The impact of insulin requirement on mortality and morbidity in non-diabetic covid-19 patients in the intensive care unit: A retrospective, observational study

Abstract

Background

COVID-19 ranges from asymptomatic cases to severe disease with high mortality. Corticosteroids are crucial in treatment, reducing mortality and morbidity. However, the use of corticosteroids poses additional challenges in maintaining glycemic control in COVID-19 patients This study aims to eva-luate the impact of insulin requirement on mortality and morbidity in non-diabetic ICU patients and investigate its correlation with disease severity.

Methods

This retrospective cohort study included non-diabetic COVID-19 patients aged ≥ 18 years admitted to the ICU of Prof. Dr. Cemil Taşcıoğlu City Hospital (Turkey) between September 1, 2020, and May 31, 2021. Patients requiring ≥ 24 h of insulin therapy were compared with those who did not need insulin. Data on demographics, severity scores (SOFA, APACHE II, SAPS II), insulin initiation and duration, corticosteroid therapy, mechanical ventilation, antiviral and immunomodulatory treatments, laboratory markers, and infection parameters were analyzed. Mortality and incidence of new-onset diabetes mellitus within the first six months post-discharge were assessed. Statistical analyses were performed using SPSS v22.0, with p < 0.05 considered statistically significant.

Results

Patients with insulin requirements had higher SOFA (p = 0.001), APACHE II (p < 0.001), and SAPS II (p = 0.041) scores, along with increased mechanical ventilation duration (p < 0.001). While corticosteroid type had no effect, > 1 mg/kg/day methylprednisolone or equivalent dexamethasone significantly increased insulin demand (p = 0.002). Among laboratory markers, only peak CRP levels were significantly higher in insulin-requiring patients (p = 0.001). ICU and total hospital stays were significantly longer in the insulin group (p < 0.001). Although in-hospital mortality was similar, 6-month mortality was significantly higher in insulin-requiring patients (p = 0.022). New-onset DM rates were 4.2% in the non-insulin group vs. 31.1% in the insulin group (p = 0.001).

Conclusions

Insulin requirement in non-diabetic COVID-19 ICU patients is a predictor of 6-month mortality. High-dose corticosteroids exacerbate glycemic dysregulation, increasing insulin needs. SARS-CoV-2-induced beta-cell damage and hyperinflammation-related stress hyperglycemia elevate the risk of post-discharge DM. Close monitoring and diabetes screening are essential in this population.

Background

Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was first identified in December 2019 in Wuhan, China, and rapidly escalated into a global health crisis [1]. The clinical presentation of COVID-19 varies widely, ranging from asymptomatic cases to mild flu-like symptoms [2], while some patients progress to severe pneumonia, acute respiratory distress syndrome (ARDS), and multiorgan failure [3]. A hallmark of severe COVID-19 is the excessive inflammatory response, which plays a critical role in disease progression and adverse outcomes [4].

Systemic inflammation in COVID-19 is associated with various metabolic disturbances, including hyperglycemia, even in patients without a prior diagnosis of diabetes mellitus (DM) [5]. Elevated glucose levels not only serve as a marker of stress response but also contribute to the inflammatory burden, exacerbating disease severity and increasing the risk of complications [6]. Among the therapeutic strategies employed in critically ill COVID-19 patients, corticosteroids (CS) have demonstrated significant mortality benefits due to their potent anti-inflammatory effects [7]. However, corticosteroid use is also linked to impaired glucose regulation, insulin resistance, and an increased risk of secondary infections, raising concerns about its metabolic consequences in intensive care unit (ICU) patients [8, 9].

Considering these interactions, the impact of corticosteroid-induced hyperglycemia in non-diabetic COVID-19 patients remains a crucial area of investigation. Despite normal hemoglobin A1c (HbA1c) levels at ICU admission, some patients develop insulin-requiring hyperglycemia, potentially influencing morbidity and mortality [10]. Our hypothesis is that insulin-requiring patients in the ICU have worse clinical outcomes due to the combined effects of hyperglycemia and systemic inflammation. Therefore, this study aims to explore the correlation between insulin requirements, mortality, morbidity, and disease severity in non-diabetic COVID-19 patients receiving corticosteroid therapy in the ICU.

Materials and methods

This study was approved by the Turkish Ministry of Health Clinical Research Ethics Committee (Decision No. 77194, dated 09.11.2021) and the Ethics Committee of Prof. Dr. Cemil Taşcıoğlu City Hospital, Health Sciences University (Decision No. 447, dated 03.01.2022). The study was conducted in accordance with the Helsinki Declaration and European Union Council directives (ETS 123; 86/609/EEC). Due to the pandemic, verbal informed consent was obtained from patients or their relatives. This study received no financial support.

