New onset of type 1 and type 2 diabetes post-COVID-19 infection: a systematic review

ABSTRACT

COVID-19 may primarily cause respiratory symptoms but can lead to long-term effects known as long COVID. COVID-19-induced diabetes mellitus was reported in many patients which shares characteristics of types 1 and 2 (T1DM and T2DM). This study aims to identify and analyse the reported cases of new onset diabetes post-COVID-19 infection. Several databases were used to conduct a comprehensive literature search to target studies reporting cases of T1DM or T2DM post-COVID-19 infection. Screening, data extraction, and cross-checking were performed by two independent reviewers. Only 43 studies met our inclusion criteria. Our results revealed that the overall prevalence of new onset diabetes post-COVID-19 was 1.37% with higher prevalence for T2DM (0.84%) as compared to T1DM (0.017%) while the type of diabetes was not reported in 0.51% of the cases. Several risk factors for developing diabetes post-COVID-19 infection were identified including the type of SARS-CoV-2 variant, age, comorbidities, and the vaccination status. The direct viral attack of the pancreatic beta cells as well as inflammation and the anti-inflammatory corticosteroids were proposed as possible mechanisms of the COVID-19 induced diabetes. A multidisciplinary approach involving endocrinologists, primary care physicians, and infectious disease specialists should be implemented in the management of post-COVID patients to address both the acute and long-term complications, including metabolic changes and risk of diabetes.

Introduction

Coronavirus disease 2019 (COVID-19), the infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was first detected in December of 2019 and spread quickly around the world – with the World Health Organization (WHO) officially declaring the disease a pandemic on the 11th of March 2020 [1]. The most common symptoms of COVID-19 are respiratory and resembling that of a cold or flu, but the severity of those symptoms varies significantly. Those infected with COVID-19, including those with no or minor symptoms, can then develop a variety of conditions post-COVID, known as long COVID [2–4].

Long COVID is one of the many names given to the symptoms or conditions that some patients experience or are diagnosed with a few months to a year, or more, after their COVID-19 diagnosis. Other names include chronic COVID, long-haul COVID, and post-acute sequelae of SARS-CoV-2 (PASC) [4]. PASC include cerebrovascular, cardiovascular, and thrombotic disease, and diabetes, among others [3].

Diabetes mellitus (DM) is a rampant disease in which the diagnosed is unable to control their blood levels of glucose. With two primary subtypes, type 1 (T1DM) and type 2 (T2DM), it is still yet a leading cause of morbidity and mortality with seemingly increasing prevalence [5]. The two types of diabetes are metabolic disorders that come from the inability to secrete (T1DM) or react to (T2DM) insulin. T1DM is characterized by the eradication of the beta cells in the islets of Langerhans in the pancreas, typically developing from an autoimmune process, that leads to a non-existent or very low amount of insulin. T2DM is characterized by a more gradual onset where the diagnosed becomes less sensitive and more resistant to insulin. This resistance while multifactorial regularly results from obesity and aging [6].

COVID-19 induced diabetes seems to fit both subtypes, with evidence of decreased insulin release and increased insulin resistance – warranting a look into the possible molecular pathogenesis and mechanisms of the disease. This systematic review investigates COVID-19 induced diabetes and the available evidence in the literature and the possible mechanisms of this sequela.

Methods

The preferred reporting items for systematic reviews and metanalysis (PRISMA) statement were used to develop the protocol of this systematic review [7].

Information sources and search strategy

This study is part of a project investigating a wide range of long-term and severe complications of COVID-19. A comprehensive search that prioritized sensitivity to retrieve all relevant studies was conducted. The following databases were searched in October 2023: PubMed, Medline (Ovid, 1946–Current), Embase (Ovid, 1974–2021), Scopus, Web of Science, Science Direct, and Cochrane Library. The search was designed around keywords and controlled vocabulary that focused on “Long Covid” and variants (see Appendix I for full search details). No language or date restrictions were used. All database search results were imported into EndNote (version 19) and exported to Covidence, where duplicates were removed prior to initial screening.

Eligibility criteria

No restrictions were made based on country, age, or gender. Duplicates were removed and any articles without primary data, such as review articles, or were not in English were excluded. Only peer reviewed full articles were included, while conference abstracts and pre-prints were excluded. During the full text screening, any studies that reported new onset diabetes post-COVID-19 infection were included if a clear diagnosis was reported. We only included severe and/or long-term complications. The inclusion criteria related to this point included any patients who developed diabetes (type I or 2) after recovering from COVID-19. If diabetes was diagnosed after at least a month after COVID-19 diagnosis or if the study reports that anti-SARS-CoV-2 immunoglobulin-G (IgG) but not IgM antibodies were detected, the study was included. The studies that reported diabetes diagnosis during the active COVID-19 infection were included only if the patient did not recover during the study time frame. Any cases of new onset diabetes that were diagnosed during the active infection of COVID-19 and recovered within less than 12 weeks were excluded.

Study selection and data collection

Title and abstract screening and full text screening were conducted by two independent reviewers using Covidence and disagreements were resolved by consensus. Similarly, data was extracted and crosschecked by two independent reviewers.

Data items

Demographic and clinical data, including age, sex, comorbidities, treatments, and outcomes, were collected. Continuous variables were expressed as mean ± standard deviation or range of results. Categorical variables were expressed as percentages.

Risk of bias and quality assessment

Different methods were used to assess the studies depending on the type of study. The Newcastle-Ottawa Quality Assessment Scale was used to assess the cohort studies [8]. The scale developed by Murad et al. [9] was used to assess the case reports and case series. Quality assessment was conducted and crosschecked by two independent reviewers.

Data analysis

The data was split into three groups, T1DM, T2DM, and diabetes with type not reported. The number of patients in the study were split into four categories: total number of COVID-19 patients, total number of patients who developed T1DM, total number of patients who developed T2DM, and total number of patients who developed either T1DM or T2DM. To calculate the prevalence rate of new onset diabetes among the COVID-19 populations, the number of patients in all studies was summed excluding the studies that did not report the number of new onset diabetes post-COVID-19. This is because some studies either did not report the number of diabetic patients or reported different values such as the risk ratio (RR), odd ratio (OR), or hazard ratio (HR). The rate of total new onset diabetes, T1DM and T2DM were calculated against the total number of COVID-19 patients as previously described.

