Long COVID Endocrine and Metabolic Sequelae: Thyroid Autoimmunity and Dysglycemia Four Years After SARS-CoV-2 Infection

Abstract

Background: Endocrine disturbances are increasingly recognized as components of long COVID, yet long-term data remain limited. This study evaluated the prevalence of dysglycemia and thyroid autoimmunity four years after SARS-CoV-2 infection in adults without previously known endocrine disease. Methods: We conducted a retrospective longitudinal 4-year evaluation of adults hospitalized for COVID-19 between 2020 and 2021. Of 1009 eligible patients without prior diabetes or thyroid disease, 96 completed a standardized 4-year post-infection evaluation. Acute-phase data included COVID-19 severity, admission glucose, inflammatory markers, imaging findings, and treatments. The 4-year evaluation comprised fasting plasma glucose, thyroid function tests, anti-thyroid antibodies (anti-TPO, anti-Tg), and thyroid ultrasonography. Baseline HbA1c, thyroid autoantibodies, and thyroid imaging were not available. Results: At four years post-infection, 27.1% of patients exhibited dysglycemia compatible with type 2 diabetes mellitus, 41.6% showed thyroid autoimmunity, and 15.6% presented with both conditions. Overall, 47.9% developed at least one endocrine alteration. Admission hyperglycemia strongly predicted long-term dysglycemia (OR 6.67; 95% CI: 1.45–30.58), and diabetes prevalence increased with acute disease severity. Thyroid autoimmunity was frequent but not associated with initial COVID-19 severity. Conclusions: Four years after SARS-CoV-2 infection, a substantial proportion of patients exhibited persistent metabolic and autoimmune alterations, supporting a long COVID immunometabolic phenotype. In the absence of baseline endocrine data, the reported findings reflect long-term endocrine alterations identified at the 4-year evaluation, with a potential role of SARS-CoV-2 infection. These findings highlight the importance of baseline metabolic and thyroid assessment—including HbA1c and thyroid autoantibodies—in hospitalized COVID-19 patients and underscore the need for structured long-term endocrine monitoring.

1. Introduction

The COVID-19 pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has constituted one of the most profound global health challenges in recent decades. Although the acute phase of the pandemic has receded due to widespread natural and vaccine-induced immunity, a considerable proportion of individuals continue to experience persistent or delayed manifestations extending beyond the initial infection. These long-term effects, collectively referred to as post-acute sequelae of COVID-19 (PASC) or long COVID, encompass heterogeneous symptoms affecting multiple organ systems [1,2,3,4].

Endocrine and metabolic disturbances have increasingly been recognized as core components of long COVID, reflecting the multisystemic nature of SARS-CoV-2 infection. Clinical manifestations such as fatigue, cognitive impairment, cardiovascular symptoms, impaired glucose regulation, and thyroid abnormalities have been reported during long-term follow-up [4,5,6]. The susceptibility of the endocrine system may be partly explained by the abundant expression of angiotensin-converting enzyme 2 (ACE2) in pancreatic β-cells and thyroid follicular cells, enabling potential viral tropism. Additional mechanisms—including cytokine-mediated inflammation, oxidative stress, endothelial dysfunction, and persistent immune dysregulation—have been proposed to contribute to chronic endocrine sequelae [7,8,9]. The relationship between SARS-CoV-2 infection and glucose metabolism is both bidirectional and complex. Individuals with diabetes mellitus are more likely to develop severe forms of COVID-19, while SARS-CoV-2 infection itself may induce sustained dysglycemia, β-cell dysfunction, or unmask latent metabolic disease [10,11,12,13]. Large cohort studies have reported an increased incidence of post-COVID dysglycemia and diabetes extending months to years after infection [14,15,16].

Thyroid involvement has also been widely documented, ranging from transient subacute thyroiditis to persistent autoimmune thyroid disease. ACE2 and TMPRSS2 expression within the thyroid may facilitate direct viral injury, whereas post-infectious immune activation can promote anti-TPO and anti-Tg antibody production [17,18]. Although thyroid abnormalities have been described in the first 6–24 months after SARS-CoV-2 infection, very few studies have extended their follow-up beyond three years, and the long-term evolution of post-COVID thyroid autoimmunity remains insufficiently understood [19,20,21,22,23].

Beyond peripheral endocrine organs, emerging evidence suggests that SARS-CoV-2 infection may also affect the hypothalamic–pituitary axis, leading to central endocrine dysregulation as part of the long COVID spectrum. Alterations in pituitary hormone secretion and hypothalamic regulation have been described after COVID-19, supporting a broader neuroendocrine involvement in post-acute sequelae. However, pituitary function has not been systematically evaluated in most long-term studies, including the present investigation [24].

A structured evaluation at four years post-infection offers an opportunity to determine whether these endocrine abnormalities resolve, persist, or newly develop as part of the long COVID phenotype. This extended interval exceeds the expected duration of stress-induced hyperglycemia or post-viral thyroiditis, allowing for the identification of potentially chronic or delayed endocrine sequelae.

Based on these considerations, the present study reassessed a cohort of adults hospitalized for COVID-19 in 2020–2021 to evaluate long-term endocrine and metabolic outcomes four years after infection. The primary objective was to determine the 4-year prevalence of post-COVID endocrine alterations, specifically: (i) dysglycemia compatible with type 2 diabetes mellitus and (ii) thyroid autoimmunity, in individuals without previously known endocrine disease. A secondary objective was to examine the association between acute COVID-19 severity, admission hyperglycemia, and the risk of long-term endocrine sequelae as components of the long COVID spectrum.

2. Materials and Methods

2.1. Study Design and Setting

This retrospective observational cohort study was conducted at the Clinical Hospital of Pneumophthisiology and Infectious Diseases in Brașov, Romania. The study population was identified from the hospital’s electronic records and included adults (≥18 years) hospitalized with confirmed SARS-CoV-2 infection by RT-PCR or antigen testing between 1 August 2020 and 31 July 2021. From this database, 1009 adults without documented diabetes mellitus or thyroid disease at the time of admission were considered eligible for long-term follow-up. Initially, the post-COVID follow-up phase aimed to include only patients who had experienced severe or critical forms of COVID-19, based on the assumption that they might be at increased risk of long-term endocrine alterations. Due to limited participation, the eligibility criteria were subsequently expanded to include all patients from the initial cohort, regardless of acute disease severity.

Between January 2024 and June 2025, eligible patients were contacted by telephone and invited to return for standardized clinical and paraclinical reassessment. Of the 1009 eligible individuals, 96 ultimately completed the 4-year post-infection evaluation. The substantial non-random loss to follow-up, resulting from mortality, inability to contact patients, or refusal to participate, may affect cohort representativeness. The patient selection process and follow-up flow are illustrated in Figure 1.

The study population comprised adults hospitalized for COVID-19 between 2020 and 2021, with a relatively balanced sex distribution and a predominance of middle-aged and older individuals. Patients spanning the full spectrum of acute COVID-19 severity, from mild to critical disease, were included, allowing for comparisons according to clinical severity at presentation.

The study received ethical approval from the Ethics Committee of the Clinical Hospital of Pneumophthisiology and Infectious Diseases in Brașov (approval no. 9328/20 June 2024) and from the Ethics Committee of “Lucian Blaga” University of Sibiu (approval no. 16/25 November 2022). The initial COVID-19 diagnosis during hospitalization was established based on clinical presentation and confirmed through RT-PCR or antigen testing.

