Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a non-segmented, enveloped, positive-strand RNA virus causing a global pandemic of COVID-19 (1). COVID-19 is known to cause wide-ranging clinical complications, most notably respiratory failure. A number of studies have reported a close interaction between COVID-19 and metabolic disruption including diabetes mellitus (DM) (1). Type 1 diabetes mellitus (T1DM) is considered to be autoimmune in etiology and is characterized by insulin deficiency resulting from destruction of pancreatic ß-cells (2). Type 2 diabetes mellitus (T2DM) is characterized by persistent elevated blood glucose concentrations (hyperglycemia) in the setting of insulin resistance and hyperinsulinemia and accounts for nearly 90% of DM cases worldwide (3). Recent literature demonstrates a bidirectional relationship between diabetes and COVID-19 (4). Notably, a recent systematic review and meta-analysis the prevalence of DM increases with COVID-19 severity, and diabetes accounts for 9.5% of severe COVID-19 cases and 16.8% of COVID-19 related deaths (5). Importantly, a 2022 systematic review and meta-analysis of 10 articles spanning 11 patient cohorts (47.1 million total individuals) analyzed the relative risk of incident DM in patients with COVID-19. Investigators found a 64% greater risk of incident DM in patients with COVID-19 compared to non-COVID-19 infected controls (6).

A 2021 study showed patients with DM who tested positive for COVID-19 have an increased risk for intensive care unit (ICU) admission (17.6% vs. 7.8%) and mortality (20.3% vs 10.5%) compared to patients without DM who tested positive for COVID-19 (7). The degree of hyperglycemia appears to modulate COVID-19 severity, and a higher hemoglobin A1c (HbA1c) is associated with increased mortality (8). Despite an overall decline in mortality rates among individuals with COVID-19, poor outcomes persist among individuals with comorbid chronic disease (9). A retrospective study of 453 hospitalized patients that evaluated the association between the degree of hyperglycemia and the risk of all-cause mortality among hospitalized patients with COVID-19 found that patients with pre-existing DM or new-onset DM were more likely to experience complications, including acute kidney injury, compared to COVID-19 patients with normoglycemia (15.3-17.0% vs 1.5-3.1%), hypoalbuminemia (36.7% - 39.4% vs 10.8-19.4%), acute respiratory distress syndrome (ARDS) (3.1%-10.5% vs 0.8-3.1%) and severe COVID-19 complications (i.e., kidney disease, ischemic heart disease) (82.7-89.4% vs 61.4-72.15) (10).

Antidiabetic agents have both anti-inflammatory and immunomodulatory effects (11). Numerous studies, including systematic reviews and meta-analyses, have evaluated the effect of antidiabetic agents and mortality rates among individuals with DM (9). To date, the effect of antidiabetic agents on patient outcomes in hospitalized individuals with a dual diagnosis of DM and COVID-19 remains unclear. This study aimed to explore the relationship between different antidiabetic drug classes and patient outcomes among patients with DM who are hospitalized for COVID-19.

We conducted a systematic review of existing literature on outcomes for patients who were hospitalized for COVID-19. The review was registered in Prospero (CRD42023476297). Findings are reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (12).

The search strategy yielded a total of 6,639 articles for review. Removing duplicates left 4,898 articles for title and abstract screening ( Figure 1 ). The interrater reliability between the first and second coder, represented by the Cohens k coefficient, was 0.68 indicating substantial agreement (16). One hundred and eighteen articles were included for full-text review, Cohens k coefficient was 0.94, indicating substantial agreement (16). Of the articles retrieved for full-text review, 83 articles were excluded for not meeting study inclusion criteria (i.e. no inpatient pharmacologic management (n=46), outcomes not reported (n=12), abstract only (n=9), not primary research (n=6), poor quality (n=4), no diagnosis of DM (n=3), pediatric population (n=1), not published in English (n=1) and pre-print (not peer-reviewed n=1). A total of 35 articles were retained for analysis. The included studies reported research from 14 different countries. Notably, many of the studies originated from China (15/35, 42.8%) (11, 17–30). The remaining 20 studies were from the U.S.: n=3 (31–33), Iraq: n=3 (34–36), Israel n=2 (37, 38), Italy: n=2 (39, 40), Philippines: n=1 (41), Turkey: n=2 (42, 43), Brazil: n=1 (44), Indonesia: n=1 (45), Iran: n=1 (46), Denmark: n=1 (47), Poland: n=1 (48), Qatar and Kuwait: n=1 (49) and multinational=1 (50).

