This study was a retrospective cross-sectional analysis of cases, including all instances of first-time diagnosis of T1D identified in the medical records of Vietnam National Children’s Hospital. The study period was from January 1, 2023, to December 31, 2023. The inclusion criteria were (1) aged between 6 months and 17 years; (2) diagnosed with DM for the first time upon hospital admission, with either hemoglobin A1c (HbA1c) >6.5% or classic symptoms of diabetes or hyperglycemic crisis with plasma glucose concentration ≥11.1 mmol/L (200 mg/dL), fasting plasma glucose ≥7.0 mmol/L, or 2-hour postload glucose ≥11.1 mmol/L (≥200 mg/dL) during an OGTT (oral glucose tolerance test) [13]; (3) classified as T1D, defined as early-onset in a child not overweight or obese or with insulin resistance symptoms and/or who was positive for pancreatic islet autoantibodies.

The study excluded participants with diabetes due to another disease: any other form of nonautoimmunerelated insulin-dependent diabetes, drug-induced diabetes, diabetes in patients with thalassemia, leukemia, monogenic diabetes, or exocrine pancreatic insufficiency. The study design and inclusion/exclusion criteria are illustrated in Fig. 1.

We collected demographic and clinical information from medical records, including age at diagnosis, gender, address, self-referred or transferred from another hospital, family history of DM, and diabetes-related symptoms such as polydipsia, polyuria, polyphagia, unintentional weight loss, diabetic ketoacidosis, and ketonuria. Laboratory results were also recorded, including HbA1c levels and pancreatic islet autoantibody status.

Anthropometric measurements, including weight and height, were performed at diagnosis and converted to z-scores based on World Health Organization (WHO) Child Growth Standards (2007). Subsequently, the researchers calculated the z-scores for weight-for-height or body mass index (BMI)-for-age. For children younger than 5 years, overweight is defined as weight-for-height greater than 2 standard deviations (SDs) above the WHO Child Growth Standards median, and obesity is weight-for-height greater than 3 SDs above the median. For children aged 5–19 years, overweight is BMI-for-age greater than one SD above the WHO Child Growth Standards median, and obesity is more than 2 SDs above the median [14]. Family history was categorized into 2 levels: First-degree relatives including parents and siblings diagnosed with DM and second-degree relatives of grandparents, aunts, or uncles with the diagnosis of DM.

The criteria for diagnosing DKA were glucose levels >11 mmol/L, pH <7.3 or bicarbonate <18 mmol/L, and elevated blood ketones or ketonuria according to ISPAD 2022 guidelines [15]. HbA1c levels were recorded from all results performed at our qualified laboratory at the time of diagnosis. Ketonuria was documented from electronic records or referral documents, depending on which data were collected closest to the time of diagnosis.

Autoantibody values were retrieved from electronic records, with reference ranges provided by the Biochemistry Department of Vietnam National Children's Hospital. Autoantibodies were measured using various assays: anti-insulinoma-associated–2 (anti-IA2) (measured by Maglumi) was considered positive if ≥28.0 U/mL, insulin autoantibody (IAA) (measured by Maglumi) was positive if >20.0 U/mL, anti-zinc transporter 8 (anti-ZnT8) (ELISA) was positive if >15.0 U/mL, anti-glutamic acid decarboxylase (GAD) 65 (measured by Maglumi) was positive if >17.0 U/mL, and islet cell antibodies (ICA) (measured by Maglumi) was positive if >28.0 U/mL. Thyroid autoantibodies were detected using quantitative electrochemiluminescence immunoassay with Roche Cobas. Positive results for Trab (thyroid-stimulating hormone receptor antibody), thyroid peroxidase antibody, thyroglobulin antibody were defined as values greater than 1.22 IU/L, 34 IU/mL, and 115 IU/mL, respectively.

The study utilized the 2023 population report from the General Statistics Office of Vietnam [16] and the projected child population distribution across regions in 2023 [17] to estimate the incidence rate for each area in Northern Vietnam, which includes the Red River Delta, the northern midlands, and mountainous areas.