2.2. Inclusion and Exclusion Criteria

Eligible participants were adults (≥18 years) hospitalized with confirmed COVID-19 between 2020 and 2021 who agreed to participate in the 4-year follow-up evaluation. Patients were included if they attended the scheduled reassessment between January 2024 and June 2025. Individuals were excluded if they were younger than 18 years, pregnant at the time of reassessment, deceased during the interim period, declined participation, or could not be contacted.

Dysglycemia at evaluation was defined according to ADA criteria as fasting plasma glucose ≥ 126 mg/dL. HbA1c measurements were not available at baseline or follow-up; therefore, classification relied exclusively on fasting plasma glucose values. Thyroid autoimmunity was defined by (i) anti-thyroid peroxidase (anti-TPO) antibody levels > 10 IU/mL and/or anti-thyroglobulin (anti-Tg) levels > 95 IU/mL (Snibe Diagnostic, Shenzhen, China; manufacturer cut-offs) and/or (ii) ultrasound findings compatible with chronic autoimmune thyroiditis according to established criteria [25].

2.3. Data Collection and Procedures

Data were collected at two time points: (i) during the acute COVID-19 hospitalization and (ii) at a 4-year post-infection evaluation (

Table 1. Variables collected at admission and at 4-year post-COVID evaluation.

Time Point	Variable Type	Variables Collected
Acute COVID-19 Admission	Demographic Data	Age, sex
	Medical History	Cardiovascular, pulmonary, renal, metabolic, hepatic comorbidities
	Clinical Data	Presenting symptoms, disease severity, oxygen therapy or non-invasive ventilation requirement, length of hospital stay, treatments administered, clinical outcome
	Laboratory Parameters	Admission glucose, inflammatory markers (CRP, ferritin, fibrinogen, ESR), hematologic and coagulation markers (leukocytes, lymphocytes, platelets, D-dimer), tissue injury markers (AST, ALT, LDH), renal function (urea, creatinine)
	Imaging Investigations	Chest X-ray or chest CT (at admission, during hospitalization, and at discharge), lowest recorded peripheral oxygen saturation (SpO2 in room air)
4-Year Post-COVID Evaluation	Demographic Data	Age, sex
	Medical History	Cardiovascular, pulmonary, renal, metabolic, hepatic comorbidities—previously existing or developed after acute COVID-19
	Laboratory Parameters	Fasting glucose, inflammatory markers (CRP, ferritin, fibrinogen, ESR), hematologic and coagulation markers (leukocytes, lymphocytes, platelets, D-dimer), tissue injury markers (AST, ALT, LDH)
	Thyroid Function and Autoimmunity	TSH, FT4, FT3, thyroid autoantibodies (anti-TPO, anti-Tg)
	Imaging Investigations	Thyroid ultrasound (identification of autoimmune thyroiditis features, thyroid nodules, thyroid cysts)

). Although none of the patients had a known history of endocrine or metabolic disease at the time of hospitalization, baseline thyroid autoantibodies and HbA1c were not assessed during the acute admission. Therefore, the study cannot exclude the presence of previously unrecognized subclinical endocrine abnormalities, and the findings reflect endocrine alterations identified at the 4-year evaluation, without allowing for firm conclusions to be made regarding the timing or causality of these alterations.

2.3.1. Acute Phase Data (Hospital Admission, 2020–2021)

Clinical data collected during hospitalization included demographic characteristics, comorbidities, presenting symptoms, disease severity, treatment received, oxygen or ventilatory support, length of stay, and clinical outcome. Laboratory parameters included admission blood glucose, inflammatory markers (CRP, ferritin, fibrinogen, ESR), hematological and coagulation indices (white blood cells, lymphocytes, platelets, D-dimer), tissue injury markers (AST, ALT, LDH), and renal function tests (urea, creatinine). Admission blood glucose measurements were obtained at hospital presentation, before the initiation of systemic corticosteroid therapy, in order to minimize the influence of treatment-related hyperglycemia. Imaging investigations included chest X-ray or baseline and follow-up thoracic CT, along with the lowest recorded peripheral oxygen saturation (SpO₂) in room air.

2.3.2. Four-Year Reassessment (2024–2025)

Four years after the acute episode, participants underwent a standardized clinical and laboratory reassessment. Collected data included demographic information, updated medical history, and laboratory tests such as fasting glucose, inflammatory and hematologic markers, tissue injury markers, and renal function parameters. Thyroid evaluation included measurements of TSH, FT4, FT3, and thyroid autoantibodies (anti-TPO and anti-Tg). Thyroid ultrasound was performed to assess structural abnormalities, including patterns suggestive of autoimmune thyroiditis, as well as the presence of nodules or cysts. All ultrasound examinations were conducted by a single senior endocrinologist using a standardized scanning protocol. Although formal intra-observer variability testing was not performed, the use of a single experienced operator ensured procedural consistency.

2.4. Thyroid Laboratory Assessment

Serum thyroid-stimulating hormone (TSH), free thyroxine (FT4), free triiodothyronine (FT3), anti-thyroid peroxidase antibodies (anti-TPO), and anti-thyroglobulin antibodies (anti-TG) were measured using a chemiluminescent immunoassay (CLIA) on a fully automated immunoassay analyzer (MAGLUMI X3, Snibe Diagnostic, Shenzhen, China). The reference ranges were as follows: TSH 0.4–4.5 mIU/L, FT4 8.9–17.2 pmol/L, FT3 2.6–6.0 pmol/L, anti-TPO < 10 IU/mL, and anti-TG < 95 IU/mL. All measurements were performed in the same certified laboratory using standardized procedures and internal and external quality control. Thyroid ultrasonography was performed using a SonoScape ultrasound system manufactured by SonoScape Medical Corp. (Shenzhen, China) equipped with a high-frequency linear transducer (7–12 MHz);

2.5. Severity Classification of COVID-19

COVID-19 severity was classified based on national and international guidelines [26,27]. For clarity, mild cases were defined as upper respiratory symptoms without radiologic pneumonia or hypoxemia. Moderate disease involved radiological pneumonia with SpO₂ > 94%. Severe cases required the presence of respiratory distress, respiratory rate > 30/min, or SpO₂ < 90%. Critical illness included respiratory failure requiring ventilatory support, acute respiratory distress syndrome (ARDS), septic shock, or major thrombotic events documented during hospitalization.

2.6. Analytical Approach

To explore the association between acute COVID-19 severity and long-term endocrine alterations, patients were grouped into four severity categories (mild, moderate, severe, and critical), and the 4-year prevalence of dysglycemia compatible with type 2 diabetes and thyroid autoimmunity was compared across groups. The association between admission hyperglycemia and dysglycemia identified at the 4-year evaluation was explored by comparing patients with and without hyperglycemia at hospital admission. Subgroup analyses were conducted to examine associations between endocrine alterations and demographic or clinical characteristics, including sex, age (<60 vs. ≥60 years), and comorbidities (hypertension, obesity, dyslipidemia). The analyses explored: (i) sex differences in thyroid autoimmunity, (ii) age-related patterns of metabolic or immunologic dysregulation, and (iii) the role of comorbidities in long-term dysglycemia.