The majority of studies (31/35, 89%) used a retrospective design (11, 17–33, 35, 36, 38–45, 47, 48, 50). Three studies reported findings from a randomized controlled trial (RCT) (31, 37, 49) and one prospective cohort study (34). Cumulatively, included studies reported on 31,766 participants. Only one study from the U.S. included data on participant self-reported racial and ethnic identity (32). Diagnosis of COVID-19 was confirmed by polymerase chain reaction (PCR) testing or chest CT. Most studies (30/35, 85.7%) included individuals with T2DM (11, 17–28, 31, 32, 34–38, 40–43, 45–50). Two studies (2/35, 5.7%%) included both T1DM and T2DM (33, 44), and three studies (3/35, 8.6%%) did not specify diabetes type (29, 30, 39).

We report findings from a systematic review of the literature on antidiabetic agents and outcomes among patients with DM hospitalized for COVID-19. We synthesized findings from 35 included articles reporting on different antidiabetic agents: metformin, alpha-glucosidase inhibitors, meglitinides, TZD, SGLT2 inhibitors, DPP-4 inhibitors, and insulin. A summary of the two predominant outcomes (i.e. mortality and ICU admission) can be found in Table 2 .

A preliminary review of studies using the JBI tool identified high quality studies suitable for a meta-analysis on the risk and prevalence of mortality among antidiabetic agent users. Overall, our findings suggest a reduced relative risk of mortality among metformin and DPP-4 inhibitor users (RR = 0.432 (95% CI = 0.268-0.695, z = 3.45, p < 0.001)). This is consistent with findings in previous meta-analyses examining mortality risk with metformin use among antidiabetic users (51–58) and DPP-4 inhibitors (52, 59, 60). The overall prevalence of mortality was 16% (95% CI = 13%–19%, z = 10.70, p < 0.001).

Fourteen studies reported on the relationship between different antidiabetic agents and patient outcomes (18, 25–31, 36, 38, 42, 47, 48, 50). Studies were primarily retrospective in nature (n=14), conducted in China (n=7) and assessed individuals with T2DM (n=12). Twelve of the studies assessed antidiabetic use and mortality noting a decrease in mortality with the use of oral antidiabetic agents (metformin, acarbose, a-glucosidase inhibitors, meglitinides, TZD, SGLT2 inhibitors and DPP-4 inhibitors) (27–29). Conversely, an increase in mortality was associated with insulin use (18, 25, 30, 38). However, when insulin was combined with oral agents a decrease in mortality was reported (28). An increase in mortality with insulin use is consistent with previous studies (52). The underlying mechanism behind this phenomenon is unclear (52). However, from a clinical perspective, insulin is often given during late-stage DM suggesting other comorbidities may be contributing to increased mortality. This is further supported by increased rates of ICU admission with insulin use (25) in contrast to a decrease in ICU admission with alpha-glucosidase inhibitors (26, 29) and other antidiabetic agents (metformin, alpha-glucosidase inhibitors, sulfonylureas, and DPP-4 inhibitors) (25).

Metformin was associated with reduced disease severity, improved oxygenation, better glycemic control and decreased D-dimer (36). Previous studies have reported on the anti-inflammatory and antiviral effects of metformin, suggesting they may have protective mechanisms in individuals with COVID-19 (61). Similar to studies reporting on all diabetic agents, studies reporting on in-patient metformin use were primarily retrospective in nature and conducted in China. In parallel to previous findings, studies evaluating metformin use reported decreased mortality (11, 17, 20–24, 39, 45). Recent data suggests that there is an increase in oxidative stress in individuals with T2DM and COVID-19 increasing the risk of inflammatory and coagulation disorders (62). However, when metformin was used a decrease in inflammatory biomarkers (D-dimer, ferritin, LDH, and CRP) was reported (22, 34, 35).

Metformin has demonstrated the ability to impede the attachment of SARS-CoV-2 to ACE2 receptors by stimulating the phosphorylation of ACE2 (63). Additionally, it has been shown to modulate the immune response by fostering the generation of anti-inflammatory regulatory T cells and macrophages (63). Moreover, metformin suppresses the activation of inflammatory signaling pathways implicated in COVID-19 (64). These findings suggest that metformin could possess anti-inflammatory and antioxidative properties, potentially mitigating the robust inflammatory response and oxidative stress associated with SARS-CoV-2 infection (65). As a result, these effects may elucidate the observed reduction in inflammatory biomarkers. Furthermore, recent literature has reported that metformin may be beneficial in reducing long-COVID (66). Our study also noted a decreased incidence of ARDS (20, 21) and reduced risk for cardiac and respiratory events (17). However, one study noted an increase in incidence of ARDS (11) suggesting more research needs to be done to gain a better understanding of the role of metformin and incidence of ARDS. Three studies reported a significant relationship between metformin and length of stay - noting a reduced duration of hospitalization (17, 24, 46). Suggesting that patients who remained on metformin may have better outcomes when compared to non-metformin users.

