Hyperglycemia in hospitalized patients with coronavirus disease 2019 (COVID-19) is linked to increased morbidity and mortality. This article reports on a novel insulin titration protocol for the management of glucocorticoid-induced hyperglycemia in hospitalized patients with COVID-19. Sixty-five patients with COVID-19 and glucocorticoid-induced hyperglycemia admitted after the protocol implementation were matched 1:1 to patients admitted before the treatment protocol rollout for analysis. In a large, diverse health system, the protocol achieved reductions in hypoglycemic events without increasing hyperglycemia or insulin use.

Patients with preexisting chronic conditions are at a higher risk of having a severe course of coronavirus disease 2019 (COVID-19) (1,2). Diabetes and hyperglycemia are important risk factors for disease severity, hospitalization, length of stay, intensive care unit (ICU) admissions, morbidity, and mortality (25). The proinflammatory state caused by COVID-19 may impair insulin secretion, increase insulin resistance, reduce peripheral glucose uptake, impair glucose metabolism, and cause β-cell dysfunction (68). Simultaneously, hyperglycemia can stimulate viral replication and augment cytokine expression, making these patients more vulnerable to severe disease (9).

Glucocorticoids (GCs) are essential in the management of moderate to severe COVID-19 because of their potent anti-inflammatory effects on immune system over- activation (10,11). The RECOVERY (Randomized Evaluation of COVID-19 Therapy) trial showed that 6 mg of daily dexamethasone resulted in reduced 28-day mortality and shorter hospitalizations in patients receiving mechanical ventilation or supplemental oxygen (12). Systemic use of GCs can induce hyperglycemia in patients with and without preexisting diabetes as a result of increased hepatic gluconeogenesis, reduced peripheral insulin sensitivity, and direct inhibition of insulin release from pancreatic β-cells (13). These effects can be seen 24–48 hours after administration (10,14). Thus, the deleterious glycemic effects of COVID-19 are further complicated by GC therapy.

Optimal treatment protocols for GC hyperglycemia have not been established. Insulin is the preferred therapy for patients with COVID-19 requiring hospitalization and is typically initiated for persistent hyperglycemia (≥180 mg/dL [10 mmol/L]) aiming for a target glucose range of 140–180 mg/dL (7.7–10 mmol/L) (15,16). Klonoff et al. (4) reported that glucose values between 141 and 180 mg/dL (7.8 and 10 mmol/L) were associated with reduced mortality in non-ICU COVID-19 hospitalizations, and Zhu et al. (2) described better clinical outcomes with euglycemic glucose values of 70–180 mg/dL (3.8–10 mmol/L) compared with hyperglycemia >180 mg/dL (10 mmol/L). Both studies highlight the relationship between glycemic control and clinical outcomes. Rigid glucose control poses a risk of hypoglycemia (<70 mg/dL [3.8 mmol/L]), which can increase mortality (17). Because of these risks and benefits, protocols for insulin dosing in GC-induced hyperglycemia may be helpful to achieve early inpatient glycemic control while minimizing hypoglycemic episodes, reducing in-hospital mortality, and reducing the need for endocrinology consultation (18). Here, we propose and evaluate the effectiveness of an insulin protocol to guide medical teams to achieve inpatient glycemic control in patients with COVID-19 treated with GCs.

Study Design

The study was carried out at the Mount Sinai Health System in New York City and included six hospitals treating COVID-19 patients. We developed a treatment protocol to aid medical teams in the management of GC-induced hyperglycemia for adult inpatients with COVID-19 as a quality improvement (QI) project. The retrospective comparator data were approved by the Mount Sinai Hospital Institutional Review Board, and the protocol was approved by the hospital’s Quality Improvement Committee. Members from the Division of Endocrinology, Diabetes, and Bone Diseases and the Division of Hospital Medicine developed treatment guidance documents based on American Diabetes Association’s standards of care for hospitalized patients with hyperglycemia and expert opinion. Endocrinology division members performed outreach to health care providers (HCPs) via e-mail and through electronic medical record (EMR) messaging from March 2021 through January 2022. Patients with COVID-19 treated with GC with hyperglycemia, defined as ≥2 fingerstick glucose (FSG) values >180 mg/dL (10 mmol/L), and admitted to the health system during this period were included and compared with historical control subjects admitted 6 months before the protocol implementation.