The potential relationship between severe or critical COVID-19, systemic inflammation, corticosteroid exposure, and long-term endocrine changes was explored descriptively, without formal causal inference, given the limited sample size.

To assess the potential coexistence of metabolic and thyroid abnormalities, the prevalence of dysglycemia was compared between patients with and without thyroid autoantibodies. Finally, the overall long-term endocrine burden identified at the 4-year evaluation was estimated by determining the proportion of patients exhibiting at least one endocrine alteration.

2.7. Statistical Analysis

Data preprocessing was performed in Microsoft Excel (Microsoft Corp., Redmond, WA, USA), followed by statistical analysis in Python 3.11 (Python Software Foundation, Wilmington, DE, USA) using pandas 2.2, NumPy 1.26, SciPy 1.11, and statsmodels 0.14. Group comparisons between categorical variables were conducted using Pearson’s chi-square test. Given the limited sample size of the follow-up cohort and the small number of outcome events, multivariable logistic regression models could not be reliably implemented without generating unstable or overfitted estimates. Therefore, all inferential results should be interpreted as unadjusted and exploratory. For analyses involving multiple hypotheses, p-values were corrected using the Benjamini–Hochberg false discovery rate procedure, with a two-sided alpha level of 0.05. Plots were created using Matplotlib 3.8 and Seaborn 0.13, and Excel was used in parallel for tabulation and cross-checking.

3. Results

Of the 1009 adults eligible for long-term evaluation and without known diabetes or thyroid disease at the time of their initial hospitalization, 96 completed the 4-year evaluation. Among the remaining 913 individuals, 115 had died within four years of the acute COVID-19 episode, 501 could not be contacted, and 297 declined in-person follow-up. This substantial non-random attrition may introduce follow-up bias; this is addressed in the Limitations Section.

3.1. Demographic and Clinical Characteristics

The demographic characteristics and acute COVID-19 severity of the study population are summarized in

Table 2. Baseline demographic characteristics and acute COVID-19 severity of the study cohort.

Characteristic	n (%)
Sex
Male	44 (45.8)
Female	52 (54.2)
Age (years)
Median (IQR)	58 (49–66)
30–39	4 (4.2)
40–49	22 (22.9)
50–59	24 (25.0)
60–69	32 (33.3)
≥70	14 (14.6)
Acute COVID-19 severity
Mild	26 (27.1)
Moderate	25 (26.0)
Severe	37 (38.5)
Critical	8 (8.3)

Legend: Baseline demographic characteristics and clinical severity of acute COVID-19 among patients included in the 4-year post-infection evaluation (n = 96).

. The final cohort consisted of 96 adults, with a balanced sex distribution and a predominance of middle-aged and older individuals. Acute COVID-19 severity ranged from mild to critical disease, allowing for comparative analyses across the full spectrum of clinical presentations.

3.2. Relationship Between Acute COVID-19 Severity and Long-Term Endocrine Alterations

3.2.1. Long-Term Dysglycemia Compatible with Type 2 Diabetes

At hospital admission, 69 patients (71.9%) presented with hyperglycemia, while 27 (28.1%) were normoglycemic. At the 4-year evaluation, dysglycemia was identified in 26 of the 96 reassessed patients (27.1%). Dysglycemia at follow-up was significantly more frequent among patients with admission hyperglycemia compared with those who were normoglycemic at admission (χ² test, p = 0.01; Figure 2).

Corticosteroid therapy was administered during hospitalization in 77 of 96 patients (80.2%), in accordance with treatment protocols for moderate-to-severe COVID-19.

When stratified by acute COVID-19 severity, the prevalence of dysglycemia at the 4-year evaluation differed across severity categories: mild (7.7%), moderate (32.0%), severe (40.5%), and critical (12.5%) (

Table 3. Long-term dysglycemia at the 4-year evaluation according to acute COVID-19 clinical severity.

Acute COVID-19 Severity	Total Patients (n)	Patients with Dysglycemia at 4 Years (n)	Dysglycemia Within Severity Group (%)
Mild	26	2	7.7
Moderate	25	8	32.0
Severe	37	15	40.5
Critical	8	1	12.5

Legend: Distribution of dysglycemia detected at the 4-year evaluation, stratified by acute COVID-19 clinical severity. Values are expressed as absolute numbers and percentages within each severity category.

).

3.2.2. Post-COVID-19 Thyroid Alterations: Functional and Structural Changes

At the 4-year follow-up, thyroid evaluation revealed a high prevalence of both structural and immunological abnormalities across all categories of acute COVID-19 severity. Ultrasound features compatible with autoimmune thyroiditis, thyroid nodules, and thyroid cysts were identified in patients from all severity groups. In parallel, serological testing demonstrated frequent positivity for thyroid autoantibodies, including anti-thyroid peroxidase (anti-TPO), anti-thyroglobulin (anti-Tg), and dual antibody positivity.

The distribution of thyroid structural findings and autoantibody positivity stratified by acute COVID-19 severity is summarized in

Table 4. Thyroid structural and autoimmune abnormalities according to clinical severity of acute COVID-19.

Clinical Severity	Autoimmune Thyroiditis n (%)	Thyroid Nodules n (%)	Thyroid Cysts n (%)	Anti-TPO Positive n (%)	Anti-Tg Positive n (%)	Double Positivity (Anti-TPO + Anti-Tg) n (%)
Mild (n = 26)	2 (7.7)	8 (30.8)	4 (15.4)	10 (38.5)	12 (46.2)	4 (15.4)
Moderate (n = 25)	6 (24.0)	5 (20.0)	6 (24.0)	9 (36.0)	7 (28.0)	5 (20.0)
Severe (n = 37)	6 (16.2)	6 (16.2)	13 (35.1)	12 (32.4)	9 (24.3)	3 (8.1)
Critical (n = 8)	2 (25.0)	1 (12.5)	0 (0.0)	5 (62.5)	2 (25.0)	1 (12.5)

Note: Values are presented as number (percentage). Autoimmune thyroiditis was defined based on ultrasonographic features consistent with chronic autoimmune thyroiditis. Anti-TPO and anti-Tg positivity were defined according to laboratory reference ranges.

. Overall, autoimmune thyroiditis on ultrasound was observed in 7.7–25.0% of patients across severity groups, while thyroid nodules and cysts were also commonly detected. Anti-TPO and anti-Tg antibody positivity was frequent in all groups, with the highest proportion of anti-TPO positivity observed among patients who had experienced critical COVID-19.

Importantly, thyroid abnormalities were identified not only in patients with moderate-to-critical disease but also among individuals who had experienced mild acute COVID-19.

3.3. Identification of Risk Factors for Post-COVID Endocrine Sequelae

At the four-year follow-up, glycemic abnormalities compatible with type 2 diabetes were identified in 26 of 96 patients (27.1%). These abnormalities were more frequently observed in patients aged ≥ 60 years (p = 0.04), in those with hypertension (p = 0.002), and in patients who had experienced severe or critical acute COVID-19 (p = 0.005). Female sex (p = 1.00) and obesity (p = 0.38) were not significantly associated with long-term glycemic abnormalities.