Three studies evaluated individuals with T2DM and reported on the use of DPP-4 inhibitors and patient outcomes. Our analysis demonstrated a reduced risk of mortality with DPP-4 inhibitor use. However, there were only a few studies assessing this relationship, thus further research needs to be conducted to fully understand the association between the two. Previous studies have suggested that DPP-4 is a ubiquitous glycoprotein that could act as both a cell membrane protein and soluble enzymatic protein after cleavage and release. Recent data suggests that DPP-4 inhibitors could alter the outcomes in individuals positive for COVID-19 infection and alter oxygen requirements through their effects on the cardiovascular system (67). However, our findings suggest a varied results regarding DPP-4 inhibitors with one study reporting a decline in ICU admission (40), in contrast to one reporting no significant difference (37). Only one study reported on inflammatory biomarkers noting a reduction in PCT and CRP (40). Furthermore, no significant difference was found when evaluating ARDS or acidosis. Our findings on DPP-4 inhibitors are limited to only a few studies, future research needs to be done to determine underlying mechanisms contributing to patient outcomes in individuals diagnosed with DM and COVID-19.

One study focused solely on the use of insulin and patient outcomes, as previously noted increased mortality was associated with insulin use. While insulin is the recommended treatment for individuals hospitalized with DM (68), our findings suggest combining insulin with an oral antidiabetic agent may have positive effects on patient outcomes in patients with DM hospitalized for COVID-19. One study examined the effect of pioglitazone on T2DM hospitalized for COVID-19, however no significant findings were identified (49). Such findings are similar to previous literature examining drug efficacy on mortality (69). Last, one study examined dapagliflozin use in individuals with DM hospitalized for COVID-19 and their study reported lower ICU rates and reduced mortality among users (43). Tisch et al., conducted a systematic review examining the potential use of SGLT2 inhibitors in patients with acute illness (70). Findings are similar noting a reduced risk for mortality, however in individuals with prior use of SGLT2 inhibitors there was no association with ICU admission (70). We recognize that viewing individual medications is one dimensional and these outcomes are likely related to many other factors, such as disease severity, diabetes duration, age, comorbidities, medications and SARS-CoV-2 variant. Therefore it is important that further studies investigate the interactions between antidiabetic medications and other potential confounding factors.

A relative strength of this study is the comprehensive and systematic review and meta-analyses of the literature and synthesis of current evidence providing insight into clinical outcomes in patients with DM hospitalized for COVID-19. It merits noting that this work has several limitations. First, most (n=31/35, 88.6%) included articles reported on retrospective studies. In addition, most studies lacked detailed data on patient demographics, and some did not specify the type of diabetes. Furthermore, the meta-analyses were limited by data reported in studies and information solicited from corresponding authors. It is also important to note that the studies did not provide detailed information on the variant underlying COVID-19. This data may also be impacted by the strain of COVID-19, recent literature reports a decline in the number of individuals with DM with evolving variants (71). Additionally, patient outcomes are also contingent upon the care and inpatient management. The findings on insulin and patient outcomes should be interpreted with caution. For example, many patients are often taken off their outpatient medications and placed on insulin during hospitalization in line with current American Diabetes Association (ADA) recommendations (72). The ADA guidelines generally recommend insulin as the preferred treatment during inpatient hospital admission, with resumption of outpatient medications following discharge (72). Notably, patients on insulin may have more severe DM or require tighter glycemic control during acute illness (e.g., COVID-19). As the pandemic progressed, understanding of the sequelae grew and clinicians gained valuable experience providing improved management. Thus, as clinical practice advanced, outcomes were likely affected. So, comparing outcome data from early days of the pandemic (i.e., Wuhan strain) with data collected on hospitalizations secondary to subsequent variants (i.e., Delta, Omicron variants) is challenging. Moreover, most studies were from China, so further research is warranted to assess patient outcomes across varied health systems as diabetes management may vary across countries and health finance structures.

Synthesizing the findings from the systematic review and meta-analyses reveals that metformin use and DPP-4 inhibitor use are associated with decreased mortality in individuals with DM who were hospitalized for COVID-19. Data also suggest that insulin treatment is associated with increased mortality rate. However, our findings may relate to other patient-level factors including age, comorbidities, SARS-CoV-2 variant, severity of COVID-19, and duration of diabetes. COVID-19 induces a potent inflammatory response. It is plausible that the anti-inflammatory properties of metformin may confer a protective role for patients with DM who are hospitalized for COVID-19. The beneficial effects of metformin and DPP-4 inhibitors suggest that patients who are established on these antidiabetic agents should remain on their current treatment (i.e., not changed to insulin) if hospitalized for COVID-19. Given the changing landscape of COVID-19 and the emergence of new variants, further studies and randomized control trials are needed to confirm and validate these findings.