Protocol Description

The protocol was made available to all frontline HCPs to serve as guidance on how to manage GC-induced hyperglycemia in adult COVID-19 patients in non-ICU settings. Before rollout, frontline HCPs prescribed and titrated insulin orders and consulted the inpatient endocrinology service based on their discretion; protocols for intravenous insulin were used in the ICU and for treatment of diabetic ketoacidosis (DKA). The new protocol was accessible within the EMR system. To use the protocol, five patient-specific history elements are required: history of diabetes (yes/no), home diabetes medication regimen with insulin total daily dose (TDD; if applicable), A1C, weight, and estimated glomerular filtration rate (eGFR). Based on these parameters, the protocol offers guidance via an algorithm for insulin starting dose and titration suggestions based on glycemic values with an FSG target range of 140–180 mg/dL (7.7–10 mmol/L) and discharge recommendations. The protocol includes insulin starting doses that range from 0.1 to 0.6 units/kg for both basal and bolus regimens and allows adjustment based on oral intake status. Guidance is also provided within the protocol regarding when users should contact the inpatient endocrinology service, including for persistent hyperglycemia (FSG >250 mg/dL [13.8 mmol/L]) despite following the algorithm for 48 hours, refractory hypoglycemia, and development of DKA. The protocol and algorithm pathways are provided in the Supplementary Material. General medicine teams who accessed the protocol were composed of attending hospitalists, residents, nurse practitioners, and/or physician assistants.

Study Groups

Patients were included in analysis if they were adults ≥18 years of age, admitted to a medicine team at one of the six hospitals within the health system, and not actively being managed by the inpatient endocrinology service. Patients were excluded if they transferred to the ICU after spending <48 hours on medicine floors, required total parenteral nutrition or tube feeding, were followed by the inpatient endocrinology service after <48 hours on medicine floors, required an insulin drip or were using insulin pump therapy, or did not have an A1C drawn during admission or receive insulin during their hospitalization.

Outcomes and Covariates

Baseline clinical and demographic characteristics such as age, sex, weight, insurance status, race, ethnicity, obesity, BMI, admission creatinine, home medication, diabetes history, most recent A1C, and comorbid conditions such as asthma, cancer, chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), coronary artery disease (CAD), heart failure (HF), and hypertension (HTN) were recorded. In-hospital glycemic and outcome parameters included peak insulin dose reported as TDD in units, FSG, length of stay (LOS), discharge on insulin, and disposition. The highest insulin TDD was used as a surrogate for total admission insulin requirements. Hypoglycemia was defined as FSG <70 mg/dL (3.8 mmol/L), euglycemia as FSG 70–180 mg/dL (3.8–10 mmol/L), mild hyperglycemia as FSG 180–250 mg/dL (10–13.8 mmol/L), and severe hyperglycemia as FSG >250 mg/dL (13.8 mmol/L). To account for the difference in quantity of total hospitalization FSG values between groups and among individual patients, proportions of FSG values per patient were calculated, and descriptive statistics of the proportions are reported.

Statistical Analysis

Two-tailed significance testing was performed with α = 0.05. G*Power software was used to estimate a minimum sample size of 64 patients per group for a β = 80% with a small-to-medium effect size for mean hospitalization FSG (19). Patients were matched 1:1 on sex, weight ± 8 kg, and then age ± 3 years. All data were analyzed in SPSS, v. 28.0.1.0, statistical software (20). Variable distributions were examined for normality using the Kolmogorov-Smirnov test and visual examination of histograms. Age was the only normally distributed continuous variable across groups. Analysis of baseline characteristics and differences in glycemic metrics and clinical outcomes between the protocol and historical comparator groups were compared using the Student t test for continuous normally distributed variables. The Wilcoxon rank sum test was used for continuous nonparametric data, and χ2 testing was used for categorical variables. Multivariable linear regression with backward stepwise selection was used to examine the association between glycemic control and baseline characteristics. A nonadjusted P value <0.05 was regarded as statistically significant.