In multivariable comparisons, hypertension (p = 0.002) and severe/critical acute COVID-19 (p = 0.005) remained significantly associated with post-COVID glycemic abnormalities, while the association with age ≥ 60 years did not reach statistical significance (p = 0.08). Detailed statistical results are presented in

Table 5. Association of demographic and clinical factors with long-term glycemic abnormalities and thyroid autoimmunity after COVID-19.

Factor Evaluated	Glycemic Abnormalities Present (n = 26)	No Glycemic Abnormalities (n = 70)	Thyroid Autoimmunity Present (n = 40)	No Thyroid Autoimmunity (n = 56)
Severe/critical COVID-19	p = 0.005	p = 0.13	p = 1.00	p = 0.63
Hypertension	p = 0.002	p = 0.05	p = 0.27	p = 0.27
Obesity	p = 0.38	p = 0.68	p = 1.00	p = 1.00
Female sex	p = 1.00	p = 1.00	p = 0.35	p = 0.43
Age ≥ 60 years	p = 0.08	p = 0.22	p = 0.46	p = 0.52

Legend: Statistical comparison of demographic and clinical variables associated with long-term glycemic abnormalities and thyroid autoimmunity after COVID-19. Statistically significant associations were observed for hypertension and severe/critical acute COVID-19 in relation to glycemic abnormalities. No evaluated variable was significantly associated with thyroid autoimmunity.

.

Thyroid autoimmunity, defined by the presence of anti-TPO and/or anti-Tg antibodies with or without ultrasound features of autoimmune thyroiditis, was identified in 40 of 96 patients (41.6%). Among these, 24 patients (64.9%) showed isolated serological autoimmunity, while 13 patients (35.1%) presented ultrasound findings consistent with autoimmune thyroiditis.

No significant associations were observed between thyroid autoimmunity and female sex (p = 0.35), age ≥60 years (p = 0.46), hypertension (p = 0.27), obesity (p = 1.00), or acute COVID-19 severity (p = 1.00). All comparisons are summarized in Table 5.

3.4. Association Between Thyroid Autoimmunity and Long-Term Glycemic Abnormalities

A descriptive analysis was performed to assess the coexistence of thyroid autoimmunity and long-term glycemic abnormalities four years after SARS-CoV-2 infection. Thyroid autoimmunity was identified in 40 of 96 patients (41.6%), while long-term glycemic abnormalities were present in 26 patients (27.1%).

Both thyroid autoimmunity and glycemic abnormalities were observed in 15 patients (15.6%). Among patients with thyroid autoimmunity, 15 of 40 (37.5%) also exhibited glycemic abnormalities. Conversely, among patients with glycemic abnormalities, 15 of 26 (57.7%) demonstrated concomitant thyroid autoimmunity.

Overall, 46 of 96 patients (47.9%) presented at least one endocrine alteration four years after infection, either thyroid autoimmunity, long-term glycemic abnormalities, or both. The distribution of endocrine findings is presented in

Table 6. Distribution of long-term endocrine findings four years after acute SARS-CoV-2 infection.

Endocrine Outcome	n (%)
Thyroid autoimmunity only	30 (31.2)
Glycemic abnormality only	26 (27.1)
Both thyroid autoimmunity and glycemic abnormality	15 (15.6)

.

4. Discussion

The findings of this study indicate that SARS-CoV-2 infection may have sustained endocrine implications extending several years beyond the acute illness. Our results reinforce emerging evidence that COVID-19 can disrupt endocrine and metabolic homeostasis, affecting organs such as the pancreas and thyroid [7,10].

4.1. General Considerations on the Persistent Endocrine Implications of SARS-CoV-2 Infection

The present study provides a detailed assessment of endocrine status in previously hospitalized COVID-19 patients, evaluated four years after SARS-CoV-2 infection. To our knowledge, this represents one of the longest observational assessments addressing post-COVID endocrine health, extending beyond the 6–24-month timeframe explored in most existing studies and enabling the identification of late-emerging metabolic and autoimmune disturbances [5,6].

At the time of reassessment, 27% of patients exhibited long-term glycemic abnormalities, despite having no known history of diabetes at baseline. In the absence of baseline HbA1c or pre-infection glycemic parameters, the designation of “new-onset diabetes” could not be applied. Even so, the prevalence of glycemic impairment four years after infection mirrors the elevated risk of post-COVID dysglycemia documented in several large epidemiological studies [13].

A particularly relevant finding is the association between acute-phase hyperglycemia and long-term glycemic abnormalities. Among patients who presented with hyperglycemia at admission (71.9%), 34.8% met glycemic criteria compatible with long-term dysglycemia, compared with only 7.4% of initially normoglycemic patients. These results support the hypothesis that COVID-19-related stress hyperglycemia reflects underlying metabolic vulnerability, amplified by systemic inflammation, corticosteroid exposure, and the host immune response [12,28].

Regarding thyroid involvement, thyroid autoimmunity was identified in 41.6% of patients at the four-year assessment, and structural abnormalities, including nodules, cysts, or ultrasound patterns consistent with autoimmune thyroiditis, were likewise common. Anti-thyroid antibody positivity was frequent, with 38.5% of patients exhibiting elevated anti-TPO antibodies and 46.2% showing anti-thyroglobulin positivity, while 13.5% demonstrated ultrasound-confirmed features of autoimmune thyroiditis. These prevalence rates substantially exceed those expected in the general adult population (≈10–15%), as reported in large community-based epidemiological studies such as the Whickham surveys and the NHANES III study [29,30,31], and are consistent with findings from Asian and European cohorts assessed 6–12 months post-COVID [18,21,32]. The persistence of these abnormalities several years after infection suggests that thyroid autoimmunity may represent a delayed or chronic process, possibly driven by prolonged immune activation, molecular mimicry, or post-viral dysregulation [19,20].

Although moderate, severe, and critical COVID-19 were associated with numerically higher autoimmune markers, we did not identify a statistically significant relationship between acute severity and thyroid autoimmunity, echoing previous reports that thyroid autoimmunity may arise even after mild infection [22,23]. Overall, the findings indicate that a substantial proportion of previously hospitalized COVID-19 patients exhibit persistent endocrine alterations—metabolic disturbances, thyroid autoimmunity, and occasionally structural thyroid changes—four years after infection [22,23,33].

Several limitations must be acknowledged. The small final sample size, resulting from substantial attrition of the initially eligible cohort, raises concerns regarding selection and survivorship bias and limits the generalizability of the findings. In addition, the limited number of outcome events precluded adjusted multivariable analyses, and residual confounding cannot be excluded, particularly for glycemic outcomes influenced by age and cardiometabolic risk.

Another important limitation is the lack of systematic baseline endocrine characterization. Although patients with known diabetes or thyroid disease were excluded, no baseline HbA1c, thyroid autoantibodies, or thyroid imaging were available at hospital admission. Consequently, it cannot be definitively established whether the endocrine abnormalities observed represent incident post-COVID conditions or the progression of previously unrecognized subclinical disease.

Information on SARS-CoV-2 reinfections and viral variants during the four-year interval was not available, and data on family history of diabetes, thyroid disease, or autoimmune disorders were lacking. Furthermore, baseline biochemical markers beyond routine parameters were not explored. In this context, factors such as vitamin D deficiency—highly prevalent in both acute and long COVID and increasingly recognized as a modulator of immune and metabolic function—or serum calcium, proposed as a surrogate marker of disease severity and systemic inflammatory stress, may represent relevant variables for future studies [9].