All new T1DM cases were identified during outpatient or emergency visits prior to hospitalization. Because of the limited population data, we also calculated the clinical incidence rate using the total number of new T1DM cases divided by the annual number of outpatient and emergency visits.

Mean value (±SD) and median (interquartile range) for continuous variables were calculated according to distribution, while proportions were estimated for categorical variables. The normality of continuous data was assessed using the Shapiro-Wilk test, with P<0.05 indicating a significant deviation from normal distribution. To compare continuous variables, we used the independent sample t-test and the Mann-Whitney U-test based on the distribution of the variables and the number of groups being compared. Categorical variables were compared using the chi-square test. A P-value less than 0.05 was deemed statistically significant.

This retrospective study was conducted according to the Helsinki Declaration and was approved by the Institutional Review Board (IRB) of Vietnam National Children's Hospital (2093/BVNTW-HDDD). As a retrospective review of medical records, all information was confidential, and the study did not influence the diagnostic or treatment processes of the patients involved.

Our study included 64 children newly diagnosed with T1D, with a mean age of 9.1±3.74 years, ranging from 1.05 to 16.16 years. Among them, there were 34 male children (53.1%) and 30 female children (46.9%), a male-to-female ratio of 1.1:1. Most patients (48.4%) were 6 to less than 11 years of age at diagnosis, 32.8% were 11 years or older, and 18.7% were younger than 6 years. Males comprised 53.1% of the cohort. A family history of diabetes was noted in 17 children, including 3.1% with first-degree and 23.4% with second-degree relatives affected. Anthropometric data showed a median height z-score of -0.55 (range, -3.85 to 1.44) and a BMI z-score of -0.63 (range, -5.82 to 4.38).

The distribution of patients across regions reveals notable variations. The Red River Delta, excluding Hanoi, had the highest number of patients, accounting for 45.3% of the total cases. The Hanoi capital region followed with 31.2%, while the northern midlands and mountain areas contributed 23.5% of the cases. However, the estimated incidence rate per 1,000,000 children does not align with this pattern. The estimated incidence rate in the northern region is 0.77 per 100,000 children, with the highest rate observed in Hanoi, followed by the Red River Delta region (Table 1, Fig. 2).

From 2020 to 2023, the number of newly diagnosed T1D cases increased slightly from 53 to 64 despite substantial fluctuations in annual total visits. The clinical yearly incidence rate per 100,000 visits peaked in 2021 at 8.07, possibly reflecting reduced healthcare access during the coronavirus disease 2019 (COVID-19) pandemic. From 2022 onwards, this rate stabilized at 5.21–5.39 (Table 2).

The clinical characteristics of our cohort reveal distinct patterns in symptom presentation, referral, nutritional status, and diagnostic indicators (Table 3). Most children, 89.1%, presented with polyuria-polydipsia syndrome accompanied by weight loss, highlighting it as a typical symptom profile of late diagnosis. In contrast, only 10.9% had polyuria-polydipsia syndrome without weight loss.

Regarding admission, 62.5% of children were referred from other health facilities, while 37.5% were admitted directly, indicating a higher referral tendency. Nutritionally, 15.6% of children were underweight, and 23.5% were overweight or obese, reflecting a significant proportion of overweight or obesity within the cohort.

Diabetic ketoacidosis at diagnosis was present in 57.8% of cases, signifying that a significant proportion of patients experienced severe metabolic derangement upon diagnosis. Additionally, 85.5% tested positive for ketonuria and 93.2% for glycosuria, indicating high prevalence of ketones and glucose in urine. The mean hemoglobin A1c at diagnosis was 13.0%±2.85%, underscoring a high level of disrupted glycemic control at the time of diagnosis.

The comparison between diabetes ketone acidosis (DKA) and non-DKA groups among newly diagnosed T1D patients revealed no significant differences in gender, age, BMI z-score, HbA1c, antibody positivity, or family history (P>0.05). Although slightly higher referral rates and HbA1c levels were observed in the DKA group, these differences were not meaningful.

