factor from this analysis, we elected to control for the potential endogeneity 7 of FiO2 (which would otherwise invalidate the statistical analysis), we ran a first stage regression of FiO2 on a set of exogenous regressors (clinical frailty score, admission temperature, respiratory rate, heart rate and systolic blood pressure), which showed a good fit (R 2 of 30%). Thereafter the logit specifications were run in STATA, adding the residuals from the first stage regressions. Both regressions displayed a good fit, with adjusted R 2 equal to 36% for the composite primary endpoint of death and/or ICU admission and 31% for 30‐day mortality alone. This increased our confidence that our findings were unbiased.
3. RESULTS
During Wave 2, 1372 adults were admitted to ICHNT hospitals with COVID‐19 compared with 889 admissions in Wave 1. 6 There was no improvement in the composite endpoint of death/ICU admission between Wave 1 and Wave 2 (Wave 1: 34.7% vs. Wave 2: 33.6%; P = 0.60). However, there was a significant reduction in 30‐day mortality alone in Wave 2, both for the entire cohort (Wave 1: 27.5% vs. Wave 2: 17.2%; P < 0.001, odds ratio [OR] 0.55 [95%CI 0.45‐0.67]) and for patients with diabetes (Wave 1: 36.1% vs. Wave 2: 22.3%; P < 0.001, OR 0.51 [95%CI 0.37‐0.70]).
Demographic and clinical characteristics of those admitted across both waves were similar. Duration of hospital admission was longer in Wave 2 (median [interquartile range] Wave 2: 11 [6‐20] days vs. Wave 1: 8 [4‐16] days; P < 0.0001). Diabetes was a recorded comorbidity for 337/889 patients (38%) in Wave 16 and for 456/1372 patients (33%) in Wave 2, with the majority having type 2 diabetes (Wave 1: 96% vs. Wave 2: 98%). The 2020/2021 prevalence of diabetes amongst adults in our local area was 3.5%. 8 Supporting Information Table S1 details the univariate analysis of clinical and demographic features of patients with diabetes associated with ICU admission/death within 30 days of COVID‐19 diagnosis for Wave 2, which are very similar to our previously published findings for patients with diabetes in Wave 1. 6
Similar to Wave 1, 6 univariate analysis of all patients in Wave 2 demonstrated that male gender (OR 1.56 [95%CI 1.29‐1.90]; P < 0.001), hypertension (OR 1.54 [95%CI 1.28‐1.87]; P < 0.001), heart failure (OR 2.65 [95%CI 1.83‐3.87]; P < 0.001), increased frailty (Clinical Frailty Score ≥7: OR 2.43 [95%CI 1.72‐3.44]; P < 0.001), impaired renal function (admission estimated glomerular filtration rate [eGFR] <60 mL/kg/min: OR 2.39 [95%CI 1.95‐2.91]; P < 0.001) and hyperglycaemia (admission capillary blood glucose ≥10 nmol/L [180 mg/dL]: OR 1.50 [95%CI 1.07‐2.07]; P = 0.02) were associated with an increased risk of death/ICU admission. Unlike in Wave 1, 6 in Wave 2, Black patients had a higher risk of ICU admission/death compared with White patients (OR 1.52 [95%CI 1.06‐2.17]; P = 0.03).
Multivariate analysis of clinical factors associated with the composite risk of death/ICU admission for all 2261 patients are shown in Table 1. Ischaemic heart disease and renal failure were independent predictors of poor outcome, over and above a diagnosis of diabetes per se, suggesting that multimorbidity rather than any one risk factor alone is the major driver for COVID‐19 mortality. 9 For the composite outcome of ICU admission/death (Table 1) as well as the single outcome of mortality alone (Table 2), the contributions of the major clinical risk factors to the odds of severe disease were similar to those which have been reported. We present the logit coefficient for the independent effect of dexamethasone usage for patients with and without diabetes, which allows us to conclude that the beneficial effect is not different between these two groups (all other cofactors being equal). Dexamethasone use was associated with a lower probability of ICU admission/death (Table 1), and its effect was very similar in patients with and without diabetes (P = 0.82 using an F‐test for equality of the logit coefficients). However, dexamethasone use was not associated with a reduction in mortality in patients with diabetes and in patients without diabetes (Table 2).
TABLE 1.