This single-center, retrospective study included ICU patients aged over 18 years who were hospitalized for more than 48 h with a COVID-19 diagnosis and received corticosteroid treatment. Patients with a history of chronic disease requiring corticosteroid therapy before COVID-19, those with a pre-existing diagnosis of DM, patients with HbA1c > 5.7% at hospital admission, and those with incomplete medical records were excluded from the study.

Study design

Patients diagnosed with COVID-19 and admitted to the ICU between September 1, 2020, and May 31, 2021, were retrospectively evaluated. The study compared patients requiring insulin therapy to those who did not require insulin therapy.

ICU length of stay was defined as at least 48 h, while insulin requirement was defined as insulin therapy initiated according to ICU blood glucose regulation protocols and continued for more than 24 h. According to the ICU blood glucose regulation protocols, blood glucose levels should be maintained within the range of 110–180 mg/dL. If consecutive measurements obtained at 1-hour intervals exceed 180 mg/dL, insulin therapy is initiated. In this protocol, a 1-unit intravenous (IV) bolus of insulin is administered when blood glucose levels range between 180 and 270 mg/dL, while levels exceeding 270 mg/dL necessitate a 2-unit IV bolus. Following bolus administration, a continuous insulin infusion is initiated in accordance with the protocol algorithm. Blood glucose measurements were routinely performed by ICU nurses every four hours, and in patients receiving insulin therapy, glucose levels were monitored hourly.

Data Collection: Demographic data, including age, gender, and body mass index (BMI), were recorded. Comorbidities were categorized as follows: hypertension (HT), cardiovascular diseases (ischemic heart disease, coronary artery bypass grafting, arrhythmias, pacemaker, heart failure), respiratory diseases (asthma, chronic obstructive pulmonary disease, previous tuberculosis, pulmonary embolism, obstructive sleep apnea syndrome), cerebrovascular disease (CVD), hypothyroidism, malignancies (endometrial, colon, thyroid, breast, bladder, pancreatic, lung, hematological), neurological disorders (epilepsy, subacute sclerosing panencephalitis, myasthenia gravis), degenerative diseases (Alzheimer’s, Parkinson’s, dementia), chronic kidney disease (CKD), and benign prostatic hyperplasia (BPH).

The ASA (American Society of Anesthesiologists score) score at ICU admission, APACHE II (Acute Physiology and Chronic Health Evaluation), SOFA (Sequential Organ Failure Assessment Score), and SAPS II (Simplified Acute Physiology Score II) scores were recorded. Ventilation strategies, both invasive and non-invasive, PaO2/FiO2 ratios at ICU admission, and acute respiratory distress syndrome (ARDS) severity were documented.

Non-invasive ventilation strategies were classified as high-flow nasal oxygen (HFNO) and continuous positive airway pressure (CPAP), while the invasive ventilation strategy was defined as orotracheal intubation. ARDS severity was categorized based on the PaO2/FiO2 ratio as follows: 0-100 severe ARDS, 100–200 moderate ARDS, and 200–300 mild ARDS.

During the ICU stay, patients’ insulin requirements, dosage, initiation time, and duration were recorded, along with the type, dosage, initiation time, and duration of corticosteroid therapy administered. The patients were evaluated in two groups based on insulin requirements: Group A (control group, no insulin required) and Group B (patients requiring insulin). Changes in prognostic laboratory parameters and medical treatments, including antiviral therapy, tocilizumab, immune plasma therapy, and cytokine-filtered dialysis therapy, were determined.

The administration of immunomodulatory therapies, including tocilizumab, cytokine-filtered dialysis, and immune convalescent therapy, was recorded. Prognostic laboratory parameters, including white blood cell (WBC) count, platelet-to-lymphocyte ratio (PLR), neutrophil-to-lymphocyte ratio (NLR), ferritin, D-dimer, C-reactive protein (CRP), procalcitonin (PRC), and interleukin-6 (IL-6), were documented at ICU admission, along with their peak values and the day on which the peak occurred.

Patient follow-up was conducted until death or up to six months post-discharge, with data collected on hospital and ICU length of stay, mortality, morbidity, and new-onset diabetes mellitus (DM) diagnosis within this period.