Due to challenges following the COVID-19 pandemic, the availability of reagents for antibody testing in our facility was inconsistent and irregular. As a result, we could only perform ZnT8 antibody quantification for 60 children, GAD antibody testing for 58 children, ICA testing for 56 children, IA2 antibody testing for 41 children, and IAA testing for 40 children. The results show that 48 of 64 children (75%) were positive for at least one antibody. The positivity rate for each antibody is shown in Fig. 3.

Among the 64 children diagnosed with T1D, 35 underwent testing for all 4 antibodies: ZnT8, GAD, ICA, and IA2. The proportion of these children testing positive for at least one antibody was 27 of 35 (77%). The proportion of children positive for exactly one antibody was 11.4%; 2 antibodies, 17.1%; 3 antibodies, 14.3%; 4 antibodies, 34.7%; and all 5 antibodies, 22.9%.

Sixteen cases were incomplete (13 cases) or underwent no antibody testing (3 cases). This group was diagnosed at a mean age of 9.74±3.4 years. Clinical manifestations comprised polydipsia (87.5%), polyuria (81.2%), and weight loss (38.4%). Diabetic ketoacidosis was present in 6 cases (37.5%). The mean BMI was 15.0±2.0 kg/m² (range, 11.9–20.2), and the BMI z-score was -1.2±1.5 (range, -4.3 to 1.3). Diagnostic laboratory findings included a mean HbA1c of 13.0%±2.9% (range, 10%–18.7%), fasting blood glucose of 21.2±6.7 mmol/L, and a positive urine ketone ratio of 75%. All these cases were treated with insulin at an average dose of 0.99±0.55 (range, 0.3–1.7) IU/kg/day for an average duration of 17.3±3.7 (range, 13–23) months.

Vietnam is located in Southeast Asia; Northern Vietnam comprises the Red River Delta, northern midlands, and mountainous region. The capital, Hanoi, is located in the Red River Delta [18]. The National Children's Hospital is situated in Hanoi capital and provides healthcare services to children across Northern Vietnam. As the hospital is a referral center, our estimated incidence likely approximates the actual rate, with only a small number of cases diagnosed or managed at lower-tier hospitals.

The estimated incidence of T1D was 0.8 per 100,000 children, which is among the lowest globally. In contrast, countries such as Finland report markedly higher rates, with an incidence of 52.2 per 100,000 [19]. This substantial difference may be attributed to variations in genetic predisposition, such as the frequency of T1D-associated human leukocyte antigen (HLA) alleles, and diverse environmental factors.

In terms of clinical incidence, our study observed a rate of 5.39 per 100,000 hospital visits, which is notably higher than the 3.16 per 100,000 reported in China [20]. This difference may reflect disparities in healthcare access, diagnostic practices, or case ascertainment methods between the 2 regions.

The Red River Delta, excluding Hanoi, accounted for more than 45.3% of the cases, indicating a potential concentration of T1D children in this densely populated area. This was followed by the Hanoi capital region (31.2%), suggesting that urban centers also have a higher incidence or greater access to healthcare facilities capable of diagnosing T1D. The northern midlands and mountainous areas contributed 23.5% of cases, possibly reflecting lower incidence and healthcare access challenges. However, the incidence rate was highest in Hanoi (0.91 of 100,000 children), likely due to the capital's high literacy rate, well-developed healthcare infrastructure, and easier access to medical services. The high referral rate (62.5%) from other hospitals underscores the need for enhanced T1D management capabilities at lower healthcare levels.

The mean age of T1D diagnosis in our cohort was 9.1± 3.74 years, consistent with data from southern Vietnam reported by Quynh Thi Vu Huynh (2023) [8], indicating a relatively uniform age of onset across the country. However, this is significantly older than the ages reported in studies from Thailand (7.62 years) [21] and China (7.1 years) [22], suggesting potential differences in genetic factors, healthcare access, or detection practices.

In our study, a male predominance (53.1%) was observed, which contrasts with the female predominance reported in Greece (68%) [23]. The reason for this phenomenon was not clear, but it may be related to gender composition differences and complex environmental factors.

