The incidence of type 1 diabetes in children under 15 years of age is rising again—a nationwide study

Type 1 diabetes (T1D) is a chronic autoimmune condition which has been shown to be rising globally for many decades, yet the exact reasons for this remain unclear [1, 2]. Reported incidence rates in the literature between regions and countries can be observed in varied states: falling, plateauing or rising [1, 3‐11].

During recent decades, Finland has consistently reported the highest incidence rate (IR) of T1D in children, peaking at 64.9/100,000/year in 2006 [12, 13]. Finland’s children experienced a decrease in IRs (overall IR reducing to 52.2/100,000/year in 2015–2018) after appearing to level off during 2003–2018 [1].

A similar trend has been noted in Ireland as in other high incidence countries. Ireland experienced a sharp rise in incidence between 1997 and 2008, followed by an apparent stabilisation of rates [3, 4, 14]. The Western Australian Children’s Diabetes Database noted no significant change in their temporal trends (2003–2016) following a peak in 2003 [3]. A Welsh study on changing incidence looking at data from 1990 to 2019 found that their incidence rates peaked around 2010 at 31.3/100,000/year and then declined [5]. They concluded that the risk of developing T1D under 15 years was not increasing. The country reported to have the second highest IR worldwide is Sweden [13]. Swedish rates were reportedly plateauing in 2011 [6]; however, in a later report, rates were noted to be rising and not stabilising [7].

Most centres reporting persisting increases in their IR of T1D were those with either previously low incidence rates or those recently experiencing economic changes [8‐10]. