Twenty-five years of pediatric type 1 diabetes in Serbia: Trends in age at diagnosis, seasonal variation, and the proportion of diabetic ketoacidosis before, during, and after the COVID-19 pandemic

Abstract.

We aimed to examine 25-yr trends in the incidence, age at diagnosis, seasonal variation, and frequency of diabetic ketoacidosis among children and adolescents with newly diagnosed type 1 diabetes mellitus in Serbia, before, during, and after the COVID-19 pandemic. This nationwide, multicentre, study included all individuals aged ≤ 19 yr diagnosed with T1DM between 2000 and 2024 across 15 paediatric centres in Serbia. Clinical characteristics were compared using parametric and non-parametric tests, as appropriate. A total of 4,335 new T1DM cases were registered. The incidence increased significantly from 9.7 per 100,000 in 2000 to 25.5 per 100,000 in 2021 (annual percent change (APC) 3.12%, p < 0.001). Among children aged 0–14 yr, incidence rose until 2016 (APC 4.14%, p = 0.01), followed by a plateau through 2024, suggesting possible stabilization in younger age groups. Monthly variations in the frequency of newly diagnosed T1DM cases did not reach statistical significance (p = 0.876). Overall, 42.4% of patients presented with DKA, with a marked rise during the COVID-19 pandemic (47.7%) and persistently higher post-pandemic levels compared with pre-pandemic years. Over 25 yr, Serbia experienced a steady increase in paediatric T1DM incidence with recent stabilization among younger children. DKA remains persistently high in post pandemic years.

Highlights

● Pediatric T1DM incidence in Serbia showed a sustained increase over 25 yr, with a recent trend toward stabilization observed in the younger age group.

● No significant monthly variation in the number of newly diagnosed children with T1DM was observed during the study period.

● The age distribution of children at the time of T1DM diagnosis remained stable across the entire study period.

● A critically high proportion (42%) of children continue to present with diabetic ketoacidosis in the post-pandemic years.

Introduction

Type 1 diabetes mellitus (T1DM) is one of the most common chronic endocrine diseases in childhood and adolescence, with a continuously rising incidence worldwide. Data from European registries have shown an annual increase of approximately 3–4% in new cases over the past three decades, reflecting complex interactions between genetic predisposition and environmental triggers (1, 2). In Serbia, as in other Central and Southeastern European countries, the burden of T1DM among children and adolescents has also been increasing steadily. Previous national and regional studies (3,4,5) have described the incidence and prevalence of childhood T1DM in Serbia. However, none have analyzed the whole 25-yr period encompassing the pre-COVID pandemic and post-pandemic years. Moreover, long-term changes in age at diagnosis, monthly distribution of disease onset, and the proportion of patients presenting with diabetic ketoacidosis (DKA) have not been comprehensively evaluated. These parameters are critical indicators of disease recognition, healthcare accessibility, and public awareness over time.

Timely diagnosis of T1DM and prevention of DKA remain essential to reducing morbidity, hospitalizations, and healthcare costs. Understanding long-term trends in presentation patterns can help identify gaps in early detection and inform targeted public-health strategies, especially in the context of the COVID-19 pandemic, which may have influenced both the timing and severity of disease onset in children.

The present study aims to describe trends in newly diagnosed T1DM among children and adolescents aged ≤ 19 yr in Serbia over 25 yr (2000–2024). Specifically, we sought to analyze changes in the mean age at diagnosis, examine monthly variation in disease onset, and assess the proportion of patients presenting with DKA before, during, and after the COVID-19 pandemic.

Methods

In this study, we collected and retrospectively analysed data from all patients ≤ 19 yr with newly diagnosed T1DM across all regional centres that provide healthcare to this group in Serbia during the study period from January 1st, 2000, to January 1st, 2025. Data regarding subjects’ gender, date of birth, date of diagnosis, duration of symptoms, DKA at presentation (yes/ no), venous pH (bicarbonates when pH unavailable), glucose, and glycated hemoglobin (HbA1c) levels at diagnosis were collected retrospectively. All patients ≤ 19 yr old in Serbia have public health insurance. Newly diagnosed T1DM patients are immediately referred to a paediatric endocrinologist in the nearest regional secondary or tertiary healthcare centre. There are 15 regional centres: Mother and Child Health Care Institute of Serbia “Dr. Vukan Cupic” in Belgrade (MCHCI), University Children’s Hospital in Belgrade (UCH), Institute for Children and Youth Health Care of Vojvodina in Novi Sad, Children Clinic of Clinical Centre in Nis, Paediatric Clinic of Clinical Centre in Kragujevac, and centres in Kraljevo, Uzice, Subotica, Cuprija, Zajecar, Smederevo, Loznica, Zrenjanin, and Pancevo.

The diagnosis of T1DM and DKA was established at each center in accordance with ISPAD guidelines (6, 7) , and reevaluated during the database check. The diagnosis of DKA was established at each center in accordance with ISPAD guidelines, and reevaluated during the database check: hyperglycemia (blood glucose > 11 mmol/L) with acidosis (venous pH < 7.3 or bicarbonate < 15 mmol/L), with ketonemia or ketonuria. DKA severity was categorized as mild (venous pH < 7.3 or bicarbonate < 15 mmol/L), moderate (venous pH < 7.2, bicarbonate < 10 mmol/L) or severe (venous pH < 7.1, bicarbonate < 5 mmol/L).

The universal database form was sent to each regional centre, where the data were entered. Two sources of data were used: patients’ records from individual regional centres and hospital protocol records from each centre. The completed databases were returned to the Children’s Clinic at the Clinical Center of Nis. All databases were checked for duplicates and errors. Data from all centres were merged into one database containing all newly diagnosed paediatric T1DM patients in Serbia during the study period 2000–2024.

Informed consent was obtained from the parents or guardians of all participants for admission to hospitals and for the procedures performed during hospitalization. The Hospital Ethics Committee approved the study, and data were retrospectively collected in accordance with the 1964 Declaration of Helsinki.

The methodology of our data collection and patient acquisition is explained in detail in an already published paper (8). The total number of diagnosed patients during this study was 4,335.

Statistical analysis

Data were presented as mean and standard deviation (mean ± SD), along with minimum and maximum values. Following the assessment of data distribution normality, continuous variables between two groups were compared using the Student’s t-test or the Mann-Whitney U test, as appropriate. Comparisons of continuous values across three or more groups were performed using Analysis of Variance (ANOVA) or the Kruskal-Wallis test. The Chi-square test (χ² test) was employed for comparing categorical variables. Multivariate regression analysis was performed to estimate an association between DKA and demographic, clinical and pandemic parameters.

The standardized incidence rate was calculated using the direct standardization method, utilizing age-specific rates and the age distribution of the World Health Organization (WHO) standard population. For each age group, the crude rate per 100,000 population (number of new cases divided by the corresponding population, multiplied by 100,000) was calculated. The resulting rates were then weighted according to the respective WHO proportions to obtain weighted contributions. The weighted contributions were summed and divided by the total WHO proportion for the observed age groups, and the result was expressed as the standardized incidence rate per 100,000 population.

Trend analyses of the number of newly registered cases and the standardized incidence rate over the observation period were conducted using the Joinpoint Regression Program, Version 5.4.0.0 (Surveillance Research Program, National Cancer Institute, USA). The analysis was initiated with a model without joinpoints (or “break points”). The optimal number of joinpoints was determined by comparing models of different complexity using the Weighted Bayesian Information Criterion (WBIC). For each segment between the joinpoints, the Annual Percent Change (APC) was calculated. Statistical significance was assessed at a two-sided p-value of < 0.05.

Results

A total of 4,335 new cases of T1DM were registered among children and adolescents aged ≤19 yr in Serbia during the 25-yr study period (2000–2024).

Joinpoint regression analysis showed a statistically significant increasing trend in the number of newly diagnosed T1DM cases, with an Annual Percent Change (APC) of 1.93% (95% CI 1.17–2.69, p < 0.001) over the 25 yr (Fig. 1).

Fig. 1.

Age-standardized incidence rate of T1DM ≤14 yr per 100,000 Population, 2000–2024.

The age-standardized incidence rate increased from 9.7 per 100,000 in 2000 to a maximum of 25.5 per 100,000 in 2021 (Table 1). The overall trend in incidence was statistically significant (APC = 3.12%, 95%CI 2.29–3.95, p < 0.001), showing a steady rise until 2024.

Table 1. Age-standardized incidence rate of T1DM ≤ 19 yr per 100,000 Population, 2000–2024.

When analyzed by age group, a significant upward trend was evident in children aged 0–14 yr between 2000 and 2016 (APC = 4.14%, 95%CI 3.09–10.54, p = 0.010), followed by a plateau through 2024 (p = 0.853), suggesting potential stabilization in younger age groups after 2016 (Fig. 2).

Fig. 2.

Age-standardized incidence rate of T1DM ≤ 19 yr per 100,000 Population, 2000–2024

Of all patients, 53.2% were male and 46.8% female, with a mean age at diagnosis of 9.56 ± 4.32 yr (range 0.5–19.0). The distribution across age categories was relatively balanced: 41.8% were aged 5–11 yr, and 40.1% were aged >11 yr. There were no statistically significant differences in the frequency of age categories over time (p = 0.117) (Fig. 3).

Fig. 3.

Monthly distribution of newly diagnosed children with T1DM during the study period (Kruskal-Wallis test, p = 0.876).

Monthly analysis revealed no significant variation in the number of new T1DM cases across months (p = 0.876), suggesting an even distribution throughout the year.

Overall, 1,268 (42.4%) children presented with diabetic ketoacidosis (DKA) at diagnosis. The frequency of DKA varied significantly over time (χ², p < 0.001) (Table 2). The highest proportion was observed during 2000–2006 (55.8%), followed by a marked decline in 2007–2018 (» 39%), and a transient increase during the COVID-19 pandemic (47.7%). Although a slight reduction was observed after 2022 (42.4%), post-pandemic rates remained higher than the pre-pandemic decade average.

Table 2. Frequency of DKA at diagnosis in different time periods.

Patients presenting with DKA were significantly younger (mean 8.78 ± 4.49 yr) compared to those without DKA (9.91 ± 4.23 yr; p < 0.001). One in four patients with DKA (25.1%) was younger than 5 yr, compared with only 15.5% in the non-DKA group (p < 0.001). No sex differences were observed in DKA occurrence (p = 0.377).

Children with DKA had a shorter symptom duration prior to diagnosis (18.4 ± 18.9 d vs. 21.4 ± 23.1 d; p = 0.001), higher mean glucose (26.15 ± 8.21 mmol/L) and HbA1c (11.98 ± 2.24%) compared to those without DKA (p < 0.001 for both) (Table 3).

Table 3. Demographic and clinical characteristics according to the presence of DKA.

Further analysis by DKA severity showed that most patients had mild DKA (41.8%), followed by severe (32.9%) and moderate (25.3%) forms. Age and metabolic parameters (pH, glucose) differed significantly across severity groups (all p < 0.001), with younger children showing more severe biochemical decompensation. The multivariate model showed an association between the occurrence of DKA and the age of the subjects (OR 0.948, p < 0.001), as well as the duration of symptoms (OR 0.995, p = 0.009) (Table 4).

Table 4. Multivariable analysis between DKA and demografic, clinical and pandemic characteristics.

Discussion

In this 25-yr national study of children and adolescents aged ≤ 19 yr, we found a continuous increase in T1DM incidence in Serbia, peaking in 2021. The overall trend remained significantly upward, consistent with global and regional data showing a persistent rise in pediatric T1DM over recent decades (1,2,3,4,5). However, age-stratified analysis revealed an important distinction. While incidence increased steadily among adolescents, a plateau was observed in children younger than 14 yr after 2016. A similar pattern was observed in several population-based registries. In Finland, incidence among young children decreased after 2003 (9), whereas in Germany, Sweden, Norway, and (10,11,12), earlier continuous increases subsequently levelled off at high incidence rates. Some of these countries experienced a transient spike during the COVID-19 pandemic. They returned to their previous levels, supporting the theory of an epidemiological plateau rather than a permanent pandemic-driven shift. In contrast, neighboring Balkan countries and the United States continue to report rising childhood incidence, with no signs of stabilization (13, 14).

The stabilization of T1DM incidence in Serbian children younger than 14 suggests the achievement of “epidemiological maturity.” This indicates that the impact of urbanization and lifestyle changes has reached a plateau, exhausting the “pool” of genetically susceptible children who would develop the disease at an early age. The continued rise in the adolescent group aged 15 and older likely reflects a delayed clinical manifestation of the disease, influenced by pubertal physiological changes rather than new environmental triggers.

The proportion of patients presenting with DKA remained high (42.4%), with significant temporal variability. The marked decline in DKA during 2007–2018 was followed by a transient but pronounced rise during the COVID-19 pandemic (2020–2021), after which rates declined modestly but remained above pre-pandemic levels. Similar observations have been reported internationally, with increased frequency and severity of DKA at T1DM onset during the pandemic years (15, 16).

Several mechanisms may explain this phenomenon. Firstly, public-health restrictions, reduced access to primary care, and parental hesitancy to seek medical attention likely contributed to delayed diagnosis. Secondly, SARS-CoV-2 infection itself may have accelerated disease progression in genetically susceptible individuals. It has been hypothesized that viral infection and associated inflammatory responses could precipitate the transition from preclinical stages (Stage 2) to overt T1DM (Stage 3) by inducing β-cell stress or direct viral involvement of pancreatic tissue (17, 18). Thus, the pandemic likely acted as both a social and biological trigger for earlier disease manifestation and more severe metabolic presentation.

Delayed diagnosis of DKA in children may be related to low clinical suspicion and the nonspecific nature of early symptoms. In Serbia, blood glucose measurement and urine testing are not routinely performed in primary health care. Improving access to glucometers and urine test strips could potentially facilitate earlier recognition of diabetes and may help reduce the risk of severe presentation, although this warrants prospective evaluation.

Younger age remained the strongest predictor of DKA, with children £ 5 yr showing more severe acidosis, shorter symptoms duration, higher glucose and HbA1c levels at diagnosis. This pattern is consistent with previous reports emphasizing diagnostic challenges and nonspecific symptom recognition in preschool-aged children (19). The absence of monthly variation in our study aligns with findings from some southern European countries (20), but differs from the results of the Sweet Study (21). Although no significant monthly differences were observed, it should be acknowledged that the applied method may have limited sensitivity for detecting subtle cyclical trends.

Changes in clinical practice, such as improved laboratory accessibility (including wider availability of HbA1c testing), evolving emergency department triage protocols, and refinements in referral pathways, may have influenced the timing of diagnosis and/or DKA detection. Although these changes did not involve modifications of diagnostic definitions, they could have contributed to earlier recognition of hyperglycemia or altered case ascertainment over time.

Taken together, our findings highlight that, despite epidemiologic progress, DKA remains an important public-health concern. Sustained post-pandemic elevation in DKA frequency underscores the need for renewed awareness campaigns targeting parents, schools, and primary care providers, and for standardized diagnostic pathways that ensure early recognition of hyperglycemia in children.

The major strengths of this study include its nationwide coverage, long observation period, and standardized trend analysis, which provide reliable insight into temporal patterns of T1DM incidence and presentation in Serbia. Limitations include its partly retrospective design, incomplete biochemical data in earlier years, and absence of immunological and socioeconomic variables.

Conflict of interests

All authors have approved the submission and declare no conflicts of interest.