Finally, the single-center design and the exclusive inclusion of hospitalized patients—many with moderate to severe disease—limit extrapolation of the results to non-hospitalized individuals or broader populations with different demographic and clinical characteristics.

Despite these limitations, the present study offers valuable insight into the long-term endocrine footprint of SARS-CoV-2 infection. The persistence of metabolic disturbances and thyroid autoimmunity several years after acute illness supports the concept that SARS-CoV-2 may act as a trigger for long-lasting endocrine dysregulation, potentially mediated by systemic inflammation, oxidative stress, immune dysregulation, and possibly direct or indirect viral effects on endocrine tissues [8,9,19].

A major strength of the present study lies in the exceptionally long interval between acute infection and endocrine reassessment, providing a unique perspective on very long-term endocrine consequences of COVID-19. By jointly assessing thyroid parameters and glucose metabolism, the study demonstrates that detectable endocrine alterations may persist years after the acute episode.

These findings support the need for periodic endocrine evaluation in COVID-19 survivors, particularly among individuals who experienced severe disease or stress hyperglycemia during hospitalization [6,31]. Targeted late post-infection screening—including thyroid autoantibodies, thyroid ultrasound when clinically indicated, and fasting glucose or HbA1c—may facilitate earlier identification and management of persistent post-COVID endocrine alterations.

4.2. Long-Term Dysglycemia After COVID-19: Prevalence, Predictors, and Mechanistic Pathways

4.2.1. Prevalence of Long-Term Dysglycemia at Four Years

At the 4-year follow-up, 27.1% of patients without previously known diabetes exhibited dysglycemia compatible with type 2 diabetes mellitus, indicating a substantial long-term metabolic burden after SARS-CoV-2 infection. Although comparisons with earlier time points (6–24 months) from other cohorts show lower reported rates, the direction of association is consistent with large studies demonstrating an increased risk of chronic glycemic abnormalities following COVID-19. For instance, Wander et al. reported a 40–60% relative excess risk of diabetes at 12 months among U.S. veterans [34], while Xie and Al-Aly observed persistent elevations in diabetes risk up to two years post-infection, with a clear gradient according to acute disease severity [13,34]. European registry data similarly show elevated long-term diabetes risk compared with non-COVID respiratory infections [35]. The higher prevalence observed in our cohort likely reflects the profile of the included population: exclusively hospitalized patients, nearly half of whom had severe or critical disease, older age distribution, and a higher frequency of cardiometabolic comorbidities, factors known to predispose to metabolic deterioration [12,16].

4.2.2. Mechanistic Pathways Linking SARS-CoV-2 Infection to Persistent Dysglycemia

Several biological pathways may explain why COVID-19 is associated with long-term alterations in glucose homeostasis. Current evidence supports a multifactorial process involving direct viral effects, inflammation-driven insulin resistance, endothelial injury, and neurohormonal stress responses.

Table 7. Proposed mechanisms linking SARS-CoV-2 infection to long-term dysglycemia.

Mechanism	Pathophysiological Basis	Metabolic Consequences
Direct pancreatic β-cell injury	SARS-CoV-2 entry into pancreatic β-cells via ACE2 and TMPRSS2 leads to cytopathic effects, impaired insulin secretion, and altered β-cell gene expression [8,36,37].	Reduced β-cell reserve, inadequate insulin secretion, and long-term glycemic instability.
Cytokine-mediated insulin resistance	Pro-inflammatory cytokines (IL-6, TNF-α, IL-1β) disrupt IRS/PI3K/AKT signaling, increasing hepatic gluconeogenesis and lipolysis while reducing peripheral glucose uptake [10,38].	Persistent insulin resistance and chronic dysglycemia.
Endothelial dysfunction and microvascular injury	Oxidative stress and endothelial damage impair skeletal muscle and hepatic glucose utilization [9,39].	Exacerbation of insulin resistance and reduced glucose disposal.
Mitochondrial dysfunction	Viral interference with mitochondrial respiration impairs oxidative phosphorylation and ATP synthesis [9,38].	Decreased metabolic flexibility and impaired compensatory insulin secretion.
Neurohormonal activation (stress hyperglycemia)	Activation of the hypothalamic–pituitary–adrenal axis increases cortisol and catecholamine levels, inducing transient but severe insulin resistance [10,38].	Acute hyperglycemia and unmasking of underlying metabolic susceptibility.
Steroid-associated dysglycemia	Glucocorticoid therapy increases hepatic glucose production and induces peripheral insulin resistance [12].	Exacerbated acute dysglycemia and increased long-term metabolic risk in susceptible individuals.

Legend: Summary of biological mechanisms implicated in post-COVID dysglycemia, integrating evidence from virology, immunology, endothelial biology, and metabolic physiology [9,38,39,40].

summarizes the principal mechanisms proposed in the literature.

4.2.3. Admission Hyperglycemia as a Predictor of Long-Term Metabolic Impairment

A key finding of this study is the high prevalence of hyperglycemia at hospital admission, observed in 71.9% of patients despite the absence of previously documented metabolic disease. At the 4-year follow-up, dysglycemia compatible with type 2 diabetes mellitus was identified in 33.3% of patients who were hyperglycemic during the acute phase, compared with only 7.4% of those who had been normoglycemic at admission. This marked difference suggests that admission hyperglycemia represents an early indicator of long-term metabolic vulnerability rather than a transient, isolated phenomenon.

In the absence of baseline HbA1c measurements or pre-infection glycemic data, it is not possible to distinguish retrospectively between stress-induced hyperglycemia related to acute systemic inflammation and unrecognized premorbid metabolic dysfunction. Consequently, the present findings cannot be interpreted as evidence of definitive new-onset diabetes attributable to SARS-CoV-2 infection. Instead, they indicate that acute-phase hyperglycemia may unmask a latent susceptibility to subsequent metabolic deterioration.

Systemic corticosteroid therapy was widely administered during hospitalization, with more than 80% of patients receiving dexamethasone or equivalent agents in accordance with treatment protocols for moderate-to-severe COVID-19. While glucocorticoids are well known to exacerbate hyperglycemia during acute illness, the persistence of dysglycemia four years after infection in a substantial subset of patients exceeds what would typically be expected from transient steroid-induced metabolic effects alone. Notably, long-term dysglycemia occurred predominantly among individuals who were already hyperglycemic at admission, supporting the concept that admission glycemia is a stronger determinant of long-term metabolic outcomes than corticosteroid exposure per se [12,16].

These findings are consistent with prior reports indicating that admission hyperglycemia reflects both acute disease severity and underlying metabolic vulnerability. Zhu et al. demonstrated that glucose levels > 140 mg/dL at hospital admission were associated with worse metabolic outcomes and an increased risk of subsequent dysglycemia [41]. Similarly, other large cohort studies have shown that stress hyperglycemia during acute COVID-19 is associated with worse metabolic outcomes and may identify individuals at increased risk for subsequent dysglycemia [42].

Taken together, the present data indicate a clinically meaningful association between admission hyperglycemia, acute COVID-19 severity, and persistent dysglycemia several years after infection. Although causal inference is precluded by the observational design and the lack of pre-infection metabolic assessment, these findings support the importance of structured long-term metabolic follow-up in patients hospitalized for COVID-19, particularly those presenting with hyperglycemia during the acute phase [6,12].