Results

Figure 1 shows the flow diagram of the study protocol. After removing the duplicates, the titles and abstracts of 38,148 studies were screened, of which 612 were selected for full text screening. Only 43 studies met our inclusion criteria. Of the 569 excluded studies, 498 were irrelevant, 29 had no primary data, 26 were not peer reviewed or were abstracts only, 12 were not in English, and 4 were duplicates. Supplementary Table S1 summarizes the demographic and clinical data of the included subjects as well as the quality assessment score for each study [10–52].

Figure 1.

Screening and study selection protocol.

Types of studies and demographic data

Of the 43 included studies, 1 was a case report, 2 were case series, and 40 were cohort studies. Among the 43 studies, 13 were from the United States of America (USA), 10 from India, 2 from Italy, 2 from the United Kingdom (UK), 2 from Bangladesh, 2 from Israel, 1 each from Bosnia and Herzegovina, Brazil, China, Croatia, Hong Kong, Jordan, Kazakhstan, Mexico, Poland, South Korea, Spain and 1 from “USA, France, Germany, Singapore and Italy”.

The total number of patients reported by the included studies was 8,456,639 of which 47.72% were males excluding 1720 (0.02%) patients whose gender was not reported. Furthermore, the age of the included patients ranged from 0 to 101.6 years (see Supplementary Table S1).

Clinical data

Not all studies reported the number of COVID-19 patients who developed diabetes. Eight studies did not report the numbers of the new onset diabetes cases or reported different formats such as HR, OR, or RR. The other 35 studies reported 60,189 cases of new onset diabetes post-COVID-19 out of 4,395,528 COVID-19 patients (1.37%) (Figure 2a). Of the 60,189 reported patients who developed diabetes post-COVID-19, 3102 were prediabetic as reported by Xu et al. [17] (Figure 2b).

Figure 2.

Summary of the number of studies and number of COVID-19 patients who developed type 1 or type 2 diabetes (T1D or T2D) post-COVID-19. (a) Total number of patients who developed any type of diabetes post-COVID-19 out of the total COVID-19 patients as reported by the included studies (1.37%). (b) Total number of patients who developed diabetes post-COVID-19 with no history of diabetes or were prediabetic. (c) Number of studies that reported T1D, T2D or any type of diabetes post-COVID-19. d: Number of patients who developed T1D, T2D or any type of diabetes. All numbers are based on the studies that reported the number of patients as some studies reported cases of post-COVID-19 diabetes without reporting the numbers.

New onset T1DM was reported by 4 studies in 753 patients (see Supplementary Table S2). Table 1 presents data on the incidence and RR or HR of new onset T1DM in individuals post-COVID-19 compared to non-COVID control groups. New onset T2DM, was reported by 16 studies in 37,175 patients (3102 were prediabetic [16]) excluding 3 studies which did not report the numbers [16, 18, 22, 36] (see Supplementary Table S3). Table 2 details the incidence and RR, HR, and OR of new-onset T2DM in individuals post-COVID-19 compared to non-COVID controls. The type of diabetes was not reported by 28 studies which reported 22,261 patients who developed new onset diabetes post-COVID-19 excluding 6 studies which did not report the numbers [11, 24, 33–35, 41] (see Supplementary Table S4). Table 3 illustrates the incidence of new-onset T1DM or T2DM, following COVID-19 infection compared to non-COVID control groups. Figure 2(c) summarizes the number of studies that reported T1DM (4 studies), T2DM (16 studies), or any type of diabetes (28 studies). Furthermore, Figure 2(d) compares the reported number of patients under each category revealing the lowest number for T1DM (753 patients), 37,175 patients for T2DM and 22,261 patients for any type of diabetes.

Table 1.

Rates of new onset of type 1 diabetes mellitus (T1DM) post-COVID-19 infection as compared to a control non-COVID group.

Author	Rate for COVID group	Rate for control group	p-value/HR, RR
Kompaniyets et al. [38]	Incidence per 100,000 person-years: 122 With no previous symptoms: 349/396,336 (0.09%)	Incidence per 100,000 person-years: 112 With no previous symptoms: 641/792,672 (0.08%)	*HR: 1.10 (95% CI: 0.96–1.25)
Zisis et al. [45]	After 3 months: 0.06 per 1000	After 3 months: 0.01 per 1000	After 3 months: *RR: 4.44 (95% CI: 3.10–6.38)
	After 6 months: 0.1 per 1000	After 6 months: 0.02 per 1000	After 6 months: *RR: 4.51 (95% CI: 3.37–6.04)
	After 12 months: 0.15 per 1000	After 12 months: 0.03 per 1000	After 12 months: *RR: 4.85 (95% CI: 3.82–6.16)

HR: hazard ratio, RR: relative risk.

Vaccination status was not specified for the patients with new onset diabetes post-COVID-19 by any of the studies in this table.

*Compared to controls.

Table 2.

Rates of new onset of type 2 diabetes mellitus (T2DM) post-COVID-19 infection as compared to a control non-COVID group.

Author	Rate for COVID group	Rate for control group	p-value/HR, RR
Choi et al. [12]	10,668/348180 (3.06%)	–	HR: 1.30 (95% CI: 1.27–1.33)
Xu et al. [17]	In-hospital DM: 292/1378 (21.19%) Persistent DM: 128/868 (14.75%)	In-hospital DM: 249/4134 (6.02%) Persistent DM: 204/2715 (7.51%)	*p < 0.001 *p < 0.001
	Persistent DM: 39/939 (4.15%)	Persistent DM: 55/1342 (4.10%)	–
Estiri et al. [18]	–		^3–6 months: *OR: 1.48 (95% CI: 1.19–1.85)
	–	–	^6–9 months: *OR: 1.65 (95% CI: 1.12–2.41)
	–	–	^Under 65 years 3–6 months: *OR: 1.57 (95% CI: 1.19–2.06)
	–	–	^Under 65 years male 6–9 months: *OR: 2.34 (95% CI: 1.53–3.58)
Zhang et al. [19]	–	–	^30–89 days: *RR: 1.26 (95% CI: 1.16–1.36)
	–	–	^90 + days: *RR: 1.11 (95% CI: 1.02–1.21)
Sharma et al. [25]	326/2433 (13.4%)	–	*OR (95% CI): 1.4 p < 0.001
Xiong et al. [26]	2109/145,199 (1.45%)	1774/145,199 (1.22%)	*HR: 1.226 (95% CI: 1.151–1.306)
Kompaniyets et al. [38]	Incidence per 100000 person-years: 255 With no previous symptoms: 729/396,336 (0.18%)	Incidence per 100000 person-years: 212 With no previous symptoms: 1210/792,672 (0.15%)	*HR: 1.19 (95% CI: 1.09–1.31) p < 0.05
Zisis et al. [45]	^3months: 2.39 per 1000	^3months: 0.34 per 1000	^3months: *RR: 7.07 (95% CI: 6.59–7.59)
	^6 months: 4.28 per 1000	^6months: 0.49 per 1000	^6months: *RR: 8.81 (95% CI: 8.31–9.34)
	^12months: 6.7 3 per 1000	^12months: 0.74 per 1000	^12months: *RR: 9.12 (95% CI: 8.70–9.56)
Lu et al. [52]	22.6% of difference (95% CI: 0.18–0.20)	3.3% (influenza patients control)	–

HR: hazard ratio, RR: relative risk.

*Compared to controls.

^After COVID-19 diagnosis.

Vaccination status was not specified for the patients with new onset diabetes post-COVID-19 by any of the studies.

Table 3.

Rates of new onset of type 1 or type 2 diabetes mellitus (T1DM) or (T2DM) post-COVID-19 infection as compared to a control non-COVID group.

Author	Rate for COVID group	Rate for control group	p-value/HR, RR
Bowe et al. [11]	9.56 per 1000 persons excess burden of diabetes	–	*HR: 1.36 (95% CI: 1.31–1.40)
	Two infections: 28.92 per 1000 persons excess burden of diabetes	–	*HR: 2.09 (95% CI: 1.96–2.23)
	three infections: 50.53 per 1000 persons excess burden of diabetes	–	*HR: 2.93 (95% CI: 2.18–3.93)
Richard et al. [24]	–	–	*RR: 1.46 (95% CI: 1.00–2.13)
Ayoubkhani et al. [33]	rate per 1000 patient years: 28.7 (26.0, 31.7)	–	*RR: 1.5 (95% CI: 1.4–1.6)
Horberg et al. [35]	–	–	**30–120 days: *RR: 1.20 (95% CI: 1.03–1.38)
	–	–	**0–30 days (persisting 30–120 days): *RR: 1.96 (95% CI: 1.50–2.55)
Rezel-Potts et al. [41]	**Up to 4 weeks: 23.79 (21.57, 26.18) per 100,000 patient-weeks	–	**Up to 4 weeks: *RR: 1.81 (95% CI: 1.51–2.19)
	**Up to 4 weeks: 19.54 (18.10, 21.06) per 100,000 patient-weeks	–	**5–12 weeks: *RR: 1.27 (95% CI: 1.11–1.46)
	**13–52 weeks: 19.57 (18.91, 20.26) per 100,000 patient-weeks	–	**13–52 weeks: *RR: 1.07 (95% CI: 0.99–1.16)
Khullar et al. [43]	Hospitalized: 891/13106 (6.8%)	Hospitalized: 3062/71222 (4.3%)	–
	Non-hospitalized: 1132/49233 (2.3%)	Non-hospitalized: 3710/176659 (2.1%)	–
Xie et al. [44]	48.38 (47.04, 49.76) per 1000	–	*HR: 1.40 (95% CI: 1.36–1.44)
Zisis et al. [45]	After 3 months: 2.48 per 1000	After 3 months: 0.36 per 100	After 3 months: *RR: 6.99 (95% CI: 6.52–7.49)
	After 6 months: 4.41 per 1000	After 6 months: 0.51 per 1000	After 6 months: *RR: 8.64 (95% CI: 8.16–9.14)
	After 12 months: 6.92 per 1000	After 12 months: 0.77 per 1000	After 12 months: *RR: 8.94 (95% CI: 8.54–9.36)
Reges et al. [49]	1145/157,936 (0.72%)	1013/157,936 (0.64%)	–
Jennifer et al. [51]	All: 237/71683 (0.33%)	All: 186/71683 (0.26%)	*p = 0.028
	non-hospitalized: 191/68097 (0.28%)	non-hospitalized: 178/71339 (0.25%)	*p = 0.33
	Hospitalized: 42/3586 (1.17%)	Hospitalized: 7/344 (2.03%)	*p = 0.168

HR: hazard ratio, RR: relative risk.

*compared to non-infected control/ before infection.

^ after discharge from COVID-19 hospitalization.

**after COVID infection/diagnosis.

Discussion

Since the relationship between COVID-19 and diabetes has garnered significant attention due to the observed increase in diabetes incidence among patients post-infection, we aim to explore the multifaceted factors contributing to this phenomenon and the implications for clinical practice.

Prevalence and risk of developing T1DM or T2DM post-COVID-19

Our compiled data revealed that out of 4,395,528 COVID-19 patients, 60,189 cases of new onset diabetes (1.37%) were reported by the 35 included studies (which reported the number of new onset diabetes post-COVID-19). This rate reflects the incidents of newly diagnosed diabetes among a specific population (previously had COVID-19) rather than the overall prevalence of diabetes within the COVID-19 population. This is because all cases with a history of diabetes were excluded from the study. The duration of the incidence of our included cases ranged from 1 to 12 months. The actual rate could be higher because many studies reported the cases of new onset diabetes only for those who returned to the hospital for either re-admission or follow-up. This means that some of the previously infected patients (out of the 4,395,528 COVID-19 patients who were used to calculate the rates) may have developed new onset diabetes but never returned to the same hospital/clinic. However, this rate is still higher than the estimated rate of new onset diabetes in a general population. For example, among adults aged 18 years or older in the USA, the crude estimates for 2021 were 1.2 million new cases of diabetes (0.59%) [53]. While this numbers was estimated during the pandemic, it applies to the entire population rather than only the previously infected people which is the focus of our study. Therefore, our calculated rate of new cases of diabetes in a population with a history of COVID-19 is still higher than the estimated numbers in the USA in 2021.