Population Characteristics

This study included 130 patients with COVID-19 and GC-induced hyperglycemia; 65 patients admitted after the protocol implementation were matched 1:1 to patients admitted before the rollout. Compared with the historical comparator group, the protocol group had a significantly lower median admission creatinine (0.9 mg/dL [95% CI 0.8–1.1 mg/dL] vs. 1.2 mg/dL [95% CI 1.0–1.4 mg/dL], P = 0.03), less use of insulin at home (12 [18.5%] vs. 22 [33.8%], P = 0.05), more adjuvant COVID-19 therapy with remdesivir (8 [12.3%] vs. 1 [1.5%], P = 0.03), more African American patients (26 [40.0%] vs. 13 [20.0%], P = 0.02), and fewer White patients (7 [10.8%] vs. 21 [32.3%], P = 0.01) (Table 1). There were no significant differences between groups in BMI, median A1C, diabetes history, home diabetes medication, asthma, cancer, CKD, COPD, CAD, HF, HTN, obesity, steroid use, adjuvant therapy, or insurance status (all P ≥0.05).

Glycemic and Clinical Outcomes

There was no significant difference in overall mean hospitalization glucose in the protocol group compared with a historical comparator (205 mg/dL [11.4 mmol/L], 95% CI 203–208 mg/dL [11.3–11.5 mmol/L], vs. 201 mg/dL [11.2 mmol/L], 95% CI 198–204 mg/dL [11.0–11.3 mmol/L], P = 0.17) (Table 2). There were significantly more patients with mean hospitalization FSG in the euglycemic range in the protocol group (25 [38.5%] vs. 11 [16.9%], P = 0.006) (Figure 1), and there was a nonsignificant decrease in patients with mean glucose in the mild hyperglycemic range (33 [50.8%] vs. 23 [35.4%], P = 0.076). There were fewer FSG checks performed in the protocol group (3,744 vs. 4,804). The protocol group had a significantly lower proportion of hypoglycemic FSG values per patient (0.003 [95% CI 0.0005–0.006] vs. 0.008 [95% CI 0.003–0.012], P = 0.044). Patients in the protocol group had a higher proportion of euglycemic FSG values per patient, but this increase did not reach significance (0.413 [95% CI 0.342–0.484] vs. 0.350 [95% CI 0.283–0.415], P = 0.239). Similarly, the proportion of mild hyperglycemic FSG values per patient trended lower in the protocol group (0.272 [95% CI 0.234–0.311] vs. 0.315 [95% CI 0.268–0.361], P = 0.291). There were no significant differences in proportion of severe hyperglycemic FSG values per patient (0.328 [95% CI 0.261–0.395] vs. 0.311 [95% CI 0.242–0.380], P = 0.828).

Median LOS was not different between groups (10 days [95% CI 8–12 days] vs. 9 days [95% CO 7–12 days], P = 0.41). The protocol group had a significantly higher utilization of basal-bolus insulin therapy (63 [93.8%] vs. 43 [66.2%], P <0.0001). Protocol patients had a nonsignificant higher percentage of their peak-day TDD coming from bolus versus basal insulin (70.2% [95% CI 64.1–77.3%] vs. 62.4% [95% CI 53.3–70.1%], P = 0.08). Peak-day insulin administration was nonsignificantly lower in the protocol group (0.25 units/kg [95% CI 0.17–0.36 units/kg] vs. 0.32 units/kg [95% CI 0.23–0.55 units/kg], P = 0.36), and a higher number of patients in the protocol group were discharged on insulin (21 [32.3%] vs. 16 [24.6%], P = 0.13). The mortality rate was not significantly lower in the protocol group (12 [18.5%] vs. 17 [26.2%], P = 0.29).

Multivariate Regression Analysis

Table 3 shows the results of the multivariate linear regressions examining the relationship between the use of the protocol, demographics, and risk factors (including diabetes diagnosis and adjuvant therapy) on having an individual euglycemic FSG and individual severely hyperglycemic FSG, respectively. There was a significant negative association between home insulin use and the proportion of euglycemic FSG (−0.127 [95% CI −0.239 to −0.014], P = 0.027). There were no significant associations with the proportion of severely hyperglycemic FSG. Protocol rollout was not significantly associated with the proportion of euglycemic (0.059 [95% CI −0.049 to 0.166], P = 0.28) or severely hyperglycemic FSG (−0.019 [95% CI −0.127 to 0.90], P = 0.73).