4.2.4. Clinical Implications and Recommendations for Follow-Up

These findings underscore the clinical relevance of admission hyperglycemia as an early indicator of long-term metabolic vulnerability in COVID-19 survivors, a concept already highlighted in post-COVID endocrine follow-up recommendations [6,31]. Even when glucose levels normalize after discharge, individuals who present with hyperglycemia during acute infection should be considered at increased risk for delayed metabolic deterioration.

For this reason, structured follow-up is warranted. Current evidence supports reassessing HbA1c, fasting glucose, and lipid profile within 3–6 months after discharge, followed by annual metabolic evaluations, with intensified monitoring in patients with additional risk factors such as older age, hypertension, dyslipidemia, obesity, or prior corticosteroid exposure [6]. Early recognition of emerging glycemic abnormalities may enable timely lifestyle intervention or pharmacologic therapy, potentially reducing the progression to overt diabetes.

Overall, the present findings reinforce the notion that SARS-CoV-2 infection may induce a persistent state of metabolic susceptibility, particularly among individuals with stress hyperglycemia or severe forms of acute illness. Accordingly, integration of long-term metabolic surveillance into routine post-COVID care is advisable and aligns with current endocrine society recommendations [6].

4.3. COVID-19 and Long-Term Thyroid Dysfunction

Thyroid dysfunction represents one of the most frequently reported endocrine manifestations of COVID-19, driven by direct viral injury, cytokine-mediated inflammation, and immune dysregulation [43]. The present study demonstrates a remarkably high prevalence of thyroid structural and autoimmune abnormalities persisting four years after documented SARS-CoV-2 infection. Thyroid alterations were observed across all categories of acute disease severity, including patients who had experienced clinically mild COVID-19, indicating that long-term thyroid involvement is not confined to severe or critical forms of the disease. Although moderate, severe, and critical COVID-19 were associated with numerically increased rates of cystic changes and thyroid autoantibody positivity—most notably anti-TPO antibodies in the critical subgroup—these differences should be interpreted cautiously considering the relatively small subgroup sample sizes.

At the four-year reassessment, thyroid autoimmunity emerged as one of the most prevalent long-term endocrine sequelae of SARS-CoV-2 infection. Overall, 41.6% of individuals who were euthyroid prior to infection demonstrated evidence of thyroid autoimmunity, including anti-TPO positivity in 38.5%, anti-Tg positivity in 46.2%, and ultrasonographic features consistent with autoimmune thyroiditis in 13.5%. These values markedly exceed the background prevalence expected in European adult populations—typically 10–15% for anti-TPO antibodies, 8–12% for anti-Tg antibodies, and approximately 5–10% for ultrasound findings suggestive of chronic autoimmune thyroiditis [29,30,31]. The magnitude of this excess supports the hypothesis of increased susceptibility to autoimmune thyroid involvement following COVID-19.

These observations are directionally consistent with findings from international post-COVID cohorts, although most previous studies have evaluated patients only within the first 6–12 months after acute infection. Campi et al. reported anti-TPO positivity in 15.7% of patients at three months, representing a two-fold increase compared with pre-pandemic controls (7.7%) [32]. Additional studies from China and Italy have reported increases in thyroid autoantibody prevalence in 12–25% of patients during 6–12 months of follow-up [32,44].

The substantially higher prevalence identified in our four-year cohort suggests the presence of either persistent subclinical immune dysregulation or a delayed autoimmune phenotype, which may become clinically or biochemically apparent only after a prolonged latency period. To contextualize these findings within current pathogenic frameworks,

Table 8. Proposed mechanisms linking SARS-CoV-2 infection to autoimmune thyroiditis.

Mechanism	Pathophysiological Basis with References	Resulting Immune and Thyroid Consequences
Direct viral cytopathic injury	High expression of ACE2 and TMPRSS2 on thyroid follicular epithelial cells enables SARS-CoV-2 entry and intracellular replication, as demonstrated by transcriptomic and immunohistochemical studies [45,46,47].	Follicular cell destruction with extracellular release of thyroglobulin (Tg) and thyroid peroxidase (TPO) antigens, leading to initiation or amplification of autoimmune responses.
Cytokine-mediated immune activation	The COVID-19-associated cytokine storm (increased IL-6, TNF-α, IFN-γ, and IL-1β) disrupts immune tolerance and enhances antigen presentation; chemokines such as CXCL10 and CCL2 promote lymphocytic infiltration [22,48,49].	Activation of autoreactive T and B lymphocytes, sustained inflammatory infiltration, and progression toward chronic autoimmune thyroiditis.
Molecular mimicry	Structural homology between SARS-CoV-2 spike or nucleocapsid proteins and thyroid antigens (Tg and TPO) facilitates cross-reactive immune responses, as shown in in silico and immunological studies [50,51].	Cross-reactive antibodies and T cells target thyroid tissue, resulting in persistent autoimmune thyroid involvement.
Genetic predisposition	SARS-CoV-2-induced immune activation may unmask latent autoimmunity in genetically susceptible individuals; persistent systemic inflammation may contribute to loss of immune tolerance [19,52].	Increased thyroid autoantibody production and progression toward chronic Hashimoto-like thyroiditis.

Legend. Summary of leading mechanisms proposed to explain post-COVID autoimmune thyroid involvement, integrating current immunological, virological, and endocrinological evidence.

provides an overview of the principal biological mechanisms proposed to link SARS-CoV-2 infection with the subsequent development of autoimmune thyroiditis.

Beyond these mechanistic pathways, emerging clinical data also suggest that thyroid injury during acute COVID-19 may not always resolve completely. While most longitudinal studies indicate normalization of thyroid function within months, for example, Lui et al. reported recovery in 82.4% of patients by 3–6 months [53], a distinct subgroup exhibits persistent or newly emerging autoimmunity after recovery. Our four-year data provide important evidence for this long-term trajectory, with nearly one-third of individuals showing serological or ultrasonographic abnormalities compatible with autoimmune thyroiditis well beyond the expected recovery period.

Importantly, the persistence of thyroid autoantibodies, even in euthyroid individuals, carries meaningful clinical implications. Anti-TPO-positive patients have a significantly increased lifetime risk of developing hypothyroidism, chronic fatigue, and subtle metabolic dysregulation [25,29]. Therefore, current expert reviews, including recent analyses published in Frontiers in Endocrinology, recommend long-term thyroid monitoring in patients with a history of COVID-19, especially those who experienced moderate or severe acute illness or exhibited elevated inflammatory markers [19,44].

Recommended follow-up evaluations include measurement of TSH, FT4, FT3, anti-TPO, and anti-Tg, along with periodic thyroid ultrasonography at 12–24 months and at extended intervals in high-risk individuals. These strategies aim to detect evolving thyroid dysfunction early and reflect the growing recognition that SARS-CoV-2 may induce persistent immune–endocrine dysregulation with clinically significant long-term consequences.

4.4. Endocrine Autoimmunity as a Potential Mechanism in Post-COVID Syndrome (PASC)

Our study provides indirect support for the hypothesis that endocrine autoimmunity may represent a contributing factor to the heterogeneous clinical manifestations observed in post-acute sequelae of COVID-19 (PASC). This syndrome is defined by the persistence or emergence of symptoms beyond 12 weeks after acute infection, in the absence of alternative explanations, and includes a wide spectrum of systemic manifestations such as chronic fatigue, cognitive impairment, respiratory dysfunction, cardiovascular abnormalities, and metabolic disturbances [4,54].