New onset T1DM post-COVID-19 infection

Of a total of 4,395,528 COVID-19 patients reported by the included studies, the incidence of new onset T1DM diabetes was found to be 0.017%. Furthermore, Zisis et al. [45] and Kompaniyets et al. [38] both identified an increased risk of developing T1DM as a long-term consequence of COVID-19 infection. Kompaniyets et al. observed this risk in a younger population with a mean age of 12 (range: 0–17), aligning with the typical age group for T1DM onset. Kompaniyets et al. reported an HR (95% Confidence Interval) of 1.10 (95% CI: 0.96–1.25) for the development of T1DM in COVID-19 patients compared to controls. This analysis, derived from a large cohort, included an incidence rate of 122 cases per 100,000 person-years in the COVID-19 group compared to 112 cases per 100,000 person-years in the control group. Although the HR suggests a slight increase in risk, the confidence interval crossing 1 indicates that this result is not statistically significant. In contrast, Zisis et al. found the increased risk of T1DM in an older population, with a mean age of 44.4 (±17.7).

The effect of time post-COVID-19 was also investigated by Zisis et al. [45] who reported that the RR of developing T1DM rose over time after COVID-19 infection, with a relative risk of 4.44 (95% CI: 3.10–6.38) at 3 months, 4.51 (3.37–6.04) at 6 months, and 4.85 (3.82–6.16) at 12 months. These results indicate a substantial and persistent elevation in T1DM risk over time, with statistical significance across all time intervals. The incidence rates of T1DM in the COVID-19 group were 0.06, 0.1, and 0.15 per 1000 person-years at 3, 6, and 12 months, respectively, demonstrating a clear upward trend. This suggests that COVID-19 may act as a long-term trigger for autoimmune responses leading to T1DM. This trend was similar to their findings on T2DM, though the peak relative risk for T1DM was lower, likely reflecting the higher prevalence of new onset T2DM compared to T1DM.

New onset T2DM post-COVID-19 infection

Based on the included studies, new onset T2DM was found to be more prevalent than T1DM as 0.84% of the COVID-19 patients included in this study developed T2DM. Choi et al. [12], Estiri et al. [18], Kompaniyets et al. [38], Sharma et al. [25], Xiong et al. [26], Xu et al. [17], J. Zhang et al. [22], and Zisis et al. [45] all have found a higher risk of developing T2DM between 1 and 12 months after a COVID-19 infection. For example, Choi et al. reported an HR of 1.30 (95% CI: 1.27–1.33), indicating a 30% increased risk of T2DM post-COVID-19. This robust finding reflects the consistent association found in larger cohort studies. Xiong et al. [26] documented a T2DM rate of 1.45% in the COVID-19 group compared to 1.22% in controls, with an HR of 1.226 (95% CI: 1.151–1.306). Kompaniyets et al. [38] found an incidence of 255 per 100,000 person-years in COVID-19 patients compared to 212 per 100,000 person-years in controls, yielding an HR of 1.19 (95% CI: 1.09–1.31). These findings suggest a modest but statistically significant increase in diabetes risk post-COVID-19.

For in-hospital diabetes, the incidence was 21.19% in COVID-19 patients compared to 6.02% in controls, with a highly significant p-value of <0.001 [17]. Persistent diabetes rates were 14.75% in the COVID-19 group versus 7.51% in controls, also with p < 0.001. This highlights the heightened risk both during hospitalization and in long-term follow-up. This suggests that severe COVID-19 infections, likely due to increased inflammatory and metabolic stress, are more likely to lead to delayed or incomplete resolution of insulin resistance. Interestingly, Lu et al. [52] reported that T1DM was diagnosed in 22.6% of patients with COVID-19 compared to only 3.3% of patients with influenza (95% CI of difference: 0.18–0.20), further highlighting the unique metabolic impact of COVID-19. This study suggests that the observed risk cannot be generalized to other viral illnesses.

The effect of time post-COVID-19 infection was investigated by Zisis et al. [45] who examined T2DM risk over time, reporting rates of 2.39 per 1,000 at 3 months (RR: 7.07, 95% CI: 6.59–7.59), 4.28 per 1000 at 6 months (RR: 8.81, 95% CI: 8.31–9.34), and 6.73 per 1000 at 12 months (RR: 9.12, 95% CI: 8.70–9.56). These findings indicate a pronounced, persistent risk increase over time. Similarly, Estiri et al. [18] also examined risks over different intervals, reporting OR of 1.48 (95% CI: 1.19–1.85) at 3–6 months and 1.65 (95% CI: 1.12–2.41) at 6–9 months. Subgroup analyses revealed particularly elevated risks for younger patients under 65, with ORs of 1.57 (95% CI: 1.19–2.06) at 3–6 months and 2.34 (95% CI: 1.53–3.58) at 6–9 months for males under 65. These findings underscore the long-term metabolic impact of COVID-19, particularly in specific demographic groups. Conversely, Zhang et al. [22] reported time-dependent risks, finding an RR of 1.26 (95% CI: 1.16–1.36) at 30–89 days post-COVID-19, decreasing to 1.11 (95% CI: 1.02–1.21) at 90+ days. This suggests a tapering of risk over time, though the overall association remains significant.

New onset T1DM or T2DM diabetes post-COVID-19

The type of diabetes was not reported by 28 studies in 0.51% of the COVID-19 patients. Bhandari et al. [10], Bowe et al. [11], Richard et al. [24], Ayoubkhani et al. [33], Horberg et al. [35], Rezel-Potts et al. [41], and Xie et al. [44] all have reported a higher risk of developing diabetes, with elevated RR or HR. For instance, Xie et al. [44] recorded a diabetes incidence rate of 48.38 per 1000 persons, with an HR of 1.40 (95% CI: 1.36–1.44). This robust finding aligns with other studies, confirming an increased risk of diabetes following COVID-19 infection. Similarly, Ayoubkhani et al. [33] reported a rate of 28.7 per 1000 patient years in the COVID-19 group, with an RR of 1.50 (95% CI: 1.40–1.60) compared to controls. This highlights a significant increase in diabetes risk over a prolonged follow-up period. Reges et al. [49] observed a diabetes incidence of 0.72% in the COVID-19 group compared to 0.64% in controls. While the absolute differences are small, they represent a statistically significant increase in diabetes risk. In hospitalized patients, Khullar et al. [43] observed a diabetes rate of 6.8% in the COVID-19 group compared to 4.3% in controls, reflecting a substantial increase in risk. Among non-hospitalized patients, the incidence was lower but still elevated, at 2.3% versus 2.1% in controls. In less severe cases, Jennifer et al. [51] found a small but statistically significant increase in diabetes risk among the COVID-19 patients (0.33% vs. 0.26%, p = 0.028). However, for hospitalized and non-hospitalized patients, the increase was not statistically significant (1.17% vs. 2.03%, p = 0.168 and 0.28% vs. 0.25%, p = 0.33 respectively).