Statistical Analysis: Statistical analyses were performed using SPSS (Statistical Package for the Social Sciences, version 22.0, Chicago, IL, USA). Continuous variables were presented as mean ± standard deviation or median (minimum–maximum), while categorical variables were expressed as counts and percentages. The normality of distribution for continuous variables was assessed using the Shapiro-Wilk test. For normally distributed data, comparisons between independent groups were performed using the t-test, while the Mann-Whitney U test was used for non-normally distributed data. Comparisons involving three or more groups were conducted using the Kruskal-Wallis test. Categorical variables were analyzed using the Chi-square test. A p-value of < 0.05 was considered statistically significant.

Results

Between September 1, 2020, and May 31, 2021, a total of 305 patients without a prior diagnosis of diabetes mellitus (DM) were admitted to the ICU due to COVID-19. Among these, 37 patients died within the first 48 h of ICU admission, 24 patients were excluded due to incomplete data, and 34 patients were excluded due to chronic diseases requiring corticosteroid use. Additionally, 49 patients were excluded due to an HbA1c level > 5.7% at admission despite having no known DM diagnosis. A total of 161 patients who met the inclusion criteria were included in the study. Among these, 73 patients did not require insulin therapy (Group A), while 88 patients required insulin therapy (Group B) (Fig. 1). Patients admitted to intensive care were evaluated for eligibility criteria for the study over a 9-month period, and data were collected. They were followed for 6 months to assess mortality and morbidity after discharge. The study lasted a total of 15 months.

Fig. 1

Flowchart of the patients

The mean age of the patients was 64.75 ± 15.28 years. There was no significant difference in mean age between Group A and Group B (65.38 ± 16.57 vs. 66.72 ± 13.90, p = 0.072). Gender distribution was also similar between the groups (p = 0.064). The SOFA, APACHE II, and SAPS II scores of Group B were significantly higher than those of Group A (p = 0.001, p < 0.001, and p = 0.041, respectively). The PaO2/FiO2 ratio at admission was similar between patients with and without insulin requirement (p = 0.194). There was no significant association between ARDS severity and insulin requirement. The proportion of patients requiring insulin therapy was similar across ARDS severity categories (p = 0.194) (Table 1).

Table 1 Demographic data

Table 2 compares the ventilation type, corticosteroid use, and other medications between the patient groups. The duration of non-invasive mechanical ventilation was significantly longer in Group B than in Group A (10 [2–28] days vs. 6 [1–16] days, p < 0.001). Insulin requirement was significantly higher in patients who underwent invasive mechanical ventilation (75.9%, p = 0.022). The duration of mechanical ventilation was 12 (4–18) days in Group B and 7 (4–22) days in Group A; however, the difference was not statistically significant (p = 0.098).

Table 2 Group distribution based on treatment

No significant correlation was found between corticosteroid treatment and insulin requirement (p > 0.005). Among patients receiving methylprednisolone (80% of the cohort), 57.4% required insulin therapy (p = 0.166). Dexamethasone treatment was administered to 36.6% of patients, and insulin requirement was similar between those who received dexamethasone and those who did not (p = 0.805). Among the 16 patients who received pulse steroid therapy, 7 required insulin therapy, while 9 did not (p = 0.356). Patients who received corticosteroid doses below the equivalent of 1 mg/kg had a significantly lower insulin requirement than those who received higher doses (43.4% vs. 66.7%, p = 0.002). The duration of corticosteroid use was significantly longer in Group B (p < 0.001). In patients requiring insulin after corticosteroid administration, insulin therapy was initiated on an average of day 8.

The proportions of patients requiring insulin therapy were similar among those who received tocilizumab, immune plasma therapy, or cytokine filtration (p = 0.487, p = 0.403, and p = 0.056, respectively). The peak CRP levels were significantly higher in Group B than in Group A (p = 0.001). Other laboratory parameters, including international normalized ratio and peak values, did not show significant differences between the groups (p > 0.05) (Table 3).

Table 3 Distribution of laboratory results in patients groups

Of the patients, 57.8% (n = 93) were discharged, while 42.2% (n = 68) died. In Group B, 51.1% of patients were discharged, compared to 65.8% in Group A. The discharge and mortality rates were similar between the groups (p = 0.062). The six-month mortality rate was 44.1% for the entire cohort. The six-month mortality rate in Group B (52.3%) was significantly higher than in Group A (34.2%) (p = 0.022). Among discharged patients, 16 (17.2%) were diagnosed with diabetes. The proportion of newly diagnosed diabetes cases was significantly higher in Group B (31%) than in Group A (4.2%) (p = 0.001) (Table 4).