In this context, the thyroid and metabolic alterations identified four years after SARS-CoV-2 infection may represent endocrine components of long COVID rather than isolated post-infectious abnormalities, potentially mediated through persistent immune dysregulation. Dysregulation of the hypothalamic–pituitary–peripheral axes, residual low-grade inflammation, and sustained autoimmunity may contribute to nonspecific symptoms frequently reported in PASC, such as fatigue, reduced physical performance, and mild cognitive disturbances [24,55]. However, clinical PASC manifestations were not systematically assessed in the present study.

Thyroid autoimmunity, present in 41.6% of reassessed patients, may play a role in PASC-related symptomatology. SARS-CoV-2 has been implicated in the development of endocrine autoimmunity, particularly autoimmune thyroid disease, through mechanisms involving immune dysregulation and molecular mimicry [56,57,58]. Even in euthyroid individuals, anti-TPO and anti-Tg positivity can reflect early or subclinical autoimmune thyroid dysfunction, which may affect energy metabolism, neuromuscular function, and thermoregulation. Previous studies have shown that early-stage autoimmune thyroiditis may be associated with fatigue and impaired concentration, symptoms that partially overlap with those described in PASC [25,29].

Post-COVID metabolic disturbances, including newly diagnosed diabetes mellitus and persistent insulin resistance, may further reinforce systemic inflammation [12,59]. Chronic hyperglycemia and insulin resistance activate pro-inflammatory and oxidative stress pathways, contributing to sustained low-grade inflammation [9,10,40]. These mechanisms may partially overlap with pathophysiological pathways proposed in PASC. Nevertheless, a causal relationship cannot be inferred from the present data.

Overall, while our data document a high prevalence of endocrine abnormalities four years after COVID-19, the relationship between these findings and specific PASC manifestations remains hypothetical, underscoring the need for longitudinal studies integrating both biochemical and symptom-based assessments [59,60].

From a clinical perspective, patients presenting with persistent fatigue, exertional intolerance, or cognitive symptoms may benefit from evaluation for subclinical hypothyroidism and thyroid autoimmunity, whereas post-infectious glycemic monitoring remains essential to detect delayed-onset diabetes mellitus. Implementing multidisciplinary follow-up protocols, including infectious disease specialists, endocrinologists, and rehabilitation experts, may improve long-term outcomes and quality of life [6,60].

These findings emphasize the importance of considering endocrine autoimmunity and metabolic dysfunction as potential contributors to long-term systemic symptoms in PASC and highlight the need for integrated clinical management strategies.

4.5. Interaction Between Metabolic and Thyroid Axes and Risk Factors for Post-COVID Endocrine Sequelae

There is a well-recognized bidirectional relationship between thyroid function and glucose metabolism, with each system modulating the other through tightly integrated hormonal pathways. Thyroid hormones regulate hepatic glucose production, insulin sensitivity, and basal metabolic rate, whereas insulin and overall metabolic status influence the peripheral conversion of thyroxine (T4) to triiodothyronine (T3) [61,62]. Subclinical hypothyroidism reduces insulin clearance, decreases GLUT-4 expression, and increases insulin resistance at both hepatic and muscular levels [63]. Conversely, hyperthyroidism enhances hepatic gluconeogenesis and protein catabolism, promoting hyperglycemia [63].

In the post-COVID context—characterized by persistent inflammation, oxidative stress, and immune dysregulation—autoimmune thyroid involvement and impaired insulin sensitivity may coexist, creating a self-reinforcing metabolic–endocrine loop. Chronic inflammation can trigger both thyroid autoimmunity and insulin resistance, while hyperglycemia perpetuates systemic inflammation and oxidative stress, further contributing to thyroid injury [9,58].

In our cohort, a partial overlap between autoimmune thyroiditis and post-COVID diabetes was observed, suggesting shared pathogenic mechanisms. This is consistent with previous studies demonstrating that persistent low-grade inflammation—one of the hallmarks of post-acute COVID-19 syndrome—can sustain both metabolic deterioration and autoimmune reactivity through overlapping immune–endocrine pathways [55,58]. Recent reviews further indicate that COVID-19-related thyroid dysfunction frequently coexists with insulin resistance, reinforcing the notion of a bidirectional immunometabolic phenotype [58].

Risk factor analysis showed that the development of endocrine sequelae after COVID-19 reflects a combination of intrinsic predisposition and the inflammatory burden of the acute infection. Although female sex is a well-established risk factor for autoimmune thyroid disease in the general population, no statistically significant association was observed in our cohort, likely due to sample size limitations and the exploratory nature of the analysis [19]. Older age (≥60 years) was associated with a greater frequency of both diabetes and thyroid dysfunction, likely reflecting immunosenescence, reduced β-cell reserve, and accumulated metabolic comorbidities [6].

The severity of the acute infection emerged as one of the strongest predictors of long-term endocrine disturbances, in agreement with large cohort studies reporting a severity-dependent gradient of post-COVID diabetes risk [13]. Regarding thyroid involvement, earlier studies by Campi et al. and Lui et al. similarly demonstrated an association between heightened inflammatory responses and persistent anti-TPO elevation at 6–12 months [32,53]. Obesity and dyslipidemia were also associated with increased risk of post-COVID diabetes, acting as baseline metabolic vulnerabilities that amplify inflammation and insulin resistance [10,59].

While multivariable analyses could not be performed due to sample size constraints, the univariate associations identified here support the hypothesis that severe acute disease, systemic inflammation, and adverse metabolic profiles synergistically shape long-term endocrine risk after SARS-CoV-2 infection. Defining these profiles may help tailor targeted endocrine screening and longitudinal follow-up in high-risk individuals [64].

4.6. The Significance of the Cumulative Endocrine Burden After COVID-19

The high prevalence of endocrine abnormalities identified four years after SARS-CoV-2 infection highlights the systemic and long-term impact of COVID-19 on immune and metabolic homeostasis. Nearly half of the cohort (47.9%) exhibited at least one endocrine dysfunction, either new-onset type 2 diabetes mellitus or autoimmune thyroiditis, underscoring the multisystemic nature of post-COVID sequelae and the substantial endocrine component of this burden. These findings align with international reports describing increased risks of incident diabetes and thyroid autoimmunity following SARS-CoV-2 infection [13,33,53]. Compared with non-COVID background rates, the cumulative endocrine burden observed in this cohort appears considerably elevated. In the general population, the 3–5-year incidence of type 2 diabetes is approximately 4–7% [34,35], whereas 27.1% of our patient’s developed diabetes at the four-year mark. Similarly, population-based studies (Whickham Survey, NHANES III) report anti-TPO positivity in 10–15% of women and 4–7% of men, with 5–10% prevalence of ultrasound patterns suggestive of autoimmune thyroiditis in euthyroid adults [29,30,31]. In contrast, 41.6% of our cohort showed serological and/or structural markers of thyroid autoimmunity.