Multivariate analysis of clinical factors independently associated with the primary outcome of death or intensive care unit admission in patients hospitalized due to COVID‐19
| Variable | Death or ICU admission within 30 days | |||
|---|---|---|---|---|
| Logit coefficient | SE | OR | SE | |
| Wave 1 | −0.484 | 0.281 | 0.616 | 0.18 |
| Age | 0.183 ** | 0.00063 | 1.018 ** | 0.00662 |
| Weight (kg) | −0.00937 * | 0.00419 | 0.991 * | 0.00427 |
| Female | −0.335 * | 0.155 | 0.715 * | 0.114 |
| White | 0.194 | 0.157 | 1.02 | 0.171 |
| Diabetes | −0.289 | 0.229 | 0.749 | 0.174 |
| Stroke | −0.0506 | 0.224 | 0.951 | 0.216 |
| Heart disease (IHD +/− CCF) | 0.413 * | 0.178 | 1.511 * | 0.289 |
| Hypertension | −0.0461 | 0.163 | 0.955 | 0.159 |
| Admission eGFR | −0.0157 ** | 0.00303 | 0.984 ** | 0.0031 |
| FiO 2 (maximum during admission) | 0.0711 ** | 0.00927 | 1.074 ** | 0.0104 |
| Dexamethasone usage—patients without diabetes | −0.854 * | 0.349 | 0.426 * | 0.149 |
| Dexamethasone usage—patients with diabetes | −0.786 * | 0.357 | 0.456 * | 0.174 |
| Remdesivir | −0.0995 | 0.201 | 0.905 | 0.186 |
| Tociluzimab | −0.12 | 0.453 | 0.887 | 0.39 |
| Residual FiO2 | −0.00951 | 0.00942 | ||
| Constant | −3.254** | 0.723 | ||
Note: Unselected multivariate logistic (Logit) analysis of clinical variables that were collected for patients admitted with swab positive COVID‐19 in Wave 1 and Wave 2 combined, as applied to the composite outcome of death and/or ICU admission within 30 days. Regressors were included if accurate data was available for >80% of the combined pool of 2261 patients. For categorical variables, a positive “estimate” indicates an increased risk of the composite outcome (death or ICU admission) with that variable present, and a negative estimate indicates a reduced risk of the composite outcome if that variable is present. For continuous variables, a positive “estimate” indicates an increased risk of the composite outcome with that variable increasing, and a negative estimate indicates a reduced risk of the composite outcome if that variable is decreasing. The P value is a measure of the confidence of that variable being an independent predictor of the composite outcome corrected for all of the other regressors listed. Bold values are used to highlight variables with significant P‐values.
Abbreviations: CCF, congestive cardiac failure; eGFR, estimated glomerular filtration rate; FiO2, fraction of inspired oxygen; IHD, ischaemic heart disease; ICU, intensive care unit; OR, odds ratio.
P <0.05;
P <0.01 for the independent logit coefficient or OR.
TABLE 2.
Multivariate analysis of clinical factors independently associated with the death in patients hospitalized due to COVID‐19
| Variable | Death within 30 days | |||
|---|---|---|---|---|
| Logit coefficient | SE | OR | SE | |
| Wave 1 | 0.372 | 0.305 | 1.451 | 0.416 |
| Age | 0.0601 ** | 0.00802 | 1.062 ** | 0.00825 |
| Weight (kg) | −0.000987 | 0.00463 | 0.999 | 0.00451 |
| Female | −0.325 | 0.167 | 0.723 | 0.124 |
| White | 0.104 | 0.169 | 1.11 | 0.19 |
| Diabetes | −0.0793 | 0.231 | 0.924 | 0.191 |
| Stroke | 0.00737 | 0.223 | 1.007 | 0.2 |
| Heart disease (IHD +/− CCF) | 0.486 ** | 0.185 | 1.625 ** | 0.284 |
| Hypertension | 0.149 | 0.179 | 1.161 | 0.213 |
| Admission eGFR | −0.0176 ** | 0.00326 | 0.983 ** | 0.00316 |
| FiO 2 (maximum during admission) | 0.0329 ** | 0.0107 | 1.033 ** | 0.0115 |
| Dexamethasone usage—patients without diabetes | −0.184 | 0.378 | 0.832 | 0.311 |
| Dexamethasone usage—patients with diabetes | 0.0323 | 0.419 | 1.033 | 0.405 |
| Remdesivir | −0.397 | 0.254 | 0.673 | 0.164 |
| Tociluzimab | −1.22 * | 0.609 | 0.295 * | 0.173 |
| Residual FiO2 | 0.0147 | 0.0113 | ||
| Constant | −6.46** | 0.932 | ||
Note: Unselected multivariate logistic (Logit) analysis of clinical variables that were collected for patients admitted with swab positive COVID‐19 in Wave 1 and Wave 2 combined, as applied to the outcome of death within 30 days. Regressors were included if accurate data was available for >80% of the combined pool of 2261 patients. For categorical variables, a positive “estimate” indicates an increased risk of the outcome (death) with that variable present, and a negative estimate indicates a reduced risk of the outcome (death) if that variable is present. For continuous variables, a positive “estimate” indicates an increased risk of the outcome (death) with that variable increasing, and a negative estimate indicates a reduced risk of the outcome (death) if that variable is decreasing. The P value is a measure of the confidence of that variable being an independent predictor of the primary outcome corrected for all of the other regressors listed. Bold values are used to highlight variables with significant P‐values.