Our study evaluated the effectiveness and safety of a user-friendly and readily accessible insulin protocol in achieving glycemic control in GC-induced hyperglycemia for COVID-19 inpatients. This protocol did not require an endocrinology consultation and only required five inputs: the patient’s diabetes history, home diabetes medication regimen, A1C, weight, and eGFR. The protocol group had significantly more patients who achieved mean euglycemic hospitalization glucose values compared with the historical comparator group and had a significantly lower proportion of FSG levels in the hypoglycemic range. These clinical results suggest that our protocol was superior in reducing hypoglycemic events, effective in improving euglycemic control, and noninferior in reducing hyperglycemia and mortality.

Our patient populations were matched for sex, weight, and age and had similar rates of comorbidities. The populations differed in terms of race, home insulin use, and admission creatinine levels. Differences in race may be attributed to the predominance of certain racial groups at specific hospitals and disparities in vaccine access (vaccine history was not available) or hesitancy.

To adjust for confounders in our patient population, we performed a multivariate linear regression analysis. It found that home insulin use, which was greater in the historical comparator group (n = 22 vs. n = 12, P = 0.05), was associated with a significantly lower probability of having a euglycemic FSG value (−0.127 [95% CI −0.239 to −0.014], P = 0.027). Radhakutty et al. (21) reported similar findings for non-COVID hospitalized patients with GC-induced hyperglycemia. Prior insulin users spent more time outside of the target range of 70–180 mg/dL (3.8–10 mmol/L) (68.3 ± 7.2% vs. 39.5 ± 4.1%, P = 0.002) and had a higher mean daily glucose (234 ± 19.8 mg/dL [13.2 ± 1.1 mmol/L] vs. 176.4 ± 9 mg/dL [9.8 ± 0.5 mmol/L], P = 0.004). We hypothesize that this outcome may have stemmed from lower baseline β-cell function in prior insulin users, resulting in augmented glycemic variability and/or increased insulin sensitivity for insulin-naive patients and thus leading to more FSG values in euglycemic range (22). No other clinical risk factors or baseline patient population differences were found to confound our results.

While the protocol’s effect on glycemic metrics did not reach significance in the adjusted model, there was a nonsignificant euglycemic improvement (0.059 [95% CI −0.049 to 0.166], P = 0.281) that was consistent with our unadjusted model. We were unable to run a regression model adjusting for hypoglycemia confounders because of the minimal variability in observed outcomes. However, our unadjusted model did show a significant (P = 0.04) increase in hypoglycemic events in the historical comparator group. The historical comparator group had a larger proportion of home insulin users; thus, we anticipate that the adjusted model would show an even larger increase in likelihood of hypoglycemic events in this group because of the association between diminished β-cell function and worsened glycemic variability, as well as the finding by Radhakutty et al. (21) that prior insulin users spent more time out of the target glycemic range.

Our protocol promoted early initiation of basal-bolus insulin compared with the historical comparator group regardless of home medication regimen. Although not statistically significant, our algorithm showed a reduction in peak daily insulin requirements and an increase in bolus insulin dose (vs. basal dose) compared with the historical comparator arm, suggesting a more appropriate distribution of insulin to address GC-induced hyperglycemia. At our health system, the insulin initiation order set has a default starting doses of 0.2 units/kg for basal insulin and 0.06 units/kg for bolus insulin. In contrast, our protocol included starting insulin doses that ranged from 0.1 to 0.6 units/kg for both basal and bolus insulin, offering a more flexible regimen according to patient-specific parameters. Despite this difference, with some patients starting at higher doses of insulin compared with traditional weight-based dosing, patients in the protocol arm had a lower rate of hypoglycemic events. This result speaks to the safety of our protocol, as hypoglycemia is a feared complication of strict inpatient glycemic control during critical illness because of its association with overall worse outcomes (17).

