As of March 2024, coronavirus disease-19 (COVID-19) has claimed the lives of over 7 million people worldwide [1] . In the early reports on the emergence and clinical aspects of COVID-19, obesity and type 2 diabetes (T2D) were cited as important risk factors for the development of severe forms of the disease [2] [3] . Several mechanisms have been described as potentially being involved in the overlap of COVID-19, obesity, and T2D, suggesting that it results from multiple factors, including genetic, epigenetic, and environmental [4] [5] .

Regardless of the mechanisms behind these common epidemiological associations, the severity of the disease has imposed therapeutic puzzles, as warranting metabolic stability directly impacts COVID-19 outcomes [4] [6] . An important therapeutic question that emerged early during the COVID-19 pandemic was whether the use of insulin to control glucose levels and metabolic abnormalities in T2D patients who developed COVID-19 would be safer and more effective than other glucose-lowering agents [7] - [9] . After the first year of the pandemic, a review published in Nature Reviews Endocrinology [9] proposed that insulin was the best option as compared to inhibitors of sodium-glucose cotransporter 2 (iSGLT2), and inhibitors of dipeptidyl peptidase 4 (iDPP4). A similar recommendation was published in a review in The Lancet: Diabetes & Endocrinology [10] . However, at about the same time, a study reported that the use of insulin was associated with increased mortality in patients with COVID-19 and T2D, and suggested that this outcome was due to increased systemic inflammation and aggravated injuries of vital organs [11] . The publication of this study initiated an important debate in the field, and after three years, we are still uncertain about the impact of insulin use on T2D patients with COVID-19.

In 2022, a large meta-analysis evaluated the outcomes of treating T2D in COVID-19 patients with iSGLT2, iDDP4, metformin, glucagon-like-1 receptor antagonists (GLP1-RA), thiazolidinediones (TZD), inhibitors of alpha-glucosidase, secretagogues and insulin [12] . The study showed that iSGLT2 and GLP1-RA were the drugs with the lowest, whereas insulin was the drug with the highest risk of adverse outcomes [12] . However, the study included articles published until September 2022, which means that it was representative of a very particular phase of the pandemic when mortality was high and there was no vaccine available. At present, the spectrum of patients developing severe COVID-19 has changed as compared to the early days of the pandemic, and this could impact the responsiveness and complications attributed to insulin. Therefore, we hypothesized that the use of insulin to treat patients with T2D and COVID-19 could represent a risk for increased mortality. Thus, we conducted a systematic review to determine the current status of the outcomes associated with the use of insulin in T2D patients who develop severe COVID-19. Our results indicate that the use of insulin increases the mortality of severe COVID-19 patients with T2D.

This was a systematic review that explored the clinical impact of using insulin in hospitalized T2D patients with COVID-19 (in four articles, highlighted in Table 1 , the patients were already using insulin prior to hospitalization). Two authors (CFB and MRS), independently, identified studies published until August 31, 2023, which contained data on COVID-19 patients with T2D using or not using insulin. The search was conducted separately in four databases: PubMED-NCBI, Scopus, Web of Science, and Google Scholar. English was the chosen language for search and articles.

The selected keywords and their synonymous terms, which were selected from the MeSH database, were: 1) “insulin” OR “regular insulin” OR “soluble insulin”; 2) “diabetes mellitus, type 2” OR “diabetes mellitus type 2” OR “diabetes mellitus type II” OR “type 2 diabetes” OR “diabetes mellitus, non-insulin-dependent”; iii, “covid-19” OR “SARS CoV 2” OR “2019 novel coronavirus” OR “2019 nCoV” OR “covid-19” OR “severe acute respiratory syndrome coronavirus 2 infection” OR “coronavirus disease 2019”. All these synonyms were combined using “AND”, and the search was performed in the title and/or abstract. The search was conducted similarly in all databases.

From all the articles found in the four databases, the researchers independently excluded duplicates and selected the relevant articles, based on the title and abstract—those directly or indirectly related to the correlation of favorable and unfavorable outcomes with the use or non-use of insulin in patients T2D and with COVID-19. Once this process was concluded, the authors compared their findings (cross-reference), resulting in the final pre-selected list of 543 articles.

From this list, 480 articles were screened for one or more of the following exclusion criteria, previously defined by the researchers: editorials, meta-analyses, systematic reviews, experimental studies, comments, letters to the Editor, studies without descriptions of patients’ clinical characteristics and/or current therapeutic regimens, studies that did not follow the selected clinical criteria, duplicated articles, and texts in languages other than English. After exclusions, 63 articles remained for the study analysis. Finally, the authors independently and carefully analyzed each of the 63 articles to determine those with the necessary and accessible data, which means, the articles presented in detail the pharmacological interventions for each group of patients and the respective clinical outcomes. In case of discrepancies, a third author (LAV) was consulted. As a result, 29 articles were included in the final analysis ( Figure 1 ). Risk of bias was evaluated by three authors, independently (CFB, MRS and LAV).

For the statistical analysis, the assumption was that the use of insulin to treat T2D in patients with COVID-19 could increase mortality. It was determined the probability of the outcome, under analysis, to occur, using the odds ratio. This analysis was made for all the study designs, together and separately. The outcome under analysis was death. Patients using insulin were considered the intervention group, and patients without insulin were considered the control group. In addition, an unpaired student’s t-test was used to determine the statistical difference of the percentage of alive patients, between patients using insulin and those not using insulin. The significant difference was defined as p < 0.05. The software employed for statistical analysis was StatView for Windows.

This systematic review evaluated the outcomes of 3,070,657 patients with severe COVID-19 and T2D that required pharmacological intervention to treat hyperglycemia ( Table 1 ).