Abbreviations: CCF, congestive cardiac failure; eGFR, estimated glomerular filtration rate; FiO2, fraction of inspired oxygen; IHD, ischaemic heart disease; OR, odds ratio.
P <0.05;
P <0.01 for the independent logit coefficient or OR.
Dexamethasone was not used as a treatment for COVID‐19 during Wave 1. In Wave 2, dexamethasone was used in 68.2% of patients (n = 935), 35% of whom had diabetes (indicating no bias against patients with diabetes for the use of dexamethasone). Only 21 patients who were eligible for dexamethasone treatment (ie, required supplemental oxygen) did not receive dexamethasone: four received alternative corticosteroid therapy (prednisolone and/or methylprednisolone), five had a transient oxygen requirement, one self‐discharged and 11 rapidly deteriorated and received palliative care. Amongst patients who received dexamethasone, there was no difference in median length of stay between patients with diabetes and patients without diabetes (with diabetes 11 [interquartile range 6‐19] vs. without diabetes 10 [interquartile range 6‐18]; P = 0.18).
The doses of dexamethasone used are shown in Supporting Information Table S2. A total of 90% of patients (841/935) received 6 mg/d dexamethasone (ie, the dose used in the RECOVERY trial 1 ). Unsurprisingly, the average capillary blood glucose levels in the 72‐hour period following COVID‐19 diagnosis were higher in patients with diabetes treated with dexamethasone compared to patients with diabetes who were not treated with dexamethasone (12.3 ± 4.4 vs. 10.0 ± 4.3 mmol/L; P < 0.0001).
Hyperglycaemia meeting the threshold for additional pharmacotherapy (ie, ≥12 mmol/L) occurred in 19% of patients (178/935) who received dexamethasone (21 patients were treated with oral antihyperglycaemic agents, 122 were treated with insulin alone, 33 were treated with both insulin and oral antihyperglycaemic agents and two patients did not receive antihyperglycaemic therapy). Of all patients admitted in Wave 2, 19 (ie, 1.4% [19/1372], 11% [19/178] of all patients with dexamethasone‐induced hyperglycaemia) were classified as having dexamethasone‐induced diabetes, based on no prior diagnosis of diabetes at admission and the requirement for new antihyperglycaemic medication at discharge. Furthermore, in dexamethasone‐treated patients, 1.4% (13/935) developed diabetic ketoacidosis and 2.7% (25/935) developed hyperglycaemic hyperosmolar syndrome. Of the patients who developed dexamethasone‐induced hyperglycaemia, neither preexisting diabetes nor type of glucose‐lowering treatment received were associated with increased risk of ICU admission/death on univariate analysis (Supporting Information Table S3).
4. DISCUSSION
Mortality within 30 days of COVID‐19 diagnosis was significantly lower in Wave 2, with dexamethasone usage (but not remdesivir) independently associated with a reduced risk of ICU admission and/or death. By 10 January 2021 (in London), vaccine coverage was low (1st dose received: 29.5% of people aged ≥80 years and 1.1% of people aged <80 years; 2nd dose received: 9.1% of people aged ≥80 years and 0.1% of people aged <80 years 10 ); thus, it is unlikely that vaccination would have had a major effect on the reduced mortality seen in Wave 2 in this study.