Table 4 Relationship between insulin requirement, hospital discharge status, mortality, and diabetes mellitus diagnosis

The hospital length of stay was significantly longer in Group B than in Group A (25 [6–57] days vs. 17 [4–41] days, p < 0.001) (Fig. 2). Similarly, the ICU length of stay was significantly longer in Group B (14.5 [3–52] days) than in Group A (9 [2–32] days) (p < 0.001) (Fig. 3).

Fig. 2

Hospital stay distribution by patient groups

Fig. 3

ICU stay distribution by patient groups

Among discharged patients, the median insulin dose was 3 (2–6) IU, whereas among deceased patients, it was significantly higher at 4 (2–7) IU (p = 0.006). There was no significant difference in insulin administration time between discharged and deceased patients (p = 0.980). Among patients who died within six months, the median insulin dose was 4 (2–7) IU, whereas among survivors, it was 3 (2–6) IU, and this difference was statistically significant (p = 0.003). The duration of insulin administration was similar between deceased and surviving patients (p = 0.864). There was no significant difference in insulin dose or administration duration based on ARDS severity (p = 0.421 and p = 0.924, respectively). Similarly, there was no significant difference in insulin dose and administration duration between patients who underwent invasive mechanical ventilation and those who did not (p = 0.106 and p = 0.102, respectively) (Table 5).

Table 5 Distribution of insulin dose and administration duration

Discussion

In this study, it was observed that insulin requirement in non-diabetic patients who required intensive care unit (ICU) admission due to COVID-19 was associated with increased six-month post-discharge mortality. Additionally, regardless of the type of corticosteroid used, patients who received > 1 mg/kg methylprednisolone or an equivalent dose of dexamethasone had a higher insulin requirement. As a control group, COVID-19 patients without diabetes mellitus who did not require insulin during the same period were evaluated.

A pre-pandemic study including 1764 non-diabetic critically ill patients reported that 70% of those requiring insulin for glycemic control were male [11]. It is well established that COVID-19 is more prevalent in males due to differences in sex hormones and immune response [12, 13]. Although the gender distribution between Group A and Group B in our study was statistically similar, the number of male patients was higher in both groups. Given the limited sample size, we believe that the gender distribution was comparable between insulin-requiring and non-insulin-requiring groups.

In critically ill non-diabetic patients without renal or hepatic failure or corticosteroid use but exhibiting stress hyperglycemia, higher APACHE II, SOFA, and SAPS II scores have been reported [14]. In our study, these scoring systems showed significantly higher numerical values in Group B patients. Compared to the pre-pandemic period, stricter ICU admission criteria due to pandemic-related strain, the viral load and hyperinflammatory state of COVID-19, and the widespread use of corticosteroids, which further disrupted glucose regulation, may have contributed to this finding. Additionally, the presence of comorbidities such as hepatic, renal, or cardiac failure, which can increase mortality, may have played a role [15].

A study of 413 patients during the first wave of the COVID-19 pandemic (February-April 2020) demonstrated a stronger association between hyperglycemia and COVID-19 severity in newly diagnosed diabetic patients compared to those with pre-existing diabetes [16]. Hyperglycemia has been found to be a negative prognostic factor for patients admitted to hospitals during the COVID-19 epidemic [17, 18]. Unlike these studies, our patient group consisted of individuals who had no history of DM or hyperglycemia and had normal HbA1c levels prior to ICU admission. While it is well-known that existing DM worsens prognosis, in this study, we examined the impact of COVID-19 and corticosteroid treatment on the development of DM and its effect on prognosis in patients with previously normal blood glucose regulation.

Li et al. [19] compared non-diabetic hyperglycemic patients, newly diagnosed diabetics, and known diabetics hospitalized for COVID-19. They found that newly diagnosed diabetes exacerbated disease severity, classifying severity according to the Chinese COVID-19 Management Guidelines (mild, severe, and critical ARDS). However, limitations of this study included the absence of HbA1c measurements in some patients and cases where normal-range HbA1c did not necessarily indicate adequate glucose regulation. In our study, we confirmed the non-diabetic status of our patients using admission HbA1c levels, ensuring a more specific approach to COVID-19-associated diabetes mellitus (DM) diagnosis. Since all patients were non-diabetic, the presence of “diabetic lung” was excluded, and while the PaO2/FiO2 ratio was lower in Group B patients, this difference was not statistically significant. No significant correlation was found between ARDS severity, PaO2/FiO2 ratio, and insulin requirement. Given the evolving ICU admission criteria and the establishment of high-flow nasal oxygen (HFNO) units outside the ICU, the similarity in PaO2/FiO2 ratios between groups could be explained.