*Home-use insulin.

The outcomes of the patients were reported in 29 articles published between December 2019 and August 2023, and included detailed information about the medication that was used and the final outcome, as deceased or alive ( Figure 1 ).

First, we analyzed the results separately, in each type of study. Two case-control studies reported the outcomes of 14,503 patients ( Table 2 ); 1,941 under insulin therapy and 12,562 under other modalities of intervention except for insulin.

In these studies, there was increased mortality in the insulin group with an odds ratio of 1.92. In two case report articles that included 95 patients (31 treated with insulin and 64 with no insulin), there was increased mortality in the insulin group with an odds ratio of 4.0 ( Table 3 ).

In four cross-sectional studies, there were 972 patients (610 treated with insulin and 362 with no insulin), and there was an increased mortality in the insulin group with an odds ratio of 2.94 ( Table 4 ).

Three observational cohort studies reported the outcomes of 2,801,039 patients ( Table 5 ); 342,184 were under insulin therapy and 2,458,855 were under

other modalities of intervention except for insulin. In these studies, there was increased mortality in the insulin group with an odds ratio of 2.25.

Finally, 18 retrospective studies evaluated 254,048 patients; 67,922 were treated with insulin, and 186,126 were treated with other modalities of intervention except for insulin ( Table 6 ). In these studies, there was increased mortality in the insulin group with an odds ratio of 1.46.

Next, we used two approaches to evaluate the whole sample: the determination of the odds ratio, and a direct comparison of the two groups using unpaired Student’s t-test. In the evaluation of all 3,070,657 patients included in this study (412,688 treated with insulin and 2,657,969 treated with another approach except for insulin) there was an odds ratio of 2.93 indicating greater mortality in the insulin group. This was further confirmed by the determination of the percentage of alive patients in either group ( Figure 2 ).

In this study, we show that using insulin as a therapeutic intervention to treat hyperglycemia and metabolic abnormalities in patients with T2D who develop severe COVID-19, increases the probability of death by almost 200%, which is similar to a previous publication [12] . While deaths related to COVID-19 are also closely related to the clinical condition of the disease, the more severe forms of the illness are strongly associated with the presence of comorbidities [13] . Among these, T2D plays an important role due to its high prevalence in the population and its impact on several organs and systems that play important roles in whole-body homeostasis. As a chronic disease that is frequently poorly controlled, T2D impacts several clinical parameters that negatively influence the prognosis of COVID-19, such as the functional impairment of phagocytes, and the abnormal expression of the SARS-CoV2 receptor, ACE2, both of which directly interfere with the pathogenesis of COVID-19 [14] - [16] .

Given the high prevalence and the pathophysiological complexity of diabetes mellitus, it is not a surprise that up to 58% of all COVID-19 patients have diabetes [17] . Moreover, infected patients with diabetes have an increased 8% risk of death and are nearly 15% more likely to require intensive care as compared to individuals without this comorbidity [18] . Thus, these patients need special attention in the treatment of COVID-19, as well as rigorous management of diabetes.

Despite recent advances in the pharmacological treatment of diabetes, a considerable fraction of patients use insulin as monotherapy or in conjunction with other drugs. In addition, when patients with diabetes develop severe infectious diseases or other medical conditions that require intensive care, insulin is commonly used to maintain metabolic stability, and this is recommended in the most important guidelines for diabetes management [19] [20] . Thus, it was not a surprise that in the early days of the COVID-19 pandemic, important journals published reviews, editorials, commentaries, and expert opinions recommending insulin as the drug of choice for the treatment of hyperglycemia in severe COVID-19 [9] [10] . However, in early 2021, a study reported for the first time that using insulin to treat severely ill COVID-19 patients with diabetes resulted in increased mortality [11] . The study retrospectively evaluated 689 patients with COVID-19 and T2D and showed an increased mortality in the insulin-treated group with a hazard ratio of 5.38. The authors suggested that mortality could be due to increased inflammatory activity and increased damage to vital organs [11] .

The publication of that study [11] triggered a series of questions regarding the optimal pharmacological intervention to treat diabetes in patients with COVID-19. In addition, it stimulated the search for the mechanisms behind this putative adversity. As a whole, we found seven articles [12] [21] - [26] that performed systematic reviews/meta-analyses to determine the impact of using insulin to treat hyperglycemia in patients with severe COVID-19. In all of them, the results indicated that using insulin resulted in increased complications and/or mortality. However, the studies were all published between July 2021 and October 2022; thus, they are representative of a phase of the pandemic with high mortality rates, and when most patients had not been submitted to vaccination.

In the present study, we revisited the theme, incorporating articles published in late 2022 and 2023, thus expanding the analysis for a new phase when COVID-19 is no longer a pandemic, and when the number of people fully vaccinated is much greater than when the prior studies were performed. In addition, the selection of articles was very strict, including only those with a clear description of therapeutic interventions for the treatment of hyperglycemia, and the relation of such interventions with one out of two outcomes: alive or deceased. Our results reinforce the concept that using insulin in patients with severe COVID-19 increases mortality and have a broad impact on clinical science, which goes much beyond diabetology and endocrinology. In addition, it calls for actions aimed at defining the mechanisms that contribute to the increased rate of mortality in insulin-treated patients.

CFB and MRS performed an article search and the first round of inclusions. LAV performed the second round of inclusion. MRS and MM-P performed statistics analysis. LAV, MRS, TPP and CFB wrote the manuscript. All authors read the manuscript and provided approval.

This research received funding from the São Paulo Research Foundation (2022/00956-6 and 2021/09662-2). The study was also supported by FAPESP (2013/07607-8).

*Corresponding Author.

^#Contributed equally as first author of this study.