A family history of diabetes mellitus was noted in 26.56 % of patients, slightly higher than in southern Vietnam (19.3%) [8] and considerably lower than in the USA (31.2%) [24] but higher than in Finland (21.1%) [25]. This can be explained by the complex pathogenesis of T1D, which involves interactions between genetic predisposition and environmental factors. This leads to T-cell-mediated destruction of insulin-producing pancreatic beta cells. The rising global obesity pandemic and the increasing and ageing population compound this [4]. Additionally, the risk of T1D is 8–15 times higher in people with affected first-degree relatives [26-29] and 2 times higher in those with second-degree relatives [26,30]. Another important influencing factor is HLA; children with familial T1D more frequently carry the DRB1*0401/2/4/5-DQA1*0301-DQB1*0302 (DR4-DQ8) haplotype than other children (74.0% vs. 67.0%, P=0.02) [25].

The median BMI z-score of our cohort was -0.63 SD, which was lower than that reported in China (-0.22 SD) [20]. However, in our study, 23% of the cases were overweight or obese. The connection between childhood and adolescent obesity and the increased incidence of T1D has been established [31,32]. Higher BMI percentiles are positively associated with incident T1D among adolescents (16 to 19 years), with approximately 25% greater risk for each incremental SD in BMI [33]. Validated with a T1D genome-wide association study, Mendelian randomization corroborated a causal role for higher childhood body size on T1D risk, with an odds ratio (OR) of 1.9 (95% confidence interval [CI], 1.2–3.1) [33]. Interestingly, the study predicted a ~ 22% reduction in T1D cases if children with severe obesity reduced their body weight by ~10%, suggesting the existence of a critical window to mitigate T1D [33].

The frequency of DKA at diagnosis was 57.81%, which is lower than the rates reported in China and Turkey (64.5%, Huang et al. [22]; 62.96%, Kandemir et al. [34]). The incidence of DKA is generally high across countries, and many studies have identified young age, ethnicity, rural residence, and lack of health insurance as significant risk factors for DKA [35]. Four studies conducted in the United States, France, and Poland specifically examined the outcomes of children who were not diagnosed with T1D during their first medical consultation due to diagnostic errors. These errors occurred when children were misdiagnosed or signs and symptoms were overlooked or unrecognized during their first visit. These children had a threefold increased risk of presenting with DKA (combined OR, 3.35 [2.35 to 4.79]; P<0.001, I²=0%) [35]. When comparing the DKA and non-DKA groups, there were no differences in gender, HbA1c levels, family history, or typical clinical manifestations between the 2 groups. However, there were significant differences regarding age at onset, BMI, and referral rates. The DKA group had a younger age at onset, higher BMI, and higher referral rates than the non-DKA group. The higher referral rate may be related to the insufficient understanding of the disease by the general public and even pediatricians and the more challenging diagnosis at younger ages. This indicates the need to improve early detection and intervention protocols to reduce severe presentations.

The main symptoms before the hospital visit were polyuria,-polydipsia (89.1%), followed by weight loss (57%). A study from Taiwan showed that the most common symptoms were polyuria (96%), polydipsia (92%), dry lips (81%), and weight loss (79%) [36]. The proportion of children presenting with weight loss was significantly higher compared with reports from other regions, such as China (32.94%) [37] and Brunei (72.4%) [38]. These findings highlight delays in T1D diagnosis in Vietnam, emphasizing the need for earlier recognition of symptoms.

The mean HbA1c level at diagnosis was 13.03%±2.85%, reflecting substantial chronic hyperglycemia, higher than levels observed in Brunei (11.5%) [38] and similar to those in Thailand (13.1%) [21]. The elevated HbA1c levels suggest that many patients in our study experienced prolonged undiagnosed T1D, highlighting potential delays in diagnosis. This may also explain the higher rates of polyuria, polydipsia, and DKA observed.