A study comparing stress hyperglycemia in non-diabetic critically ill patients with normoglycemic non-diabetics and diabetic critically ill patients found no difference in the duration of invasive mechanical ventilation between groups. However, this study was conducted before the pandemic, and COVID-19 pneumonia was not a contributing factor [20]. In our study, insulin requirement was significantly higher among patients requiring invasive mechanical ventilation, although the duration of mechanical ventilation did not differ significantly between groups. Type 2 diabetic patients with poorly regulated glucose levels tend to have longer durations of non-invasive and invasive mechanical ventilation; however, the lack of data on pre-hospital glucose regulation in these patients remains a limitation [21].

Corticosteroids are a cornerstone of COVID-19 treatment, with dexamethasone, methylprednisolone, or pulse steroid therapy being commonly used. However, the impact of high-dose versus low-dose corticosteroids on glucose regulation in non-diabetic and diabetic COVID-19 patients remains unclear. Studies suggest no significant difference in hyperglycemia incidence between high-dose and low-dose corticosteroid-treated COVID-19 patients, though these studies included diabetic patients whose glucose regulation may already have been impaired [22]. A meta-analysis of pneumonia patients caused by coronaviruses found no effect of corticosteroids on hyperglycemia, although steroid dosage was not specified [23]. The CODEX study reported similar rates of insulin-requiring hyperglycemia in patients receiving high-dose dexamethasone and those not receiving dexamethasone, though diabetic patients were present in both groups, and baseline HbA1c levels were not recorded [24]. Most studies have not excluded diabetic patients, typically comparing general groups receiving high or low-dose corticosteroids. While the GLUCOCOVID study supports our findings, it also did not exclude diabetic patients or assess prior corticosteroid use or HbA1c values [25]. Our study found no significant association between corticosteroid type and insulin requirement, regardless of dosage. However, insulin requirement was lower in patients receiving < 1 mg/kg methylprednisolone or an equivalent dexamethasone dose.

Glucose dysregulation in COVID-19 patients can be influenced by corticosteroids, stress hyperglycemia, prolonged ICU stays, and direct pancreatic beta-cell damage by SARS-CoV-2. Despite the known adverse effects of corticosteroids on glucose regulation, their role in COVID-19 treatment remains crucial. A WHO REACT meta-analysis demonstrated that systemic corticosteroids reduced 28-day all-cause mortality in critically ill COVID-19 patients compared to standard care or placebo [26]. In our study, no significant association was found between the type of corticosteroid used and insulin requirement, regardless of the administered dose. However, in non-diabetic patients without comorbidities requiring corticosteroid therapy, the use of methylprednisolone at < 1 mg/kg or an equivalent dose of dexamethasone was associated with a lower insulin requirement. The mean onset of insulin requirement after corticosteroid initiation was observed to be on day 8 (range: 1–45). However, it is important to consider other factors that may contribute to impaired glucose regulation during this period.

In the study by Li et al. [19], patients with newly diagnosed diabetes and pre-existing diabetes had significantly higher CRP, D-dimer, fibrinogen, and WBC levels. However, no significant difference was observed between non-diabetic hyperglycemic and normoglycemic patients. Additionally, no significant difference was found in the length of hospital stay. The researchers did not specify the type or dose of corticosteroid used, and their study only included laboratory measurements from the first three days, focusing on patients followed up in general hospital wards.

Similarly, Zhang et al. [27] compared non-diabetic normoglycemic, hyperglycemic, and diabetic patients and found that the non-diabetic hyperglycemic group had higher levels of leukocytosis, neutrophilia, lymphocytopenia, eosinopenia, CRP, LDH, ferritin, and D-dimer. However, as previously mentioned, these values represent the average measurements over the entire hospitalization period, and the proportion of patients exceeding normal limits was not reported. Additionally, the hospital stay duration was found to be longer in diabetic patients.

In our study, prognostic laboratory parameters, including WBC, ferritin, D-dimer, CRP, PRC, platelet-to-lymphocyte ratio (PLR), neutrophil-to-lymphocyte ratio (NLR), and IL-6, were evaluated at ICU admission and on the days when their peak values were recorded. Considering the potential for secondary bacterial infections and prolonged ICU stays to elevate these values, only admission and peak values were assessed. However, apart from CRP, no significant differences were observed. The highest CRP levels were significantly higher in patients requiring insulin therapy. We attribute this finding to the longer length of hospital stay, the exclusive inclusion of ICU patients in our cohort, and the presence of secondary infections.