Multivariate analysis revealed that the independent effect size of dexamethasone on reducing COVID‐19 severity was similar for patients with and without diabetes. This suggests that the benefits of dexamethasone in people with diabetes remain substantial despite the occurrence of dexamethasone‐related hyperglycaemic complications in a fifth of patients. Dexamethasone should continue to be used with confidence in patients with diabetes and severe COVID‐19, but treatment guidelines should incorporate strategies for identification and management of steroid‐induced hyperglycaemia. 2 Service providers must also be aware of the extra resource required to manage glycaemic complications whilst treating inpatients and also post discharge. 3
CONFLICT OF INTEREST
All authors declare no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conceptualization: Victoria Salem and Tricia M. Tan. Investigation: all authors. Analysis: Walter Distaso, Pei Chia Eng, Chioma Izzi‐Engbeaya and Victoria Salem Data curation: Pei Chia Eng, Chioma Izzi‐Engbeaya. and Victoria Salem. Writing – original draft: Pei Chia Eng, Chioma Izzi‐Engbeaya and Victoria Salem. Writing – review and editing: all authors. Victoria Salem is the guarantor of this work, had full access to the data, and accepts full responsibility for integrity of the data and the accuracy of the data analysis.
PEER REVIEW
The peer review history for this article is available at https://publons.com/publon/10.1111/dom.14692.
Supporting information
Supplementary Table S1
Eng PC, Distaso W, Durreshahwar H, et al. The benefit of dexamethasone in patients with COVID‐19 infection is preserved in patients with diabetes. Diabetes Obes Metab. 2022;24(7):1385‐1389. doi: 10.1111/dom.14692
Chioma Izzi‐Engbeaya and Victoria Salem are joint senior authors.
Funding information The study was not externally funded. C.I.‐E. is funded by an Imperial‐Biomedical Research Centre (BRC) IPPRF Fellowship (P79696). V.S. is the recipient of a Diabetes UK Harry Keen Clinician Scientist Fellowship (15/0005317). The Department of Metabolism, Digestion and Reproduction is funded by grants from the Medical Research Council and the National Institutes of Health Research (NIHR), and is supported by the NIHR Imperial BRC Funding Scheme and the NIHR/Imperial Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the abovementioned funders, the UK National Health Service (NHS), the NIHR, or the UK Department of Health.
DATA AVAILABILITY STATEMENT
Full access to primary data and statistical analyses are available upon request to the corresponding author.
REFERENCES
- 1. Horby P, Lim W, Emberson J, et al. Dexamethasone in hospitalized patients with Covid‐19. N Engl J Med. 2021;8:693‐704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Rayman G, Lumb AN, Kennon B, et al. Dexamethasone therapy in COVID‐19 patients: implications and guidance for the management of blood glucose in people with and without diabetes. Diabet Med. 2021;38:e14378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Younes YR, Stockley S, Keegan L, et al. COVID‐19 and dexamethasone‐induced hyperglycaemia: workload implications for diabetes inpatient teams. Diabetes Med. 2022;39(2):e14716. doi: 10.1111/dme.14716 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Schlesinger S, Neuenschwander M, Lang A, Pafili K, Kuss O, Herder CRM. Risk phenotypes of diabetes and association with COVID‐19 severity and death: a living systematic review and meta‐analysis. Diabetologia. 2021;64(7):1480‐1491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Zhu L, She ZG, Cheng X, et al. LH. Association of Blood Glucose Control and Outcomes in patients with COVID‐19 and pre‐existing type 2 diabetes. Cell Metab. 2020;2(31):1068‐1077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Izzi‐Engbeaya C, Distaso W, Amin A, et al. Adverse outcomes in COVID‐19 and diabetes: a retrospective cohort study from three London teaching hospitals. BMJ Open Diabetes Res Care. 2021;9(1):e001858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Moran JL, Santamaria JD, Duke GJ, The Australian & New Zealand Intensive Care Society (ANZICS) Centre For Outcomes & Resource Evaluation (CORE) . Modelling hospital outcomes: problems with endogeneity. BMC Med Res Methodol. 2021;21(1):124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Diabetes. Public Health England . (2021). https://fingertips.phe.org.uk/profile/diabetes-ft/data. Accessed December 14, 2021.
- 9. Clark A, Jit M, Warren‐Gash C, et al. Global, regional, and national estimates of the population at increased risk of severe COVID‐19 due to underlying health conditions in 2020: a modelling study. Lancet Glob Heal 2020. 2020;8(8):e1003‐e1017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. NHS England. Covid‐19 montly announced vaccinations . https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-vaccinations/covid-19-vaccinations-archive. Accessed January 14, 2021.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Table S1
Data Availability Statement
Full access to primary data and statistical analyses are available upon request to the corresponding author.