Scant literature is available about the effect of insulin titration guidance documents and their effect on inpatient glycemic control in COVID-19 patients. To the best of our knowledge, this is one of the first insulin protocol algorithms for COVID-19 patients receiving GC therapy to be used in the United States. Asiri et al. (18) implemented an insulin protocol in Saudi Arabia, with which the experimental group reached well-controlled glucose levels with less hospital mortality compared with the control group (12.93 vs. 29.93%, P = 0.01). They observed a decrease in ICU admissions, intravenous insulin infusion, and need for mechanical ventilation in the experimental group, although these reductions were not statistically significant. Their protocol only used correctional sliding-scale insulin on insulin-naive patients at admission and started basal insulin (0.1 units/kg) by day 2 if the TDD was >10 units. In contrast, our protocol initiated a basal regimen of 0.1–0.2 units/kg if fasting FSG was >140 mg/dL (7.7 mmol/L) for patients without diabetes and 0.2–0.3 units/kg for patients with a history of diabetes regardless of their insulin usage at home. Interestingly, Asiri et al. noted a slightly higher number of patients who developed hypoglycemia in the protocol group versus the control group, although this increase was not statistically significant, despite following a more conservative approach to insulin dosing compared with ours. These findings stress the importance of further studies to address the optimization of insulin protocol algorithms as a management tool for GC-induced hyperglycemia.

The strengths of our protocol include its safety, improvement in euglycemia, EMR-based accessibility, and efficiency in health care resource utilization by empowering front line HCPs to manage GC-induced hyperglycemia without needing an immediate endocrinology consultation. A major limitation of our study is that we could not track the exact use of the protocol, which could have led to an underestimation of its effect, since it was not used for all patients. There also may have been bias in which front line HCPs chose to use the protocol. Another limitation was that our sample size was not large enough to appropriately power smaller effect sizes such as for mortality, LOS, or differences in peak TDD. Additionally, vaccination rates, adjuvant COVID-19 therapies, and institutional COVID-19 treatment protocols evolved and may have confounded our two groups, making the use of a historical control group a potential limitation.

Despite these factors, our protocol equips front line HCPs with a tool to assist with insulin titration for glucose management in these complex patients. Future directions include optimization of our protocol, including assessment of efficacy specifically for the vulnerable subpopulation of home insulin users and modification of their treatment algorithm accordingly. Future studies are also needed with enough power to assess for outcome variables of smaller effect sizes (e.g., mortality and LOS). Our protocol may also be adapted to the general inpatient population and studied in future QI projects.

The achievement of euglycemia is associated with improved overall outcomes in hospitalized patients with moderate-to-severe COVID-19, but despite this, guidance and data are lacking on how to accomplish this task in a timely and safe manner (3,4,10). The association between hyperglycemia, COVID-19, and worse clinical outcomes is well established. Insulin protocols offer a solution to this dilemma and serve as tools to improve glycemic management. Here, we describe a safe and effective novel insulin protocol for the management of GC-induced hyperglycemia in hospitalized COVID-19 patients. Our protocol achieved a reduction in hypoglycemic events without increasing hyperglycemia or insulin utilization and possible improvement in the rate of euglycemia.

Duality of Interest

D.W.L. receives research support from Renalytix AI. C.J.L. receives research support and product support paid to her institution from Abbott Diabetes Care, Dexcom, Insulet, and Tandem Diabetes Care. G.O. receives research support, including salary and product support, from Abbott Diabetes Care, Dexcom, Omnipod, and Tandem Diabetes Care. No other potential conflicts of interest relevant to this article were reported.

Author Contributions

A.B.K. analyzed data and wrote the manuscript. N.V. implemented the protocol and wrote the manuscript. S.J.O. analyzed data. D.B. developed the concept and implemented the protocol. N.A.S., A.S.L., and G.O. developed the concept and reviewed the manuscript. D.W.L. and C.J.L. developed the concept, implemented the protocol, and reviewed the manuscript. G.O. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation

A portion of this work was presented in abstract form at the Endocrine Society’s ENDO 2022 conference, 11–14 June 2022, in Atlanta, GA.

This article contains supplementary material online at https://doi.org/10.2337/figshare.22096607.