The average hospital stay was 25 days in Group B and 17 days in Group A, while the ICU stay was 14.5 days in Group B and 9 days in Group A, both statistically significant differences. Additionally, the high admission SOFA scores and the fact that 90.6% of patients had moderate or severe ARDS suggest that elevated prognostic markers were expected.

Wu et al. [28] demonstrated that blood glucose levels serve as an independent risk factor in their study comparing critically and non-critically ill COVID-19 patients. Hyperglycemia was found to increase the risk of disease progression in non-critical patients and mortality in critically ill patients. However, their study did not consider HbA1c levels or diabetes status. In an unadjusted analysis, newly diagnosed diabetes in COVID-19 patients was associated with increased ICU admissions and mortality. However, this study was conducted in the early stages of the pandemic, and treatment protocols had not yet been updated. Furthermore, corticosteroid use was not mentioned [16].

Li et al. [19] and Zhang et al. [27] found no statistically significant difference in mortality between diabetic, non-diabetic normoglycemic, and hyperglycemic groups. However, newly diagnosed diabetic and pre-existing diabetic patients had higher mortality rates. A limitation of these studies was the small number of ICU-admitted and critically ill patients included.

In our study, in-hospital outcomes (discharge or death) were similar between groups. However, six-month post-discharge mortality was significantly higher in Group B. Notably, there were no non-disease-related causes of mortality. The higher mortality rate may be attributed to ICU admission, elevated APACHE II, SOFA, and SAPS II scores at admission, hyperglycemia-induced coagulation cascade alterations, endothelial dysfunction, excessive inflammatory cytokine production, and impaired airway epithelial defense due to increased glucose concentration in airway secretions.

In our cohort, 51% of Group B patients were discharged in good health, yet 31.1% were diagnosed with diabetes mellitus within six months post-discharge and required antidiabetic treatment. In contrast, this rate was only 4.2% in Group A. SARS-CoV-2 has been shown to directly damage pancreatic cells, induce insulin resistance due to viral infection, and disrupt glucose regulation through corticosteroid use, hyperinflammatory response, and cytokine storm [8, 29, 30, 31]. We believe these factors contribute to the persistent diagnosis of diabetes mellitus in these patients. Therefore, non-diabetic COVID-19 patients requiring insulin therapy in the ICU should be closely monitored and evaluated for antidiabetic treatment post-discharge. It should also be mentioned, nevertheless, that six months is not a long enough period to assess reversibility or permanence of DM. This is an area where extended patient follow-up could be implemented for further evaluation.

Study Limitations: Our study has some limitations. First, due to the high workload during the pandemic, documentation inconsistencies occurred. The retrospective nature of the study was necessitated by the challenges of conducting large-scale prospective research in this setting. Given the high burden of patient admissions, ICU admissions were selectively determined based on ICU bed availability. Focusing on the treatment of ARDS with the aim of improving patient survival, however, the absence of C-peptide and insulin secretion analyses, which are crucial for diagnosing DM but are not routinely monitored during COVID-19 treatment, represents a limitation of our retrospective observational study. Further research is needed to distinguish between stress hyperglycemia and direct viral beta-cell damage. The study cohort of 161 patients is not a large sample size. Moreover, treatment approaches evolved throughout the pandemic, leading to potential variations in patient management. This study includes only patients admitted to ICUs of our hospital, which primarily managed severe/critical COVID-19 cases.

Clinicians should pay attention to glucose regulation in non-diabetic COVID-19 patients, as it may serve as a mortality indicator, similar to diabetic patients. We believe our study provides evidence supporting the development of effective glucose regulation strategies and the need for closer follow-up of patients requiring insulin therapy.

Conclusion

The need for insulin therapy in non-diabetic COVID-19 patients during ICU stay increases the risk of all-cause six-month mortality. Additionally, the use of methylprednisolone at doses > 1 mg/kg or an equivalent dose of dexamethasone, regardless of the type of corticosteroid used, leads to greater glucose dysregulation and higher insulin requirements. The combination of corticosteroid therapy, direct pancreatic damage from SARS-CoV-2, and stress-induced hyperglycemia associated with COVID-19 may contribute to persistent diabetes mellitus. Therefore, patients requiring insulin therapy should be closely monitored and evaluated for antidiabetic treatment post-discharge.