The effect of time post-COVID-19 was investigated by Rezel-Potts et al. [41] who divided their findings into time intervals: up to 4 weeks post-infection, the diabetes incidence was 23.79 per 100,000 patient-weeks (RR: 1.81, 95% CI: 1.51–2.19). Between 5 and 12 weeks, the incidence slightly decreased to 19.54 per 100,000 patient-weeks (RR: 1.27, 95% CI: 1.11–1.46). At 13–52 weeks, the incidence remained steady at 19.57 per 100,000 patient-weeks, but the RR (1.07, 95% CI: 0.99–1.16) indicated a non-significant trend. This time-dependent risk pattern suggests that diabetes risk is most pronounced shortly after COVID-19 infection and gradually stabilizes over time which contradicts the findings of Zisis et al. [45] Similarly, Horberg et al. [35] found a lower risk 1.20 (95% CI: 1.03–1.38) between 4 and 16 weeks. Rezel-Potts et al. [41] attributed this declining risk to earlier diagnosis of underlying diabetes due to hospitalization and more frequent check-ups in COVID-19 patients.

Risk factors for new onset diabetes post-COVID-19

While the previous section discussed the data based on the type of diabetes, this section comprehensively looks at any risk factors that were found to be associated with developing any type of diabetes post-COVID-19 infection.

Variants and severity

Several studies examined the effect of different factors like different COVID-19 variants. Xiong et al. [26] found that the Omicron variant had a lower risk of developing T2DM compared to non-Omicron variants, HR 1.209 (95% CI: 1.134–1.289) vs. non-Omicron, HR 1.871 (95% CI: 1.352–2.589), but it still posed an increased risk overall. Likewise, Bhandari et al. [10] found the Delta variant to be linked with the highest risk of diabetes (9.21%, p < 0.001) compared to Omicron (4.54%, p = 0.1141), and Alpha (6.19%, p = 0.0028). This difference between variants is further shown with the Omicron variant having a lower risk when assessed alone by Xiong et al. [26] who reported HR 1.23 (95% CI: 1.51–1.31). Though, this could be interpreted by the fact that the Omicron variant is associated with more attenuated disease overall, and a milder immune response [54,55]. Interestingly, however, an extensive study showed that the Omicron variant was associated with an increased risk of diabetes HR 1.1 (95% CI: 1.04–1.17) and 1.12 (95% CI: 1.06–1.18) at 3 and 6 months respectively, whereas the Delta variant was associated with a decreased risk 0.88 (95% CI: 0.82–0.94) and 0.95 (95% CI: 0.90–0.99) respectively. The opposite was observed after 12 months whereas a higher risk was associated with the Delta variant 1.84 (95% CI: 1.56–2.17) versus 1.12 (1.06–1.18) for the Omicron [45]. This contrasting finding may be explained by Omicron's ability to spread rapidly, with efficient cell-to-cell transmission and immune escape, leading to a different risk profile for diabetes and other post-infection conditions [54].

The risk of diabetes post-COVID was also found to increase in a stepwise fashion based on the level of care required during the acute phase of infection, with ICU patients facing the highest risk, 123.48 (95% CI: 107.31–141.89) burden per 1000 people at 12 months, followed by hospitalized patients 91.35 (95% CI: 83.80–99.54) as compared to the non-hospitalized patients 48.38 (95% CI: 47.04 –49.76) [44]. Furthermore, ICU patients with COVID-19 were found to have nearly double the risk of developing diabetes compared to non-ICU patients [31]. This association is further supported by studies linking severe COVID-19 cases to elevated haemoglobin A1c (HbA1c) and serum glucose levels during and after hospitalization [13,56]. This correlation is further highlighted in other studies such as Reges et al. [49] who reported the highest risk in the acute phase of the hospitalized patients HR 2.47 (95% CI: 1.86–3.29) [49]. Similarly, Xu et al. [17] and Lu et al. [52] reported rates of 21.9% and 16.7% of new diabetes diagnosis in the hospitalized COVID-19 patients versus 4.15% and 7.3% in the non-hospitalized patients. Another study then highlighted that patients with COVID-19 reinfection had an increased risk of developing diabetes compared to those without reinfection [11]. Bowe et al. [11] reported an excess burden of diabetes post-COVID-19, with rates of 9.56 per 1000 persons for those with one infection, increasing to 28.92 per 1000 persons for individuals with two infections and 50.53 per 1000 persons for three infections. Corresponding HRs increased with the number of infections: 1.36 (95% CI: 1.31–1.40) for one infection, 2.09 (95% CI: 1.96–2.23) for two infections, and 2.93 (95% CI: 2.18–3.93) for three infections. These findings suggest a dose–response relationship, where repeated COVID-19 infections exacerbate diabetes risk.

Many studies highlight that corticosteroid treatment of moderate to severe COVID-19 cases could have played a role in the pathogenesis of new onset diabetes [30,32]. The study by Xu et al. [17] showed that the corticosteroid treatment in patients with pre-diabetes was associated with a significantly higher risk of developing in-hospital diabetes mellitus (I-DM), with an HR of 2.88 (95% CI: 2.2–3.8, p < 0.005). This is indeed an important consideration to keep in mind, especially when discussing possible disease mechanisms. However, sensitivity analysis revealed that corticosteroid use had no significant effect on diabetes incidence, suggesting results consistent with leading hypotheses proposing direct COVID-19 impacts on glucose metabolism and pancreatic function [35].

Age, comorbidities, and other demographics

Estiri et al. [18] examined risks over different intervals whereas subgroup analyses revealed particularly elevated risks for younger patients under 65, with ORs of 1.57 (95% CI: 1.19–2.06) at 3–6 months and 2.34 (95% CI: 1.53–3.58) at 6–9 months for males under 65. These findings underscore the long-term metabolic impact of COVID-19, particularly in specific demographic groups which is possibly due to a stronger immune response that may drive some post-COVID sequelae [18].