The coexistence of autoimmune thyroiditis and type 2 diabetes in 15.6% of patients suggests the emergence of a post-COVID immunometabolic phenotype driven by chronic inflammation, oxidative stress, and immune dysregulation [14,57]. This clustering resembles post-viral polyautoimmune states described after Epstein–Barr virus or cytomegalovirus infections, although it appears more pronounced in the context of SARS-CoV-2 due to its multisystemic tropism [48,52].

Clinically, these data emphasize the importance of incorporating systematic endocrine screening into long-term follow-up for COVID-19 survivors. An integrated surveillance strategy, including fasting glucose, HbA1c, TSH, FT4, and anti-TPO/anti-Tg antibody testing, may enable early detection of evolving abnormalities and allow for timely interventions to mitigate the long-term metabolic and functional consequences of COVID-19 [6,33].

4.7. Study Limitations

This study has several important limitations that should be considered when interpreting the findings. First, only 96 of the 1009 initially eligible patients completed the four-year reassessment. This substantial and non-random attrition, resulting from mortality, inability to contact participants, or refusal to return for follow-up, raises concerns regarding selection and survivorship bias and limits the generalizability of the results. Patients who survived and agreed to long-term reassessment may differ systematically from those lost to follow-up, potentially leading to an overestimation of the true prevalence of long-term endocrine alterations.

Second, systematic baseline endocrine characterization was not available at the time of hospital admission. Although patients with known diabetes mellitus or thyroid disease were excluded, baseline HbA1c levels, thyroid autoantibodies, and thyroid ultrasonography were not routinely assessed during the acute phase of infection. Consequently, it cannot be definitively determined whether the endocrine abnormalities identified at the four-year follow-up represent incident post-COVID conditions or the progression or detection of previously unrecognized subclinical disease.

Moreover, serial endocrine evaluations at earlier time points (2 and 3 years) were not feasible due to logistical constraints, delayed ethical approval for long-term follow-up, and limited patient availability. Consequently, the present study captures endocrine alterations at a single extended time point rather than their temporal evolution.

In addition, information on family history of diabetes, thyroid disease, or other autoimmune disorders was not available. Given the well-established role of genetic and familial susceptibility in autoimmune thyroid disease, the absence of such data may partly explain why no baseline clinical or biochemical predictors of thyroid autoimmunity were identified in this cohort, despite the high prevalence of thyroid autoimmunity observed at long-term follow-up.

The lack of data on SARS-CoV-2 reinfections and circulating viral variants during the four-year interval further limits the ability to attribute the observed long-term endocrine outcomes to a single SARS-CoV-2 exposure. Recurrent infections or variant-specific effects may have contributed to cumulative immune and metabolic stress, which could not be accounted for in the present analysis.

Additional limitations include the absence of detailed information on lifestyle changes and metabolic trajectories during the follow-up period. Pandemic-related factors such as reduced physical activity, weight gain, dietary changes, and psychological stress may have independently influenced long-term glycemic control and immune function, irrespective of direct viral effects [64].

In addition, unrecognized thyroid dysfunction may have contributed to long-term metabolic alterations, as impaired insulin action in adipose tissue and skeletal muscle has been demonstrated in hypothyroid states [65].

Finally, multivariable regression analysis was not feasible due to the limited number of outcome events and incomplete baseline information, particularly regarding pre-infection metabolic status and cumulative corticosteroid exposure. Under these constraints, adjusted models would have been statistically unstable; therefore, all associations reported should be interpreted as exploratory and hypothesis-generating.

Despite these limitations, this study provides one of the longest and most comprehensive endocrine follow-up evaluations conducted to date, integrating biochemical, immunological, and ultrasonographic assessments. The findings offer valuable insights into the potential long-term metabolic and autoimmune consequences of SARS-CoV-2 infection and emphasize the need for structured, multidisciplinary follow-up in COVID-19 survivors.

4.8. Future Directions

The findings of this study highlight several key priorities for future research on the long-term endocrine consequences of SARS-CoV-2 infection. First, to accurately define the temporal trajectory and causality of post-COVID endocrine dysfunctions, prospective longitudinal studies with predefined baseline metabolic and thyroid assessments (including fasting glucose, HbA1c, TSH, FT4, anti-TPO, and anti-Tg) are urgently needed. The absence of such baseline data during the pandemic was a major limitation in determining whether detected abnormalities were new or pre-existing. Standardizing these measurements in future clinical care pathways for hospitalized COVID-19 patients may substantially improve risk stratification and the interpretation of long-term outcomes.

Second, multicenter and population-based cohorts with larger sample sizes are warranted to validate the associations observed here and to examine potential modifiers such as age, sex, comorbidities, viral variants, and treatments received during the acute phase. International collaboration and the establishment of dedicated post-COVID endocrine registries would facilitate harmonized data collection and enable comparative analyses across diverse populations and healthcare settings.

Third, mechanistic studies exploring immune–endocrine interactions are needed to clarify the biological links between SARS-CoV-2 infection, persistent autoimmunity, and long-term dysglycemia. Investigations into molecular mimicry, chronic low-grade inflammation, β-cell vulnerability, and thyroid immune activation may help identify biomarkers predictive of long-term endocrine sequelae.

Finally, clinical research should evaluate the effectiveness of structured endocrine follow-up programs for patients recovering from moderate or severe COVID-19, particularly those with hyperglycemia at admission or elevated inflammatory markers. Such programs could guide timely lifestyle interventions or pharmacologic management, thereby reducing the long-term health burden associated with post-COVID endocrine dysfunction.

5. Conclusions

Four years after acute SARS-CoV-2 infection, this single-center follow-up study suggests that COVID-19 may exert sustained effects on endocrine homeostasis, even in individuals without previously known metabolic or thyroid disease. The high prevalence of long-term dysglycemia compatible with type 2 diabetes and persistent thyroid autoimmunity observed in this hospitalized cohort supports the concept that SARS-CoV-2 can act as a prolonged trigger of immunometabolic disturbances rather than a pathogen with purely transient physiological impact.

Acute clinical severity and stress hyperglycemia emerged as important correlates of long-term metabolic impairment, while nearly one-third of patients exhibited persistent thyroid autoimmunity at four years, suggesting a delayed or evolving autoimmune process potentially driven by post-viral immune activation. Because baseline HbA1c and thyroid antibody measurements were not available, these results should be interpreted as associations rather than definitive evidence of disease onset; nevertheless, the magnitude and persistence of these abnormalities highlight an important long-term clinical signal. Taken together, our findings indicate that long COVID includes a clinically relevant endocrine component involving both metabolic and autoimmune pathways.

Given the proportion of endocrine abnormalities detected at long-term follow-up, and the impossibility of distinguishing pre-existing from post-infectious dysfunction in the absence of baseline data, these results support the clinical rationale for including baseline metabolic (fasting glucose and HbA1c) and thyroid assessments (TSH, FT4, anti-TPO, anti-Tg) at the time of COVID-19 hospitalization. Establishing such standardized baseline measurements may substantially improve the accuracy of future risk stratification and facilitate early detection of delayed endocrine sequelae.

From a clinical perspective, this study underscores the need for structured, long-term endocrine surveillance in individuals hospitalized for COVID-19, particularly those who experienced moderate-to-severe disease or hyperglycemia at admission. Periodic assessment of glucose metabolism and thyroid function may facilitate early identification and management of delayed post-infectious abnormalities, potentially mitigating the long-term health burden associated with SARS-CoV-2 infection.