The antibody positivity rate among newly diagnosed T1D patients in our cohort was 75%, which is notably higher than those reported in similar studies from other developing regions. For instance, Karavanaki et al. [23] reported an antibody positivity rate of only 25.6% in Athens, Greece, Similarly, Li et al. [20] observed a rate of 41.8% across 34 medical centers in 25 major cities in China, while another Chinese center reported a prevalence of 51.22% [39]. Studies in other Asian populations further highlight these variations. In Taiwan [40], 89.4% of children had positive GAD antibody or IA2 within the first year of diagnosis, with individual rates of 66.3% for GAD antibody and 65.3% for IA2. Similarly, a study in Iran reported that 81.7% of pediatric patients tested positive for at least one autoantibody [41]. The differences in the rates of autoantibodies across countries can be attributed to a combination of genetic, environmental, and methodological factors. Genetically, the prevalence of high-risk HLA alleles, strongly associated with T1D, varies significantly between populations. Environmental factors also play a crucial role, as environmental triggers can interact with genetic susceptibility. These factors include viral infections, dietary components, and other lifestyle factors that may differ between countries. Methodological differences in detecting and reporting T1D autoantibodies can also contribute to the observed variations. Differences in study design, population selection, and laboratory techniques can lead to discrepancies in reported autoantibody prevalence. Standardizing these methodologies is crucial for accurate comparisons across studies and populations.

In our cohort, there were 16 cases in which incomplete (13 cases) or no antibody testing (3 cases) was performed. All cases had a lean physique, with an average BMI of 15.0; none were overweight or obese. Six cases presented with DKA, and all required insulin treatment for at least 13 months. Typically, for cases without antibody results, the diagnosis of T1D can be based on the C-peptide level. The American Diabetes Association [42] recommend including C-peptide levels in the diagnostic process, mainly when the type of diabetes is unclear. Similarly, the National Institute for Health and Care Excellence [43] highlights the value of C-peptide in distinguishing between diabetes types, especially in individuals with atypical presentations. Clinical guidelines stress combining C-peptide results with other diagnostic criteria and clinical observations to differentiate T1D from T2D and decide on appropriate treatment. Recent studies revealed important relations between C-peptide concentrations in plasma and the characteristics of patients with T1D. Individuals with lower levels of plasma C-peptide were more likely to develop T1D in a prediagnostic state. In contrast, those with higher levels of C-peptide had features more closely associated with T2D [44-46]. However, due to the impact of the COVID-19 pandemic, we could not procure the reagents needed for C-peptide testing. As a result, these cases also lacked C-peptide results. Our diagnosis was entirely based on clinical progression and treatment methods.

Our study has several limitations that should be noted. Conducting the research within a single tertiary center may limit its generalizability to broader populations. Moreover, some cases may have been managed at provincial or adult endocrinology hospitals, potentially due to geographic barriers or milder disease presentations, which could influence the total number of reported cases. The lack of data on critical clinical markers, such as C-peptide levels, and the absence of long-term follow-up data hamper the ability to comprehensively evaluate the clinical spectrum and outcomes of T1D within this cohort. Additionally, due to reagent shortages and reliance on retrospective record reviews, the limited availability of autoantibody testing may adversely affect diagnostic accuracy. Estimates of incidence derived from population projections could also introduce variability. Last, the relatively small sample size of 64 patients limits the statistical power for subgroup analyses, including comparisons between patients presenting with and without DKA.

In conclusion, this study highlights the clinical incidence and characteristics of newly diagnosed T1D in a tertiary pediatric center in Vietnam, emphasizing challenges in early detection. The overall incidence rate in Northern Vietnam was 0.77 per 100,000 children, with significant regional differences across the Red River Delta, Hanoi capital, northern midlands and mountainous areas. In 2023, the clinical incidence was 5.39 per 100,000 visits, with 57.81% of cases presenting with DKA and reflecting diagnostic delays. Additionally, 89.1% of cases showed polyuria-polydipsia syndrome, and antibody positivity with at least one autoantibody was found in 75%. Regional variations suggest the influence of genetic, environmental, and healthcare factors. Improving diagnostic capabilities and access to care is crucial for better patient outcomes.