A large retrospective study by Zisis et al. [45] revealed a list of comorbidities as risk factors for developing diabetes after COVID-19 infection. These included metabolic and cardiovascular conditions such as hyperlipidaemia, hypertension, and coronary artery disease; pre-diabetes; having a family history of diabetes; autoimmune conditions like rheumatoid arthritis and systemic lupus erythematosus (SLE); vitamin D deficiency; the use of medications like statins, remdesivir, glucocorticoids, and antipsychotics; as well as being over 65, and having a body mass index (BMI) over 30.

Additionally, racial and ethnic disparities were evident in the incidence of new-onset diabetes, with non-Hispanic Black and Hispanic patients showing significantly higher rates (8.0% and 8.4%, respectively) compared to White patients (4.6%) [43]. This disparity may be attributed to the increased severity of COVID-19 infections in these populations, lower vaccination rates, and pre-existing socioeconomic and health disparities. As such, individuals with limited access to healthcare are likely to experience more severe infections and subsequent long-term effects, including new-onset diabetes.

COVID-19 vaccination status

Some studies have explored the impact of COVID-19 vaccinations on the risk of developing new-onset diabetes as a long-term sequelae of COVID-19 infection, yielding varied conclusions. Jennifer et al. [51] compared the risk of developing new-onset diabetes in individuals reinfected with COVID-19 (infected twice) to those infected only once, across different vaccination statuses. They found that unvaccinated individuals had a HR of 1.51 (95% CI: 1.25–1.82) for developing diabetes after reinfection, while those who received one vaccine had a slightly higher HR of 1.58 (95% CI: 1.30–1.90). Interestingly, individuals with two or more vaccinations had a lower HR of 1.45 (95% CI: 1.18–1.79), suggesting a potential protective effect from multiple doses, although the differences between these groups were not statistically significant. On the other hand, Bowe et al. [11] identified a significantly lower incidence of new-onset diabetes post-COVID-19 infection among vaccinated individuals (0.15%) compared to unvaccinated individuals (0.33%) (p < 0.01). This supports the idea that vaccination may reduce the risk of developing diabetes after infection. Together, these studies suggest that vaccination, particularly multiple doses, may help mitigate the risk of developing new onset diabetes following COVID-19 reinfection. It was not possible to reach a definitive conclusion about the effect of vaccination on the risk of new-onset diabetes following COVID-19 infection. This limitation is due to insufficient data in the included studies. Only a few studies reported the vaccination status of patients, and not all examined the correlation between vaccination and new-onset diabetes, often due to their study design. To draw a robust conclusion, a comparison of the rates of new-onset diabetes across cohorts with varying vaccination statuses would be necessary. However, this was not the primary objective of our study, which was not designed with inclusion and exclusion criteria specifically aimed at addressing this question. Our included studies primarily focused on comparing COVID-19 cohorts with non-COVID-19 cohorts. As a result, a dedicated study should be designed to specifically address this topic, given the conflicting findings in the current literature. While some studies suggest that vaccination reduces the risk of developing diabetes following COVID-19 infection, there have also been reports of new-onset diabetes occurring after COVID-19 vaccination. This highlights the need for further investigation to clarify the role of COVID-19 vaccination [57] in triggering or protecting against diabetes. A more comprehensive and targeted approach is required to better understand the relationship between COVID-19 vaccination and diabetes risk, considering the complexity and variability of the reported outcomes.

Mechanisms of COVID-19 induced new-onset diabetes

Most studies either report T2DM, or do not report the type of diabetes assessed, possibly fitting in with mechanistic explanations centring on inflammation and insulin resistance. However, a robust body of research explicitly identifies new-onset T1DM. Case series document DKA-like hospitalizations with confirmed autoantibodies [21] while larger retrospective cohort studies corroborate the occurrence of T1DM post-COVID, alongside T2DM cases [38,45]. These studies thus warrant investigation into contrasting mechanisms of COVID-19 induced disease involving pancreatic damage and reduced insulin production. Research indicates that viruses can induce pancreatic β-cell death or damage through various mechanisms. These include direct cell lysis, programmed cell death, inflammation leading to bystander damage or activation, autoimmunity targeting β-cells, molecular mimicry, transdifferentiation and dedifferentiation [58]. Figure 3 summarizes some possible mechanisms that may contribute in the COVID-19 induced T1DM or T2DM.

Figure 3.

Possible mechanisms for post-COVID-19 new onset diabetes.

Role of angiotensin-converting enzyme 2

To invade the host cell, SARS-CoV-2 primarily binds angiotensin-converting enzyme 2 (ACE2) with its surface spike protein S, initiating a sequence that allows it to infect cells. ACE2 is found in the lungs, as well as other key metabolic tissues that include the liver, kidney, and pancreas [59]. Furthermore, ACE2 expression has been shown to be elevated in the pancreas compared to the lungs, suggesting its potential involvement in pancreatic pathogenesis and COVID-19 related disease mechanisms [60].

ACE2, like its homolog ACE, plays a large role in renin-angiotensin-system (RAS) signalling. While ACE cleaves Angiotensin-I (Ang I) into Angiotensin-II (Ang II) ACE2 cleaves Ang II into Angiotensin 1–7 (ANG 1–7) [61]. By degrading Ang II, ACE2 can inhibit RAS hyperactivity and its harmful effects of inflammation, vasoconstriction, and reactive oxygen species (ROS) on the cellular level, and hyperglycaemia, hypertension, and cardiac dysfunction at the patient level. Specifically in the pancreatic islets, Ang II has been shown to increase inflammation, islet cell apoptosis, and decrease blood flow, leading to reduced insulin secretion [61]. Therefore, SARS-CoV-2 cell entry may lead to the reduced expression of ACE2 and predispose to RAS-mediated inflammation and damage. Accordingly, ACE2 seems protective of the pancreas, although it also facilitates SARS-CoV-2 entry into pancreatic islet cells – seemingly playing a role both in beta-cell protection and dysfunction. This makes ACE2 expression a big suspect for a key player in the pathogenesis of COVID-19 induced diabetes, but also possibly a key target for therapeutics.

Direct beta-cell invasion and injury

After hijacking ACE2 and invading pancreatic islet beta cells, it is possible that SARS-CoV-2 can directly induce cell injury. SARS-CoV-2-specific viral RNA was detected in the post-mortem pancreatic tissues, along with immature insulin granules and proinsulin. This finding suggests disruptions in beta-cell proinsulin processing, beta-cell degeneration, and hyperstimulation [62]. Though it is not traditionally classified as a cytolytic virus, SARS-CoV-2 can induce cell death through apoptosis, pyroptosis, and autophagy [63]. Steenblock et al. [64] demonstrated that in pancreatic samples from COVID-19 patients, only certain islets showed high levels of phosphorylated pseudokinase mixed lineage kinase domain-like (pMLKL) protein, a marker of necroptosis. This finding suggests that SARS-CoV-2 infection may induce necroptotic cell death in islet cells, aligning with previous reports that human coronaviruses can trigger necroptosis in host cells. Wu et al. [65] further reported that SARS-CoV-2 infection of human primary islets results in decreased insulin content and secretion, alongside an increased number of TUNEL-positive β-cells in ex vivo experiments. Additionally, phosphoproteomic mass spectrometry analysis identified activation of the c-Jun N-terminal kinase (JNK)-mitogen-activated protein kinase (MAPK) apoptosis signalling pathway as a likely mechanism underlying β-cell death following SARS-CoV-2 infection.

Inflammation and other factors leading to the indirect injury of beta cell

Another immune mechanism that could contribute to new-onset diabetes is the infamous cytokine storm seen in severe COVID-19. The surge of inflammatory cytokines, including interleukin-6 (IL-6), tumour necrosis factor-α (TNF-α), and even IL-17, directly damages beta cells, and is associated with hyperglycaemia and adipose inflammation, which contribute to insulin resistance [66–68]. In addition to the acute context of the cytokine storm, patients who have had COVID-19 were shown to have significantly different cytokine profiles months after infection, with alterations in cytokines such as fibroblast growth factor 2 (FGF-2), vascular endothelial growth factor-A (VEGF-A), epidermal growth factor (EGF), IL-12 (p70), IL-13, and IL-6. [16]. This emphasizes the intricate interplay between cytokine dynamics and diabetes progression, even in the long term. Ben Nasr et al. [62] explored the effects of SARS-CoV-2 on human pancreatic islets in 10 patients who developed hyperglycaemia after COVID-19. The patients’ serum exhibited toxicity toward human pancreatic islets which was mitigated by the application of anti-IL-1β, anti-IL-6, and anti-TNF-α, which are cytokines markedly upregulated during COVID-19. Notably, receptors for these cytokines were highly expressed on human pancreatic islets. Elevated levels of unmethylated INS DNA, a marker of beta-cell death, were observed in several patients with COVID-19, indicating ongoing cellular damage. Post-mortem pancreatic tissues analysis from hyperglycaemic COVID-19 patients revealed mild lymphocytic infiltration in the pancreatic islets and lymph nodes. These observations collectively indicate that SARS-CoV-2 may impair pancreatic islet function and survival through inflammatory mechanisms, potentially compounded by direct viral effects, contributing to the metabolic abnormalities observed in COVID-19 patients.

A link between pancreatic thrombofibrosis and new-onset diabetes in COVID-19 patients was identified by Qadir et al. [69]. Pancreatic tissue from SARS-CoV-2-infected nonhuman primates (NHPs), including African green monkeys and rhesus macaques, revealed multiple microthrombi in small veins, increased fibrosis, and endotheliitis, accompanied by elevated serum lipase levels compared to uninfected controls. Similar findings were observed in human COVID-19 patients, including those newly diagnosed with diabetes upon hospital admission. Kusmartseva et al. [70] also reported multiple thrombotic lesions in the pancreatic tissues of COVID-19 patients. Remarkably, SARS-CoV-2-infected NHPs developed diabetes 9–24 days post-inoculation, suggesting that the long-term effects of a fibrotic and thrombotic pancreas may indirectly impair β-cell function, potentially leading to late-onset diabetes in COVID-19 patients.

Transdifferentiation and dedifferentiation

Transdifferentiation refers to the process where β-cells lose their identity and acquire characteristics of other pancreatic or non-pancreatic cell types, impairing their insulin-secreting function. This phenomenon is driven by direct viral effects, inflammation, and dysregulation of cellular signalling pathways. Once inside the β-cells, the virus may disrupt transcriptional programs responsible for maintaining β-cell identity. Viral replication and stress response pathways activate cellular mechanisms that push β-cells into a dedifferentiated or transdifferentiated state. The cytokine storm induced by SARS-CoV-2, involving elevated levels of IL-6, IL-1β, and TNF-α, creates a pro-inflammatory microenvironment around β-cells. Chronic exposure to inflammatory cytokines has been shown to trigger dedifferentiation, where β-cells lose their functional phenotype and insulin-secreting capacity. Under sustained stress, β-cells may transdifferentiate into α-cells or other pancreatic endocrine cells, further reducing the pool of functional insulin-producing cells. Furthermore, the metabolic and epigenetic dysregulation common in COVID-19 patients, exacerbate oxidative stress and metabolic dysfunction in β-cells. Epigenetic changes, such as DNA methylation and histone modifications, induced by the viral infection and inflammatory environment, alter the expression of key β-cell transcription factors. This epigenetic reprogramming facilitates the conversion of β-cells to other cell types, such as glucagon-producing α-cells or even non-endocrine cells. Combined with other mechanisms like β-cell apoptosis and insulin resistance, transdifferentiation contributes to the onset of diabetes in COVID-19 patients [71–73].

Molecular mimicry and autoimmunity

SARS-Cov-2's spike protein has been found to have various mimotopes that closely resemble human antigen epitopes, and, specifically, motifs found in beta-cell proteins [66,74]. For example, a study highlighted similarities between SARS-CoV-2 spike protein, and human insulin and glutamic acid decarboxylase 65-kilodalton isoform (GAD-65) (both of which are targets of established and prevalent autoantibodies in T1DM) [75,76]. This molecular mimicry together with the hyperinflammatory environment, characteristic of severe COVID-19, driven by elevated levels of cytokines such as IL-6, IL-1β, and TNF-α, exacerbates immune dysregulation and amplifies the autoimmune response. This inflammatory milieu enhances the activity of antigen-presenting cells (APCs), which process and present viral peptides that mimic β-cell epitopes, leading to the activation of autoreactive CD4+ and CD8+ T cells. These immune cells then target β cells, resulting in apoptosis or necrosis. Additionally, the autoreactive B cells will be activated leading to the generation of autoantibody-producing plasma cells [66]. Cells of the innate immune system (macrophages, dendritic cells, and natural killer cells) can also recognize these mimotopes and antigens and contribute to beta-cell destruction through different mechanisms.