1.
Zhou
F
,
Yu
T
,
Du
R
, et al
.
Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study
.
Lancet
2020
;
395
:
1054
1062
2.
Zhu
L
,
She
ZG
,
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
:
1068
1077.e3
3.
Bode
B
,
Garrett
V
,
Messler
J
, et al
.
Glycemic characteristics and clinical outcomes of COVID-19 patients hospitalized in the United States
.
J Diabetes Sci Technol
2020
;
14
:
813
821
4.
Klonoff
DC
,
Messler
JC
,
Umpierrez
GE
, et al
.
Association between achieving inpatient glycemic control and clinical outcomes in hospitalized patients with COVID-19: a multicenter, retrospective hospital-based analysis
.
Diabetes Care
2021
;
44
:
578
585
5.
Gianchandani
R
,
Esfandiari
NH
,
Ang
L
, et al
.
Managing hyperglycemia in the COVID-19 inflammatory storm
.
Diabetes
2020
;
69
:
2048
2053
6.
Accili
D
.
Can COVID-19 cause diabetes?
Nat Metab
2021
;
3
:
123
125
7.
Lim
S
,
Bae
JH
,
Kwon
H-S
,
Nauck
MA
.
COVID-19 and diabetes mellitus: from pathophysiology to clinical management
.
Nat Rev Endocrinol
2021
;
17
:
11
30
8.
Erener
S
.
Diabetes, infection risk and COVID-19
.
Mol Metab
2020
;
39
:
101044
9.
Codo
AC
,
Davanzo
GG
,
Monteiro
LB
, et al
.
Elevated glucose levels favor SARS-CoV-2 infection and monocyte response through a HIF-1α/glycolysis-dependent axis
.
Cell Metab
2020
;
32
:
437
446.e5
10.
Brooks
D
,
Levy
CJ
.
Overview and management of glucocorticoid-induced hyperglycemia in pulmonary diseases: insight into the COVID-19 pandemic
.
International Journal of Diabetes & Metabolic Syndrome
2021
;
1
:
1
13
11.
Alessi
J
,
de Oliveira
GB
,
Schaan
BD
,
Telo
GH
.
Dexamethasone in the era of COVID-19: friend or foe? An essay on the effects of dexamethasone and the potential risks of its inadvertent use in patients with diabetes
.
Diabetol Metab Syndr
2020
;
12
:
80
12.
Horby
P
,
Lim
WS
,
Emberson
JR
, et al.;
RECOVERY Collaborative Group
.
Dexamethasone in hospitalized patients with Covid-19
.
N Engl J Med
2021
;
384
:
693
704
13.
Donihi
AC
,
Raval
D
,
Saul
M
,
Korytkowski
MT
,
DeVita
MA
.
Prevalence and predictors of corticosteroid-related hyperglycemia in hospitalized patients
.
Endocr Pract
2006
;
12
:
358
362
14.
Sosale
A
,
Sosale
B
,
Kesavadev
J
, et al
.
Steroid use during COVID-19 infection and hyperglycemia: what a physician should know
.
Diabetes Metab Syndr
2021
;
15
:
102167
15.
American Diabetes Association
.
13. Diabetes care in the hospital
.
Diabetes Care
2016
;
39
(
Suppl. 1
):
S99
S104
16.
Fleming
N
,
Sacks
LJ
,
Pham
CT
,
Neoh
SL
,
Ekinci
EI
.
An overview of COVID-19 in people with diabetes: pathophysiology and considerations in the inpatient setting
.
Diabet Med
2021
;
38
:
e14509
17.
Rodriguez-Gutierrez
R
,
Gonzalez-Gonzalez
JG
,
Zuñiga-Hernandez
JA
,
McCoy
RG
.
Benefits and harms of intensive glycemic control in patients with type 2 diabetes
.
BMJ
2019
;
367
:
I5887
18.
Asiri
AA
,
Alguwaihes
AM
,
Jammah
AA
,
Alfadda
AA
,
Al-Sofiani
ME
.
Assessment of the effectiveness of a protocol to manage dexamethasone-induced hyperglycemia among hospitalized patients with COVID-19
.
Endocr Pract
2021
;
27
:
1232
1241
19.
Faul
F
,
Erdfelder
E
,
Buchner
A
,
Lang
AG
.
Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses
.
Behav Res Methods
2009
;
41
:
1149
1160
20.
IBM Corp
.
SSPS, v. 28.0
.
Armonk, NY
,
IBM Corp.
,
2020
21.
Radhakutty
A
,
Stranks
JL
,
Mangelsdorf
BL
, et al
.
Treatment of prednisolone-induced hyperglycaemia in hospitalized patients: insights from a randomized, controlled study
.
Diabetes Obes Metab
2017
;
19
:
571
578
22.
Saisho
Y
.
β-Cell dysfunction: its critical role in prevention and management of type 2 diabetes
.
World J Diabetes
2015
;
6
:
109
124
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.