Kayhan et al. [77] examined the potential impact of SARS-CoV-2 infection on pancreatic β cells by measuring autoantibodies in 95 hospitalized COVID-19 patients. At the time of diagnosis, 4.2% tested positive for anti-islet autoantibodies, while 1.1% had anti-GAD autoantibodies, and no patients were positive for anti-insulin autoantibodies. Three months later, half of the initially positive anti-islet autoantibodies (2 out of 4) and all anti-GAD autoantibodies (1 patient) remained detectable. These findings suggest that SARS-CoV-2 may trigger an autoimmune response leading to persistent β-cell autoantibody positivity, potentially contributing to the development of autoimmune diabetes.

Contribution to insulin resistance

It follows that the inflammatory processes that follow SARS-CoV-2, alongside rampant RAS activity, can contribute majorly to insulin resistance [78]. It has been shown that the virus can also directly dysregulate metabolic profiles and induce insulin resistance [56]. In a study determining metabolic alterations in COVID-19 patients with no previous diabetes, SARS-CoV-2 altered the expression of secreted metabolic factors including myostatin, apelin, and myeloperoxidase at the transcriptional level – disrupting glucose and lipid metabolism with new-onset insulin resistance [79].

It is important to note that corticosteroid treatments warranted in moderate and severe forms of COVID-19 can also induce or contribute to hyperglycaemia and insulin resistance and are thus important to consider when discussing mechanisms of new-onset diabetes.

Quality assessment

The quality of the cohort studies, as assessed by the NOS, varied widely, with scores ranging from 3 to 9 out of 9. Thirteen studies achieved the highest score of 9, four scored 8, two scored 7 and twelve scored 6 indicating rigorous methodological design and robust data. Five studies scored 5 reflecting a moderate quality. A relatively small proportion of the studies (4 out of 40) scored between 3 (one study) and 4 (three studies) reflecting varying degrees of methodological limitations. These may include issues with comparability between groups, and the follow-up. However, the four studies which scored 3–4, had minimum effect on the overall quality of the study as their overall weight was 1125 COVID-19 and 83 patients with new onset diabetes out of the total 4,395,528 COVID-19 patients and 60,189 patients with new onset diabetes post-COVID-19 respectively. The scores of the case report and case series using Murad et al. [9] scale ranged from 3 to 4 out of a maximum of 8. However, three of the eight questions are primarily relevant to adverse drug events which do not apply to our included studies. Therefore, the three studies scored 3–4 out of a maximum of 5 indicating a moderately high quality of reporting. Regarding the heterogeneity concerns, while case reports and case series inherently lack robust methodology, their inclusion had a minimum effect on the overall quality of the study as their total weight was 6 patients with new onset diabetes out of 4,395,528 COVID-19 patients and 60,189 patients with new onset diabetes post-COVID-19.

Study limitations

This study has some limitations. One limitation was that some studies did not report the number of patients who developed diabetes post-COVID-19. Furthermore, some studies reported the total number of diabetic patients post-COVID-19 without specifying how many of them had a history of diabetes or were prediabetic. Another limitation was the unspecified type of diabetes which was reported by many studies. This made it hard to make a solid conclusion or draw a clear picture about the relative prevalence of T1DM and T2DM. However, despite of this limitation, T2DM was more prevalent based on the studies that reported the type of diabetes. Moreover, some studies reported the cases of new onset diabetes only for those who returned to the same clinic/hospital for follow-up or re-admission. This may have led to missing some of the cases who used other medical facilities to treat diabetes after recovering from COVID-19. Those patients were still counted among the 4,395,528 total COVID-19 patients against which the rates of new onset diabetes were calculated. This may lead to underestimating the rates of new onset diabetes post-COVID-19. Another factor was the vaccination status which was not reported by many studies. The relationship between vaccination and the risk of developing diabetes post-COVID-19 requires further investigation in a separate study focusing on selecting only studies that reports the vaccination status of the patients or compare new onset diabetes in vaccinated with unvaccinated cohorts.

Although a larger number of included studies may have empowered the evidence, however, many studies were excluded for being published as conference abstracts. Only peer reviewed full article journal publications were selected to maintain the quality of the included studies and to avoid duplication when the work is published in both conferences and journals.

Conclusion and recommendations

The relationship between COVID-19 and the development of T1DM and T2DM is complex and multifactorial. This study confirms an increased prevalence of diabetes following COVID-19 infection, with T2DM being more prevalent than T1DM. The risk of diabetes increases over time post-infection, with the highest prevalence observed between 3 and 12 months post-infection. Key factors influencing diabetes onset include the severity of the infection, the type of variant involved, comorbidities, and demographic factors such as age and race. Severe cases requiring hospitalization or ICU admission, as well as infection with certain COVID-19 variants like Delta and Omicron, have been associated with higher risk of diabetes. Moreover, the mechanisms behind COVID-19-induced diabetes may include direct damage of the pancreatic beta cells, inflammation, insulin resistance, and autoimmune responses. It is recommended that healthcare providers should monitor individuals post-COVID-19 for signs of new-onset diabetes, particularly those with risk factors such as hospitalization, ICU admission, severe infection, pre-existing metabolic disorders, and COVID-19 variants associated with higher risks. Given that the risk of developing diabetes increases over time, especially within the first year post-infection, regular follow-up appointments and screening (e.g. HbA1c testing) should be implemented for COVID-19 survivors, especially for those at high risk, to facilitate early detection and management. Studies suggest that COVID-19 vaccination, particularly multiple doses, may reduce the risk of diabetes onset post-infection. Vaccination campaigns should continue to be emphasized as part of broader strategies to mitigate long-term complications of COVID-19, including diabetes. Further investigation into the specific mechanisms of COVID-19-induced diabetes is needed, particularly around the autoimmune responses, and the role of inflammation. A multidisciplinary approach involving endocrinologists, primary care physicians, and infectious disease specialists should be implemented in the management of post-COVID patients to address both the acute and long-term complications, including metabolic changes and risk of diabetes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All extracted data for this systematic review are submitted as supplementary materials.

Supplemental Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/22221751.2025.2492211.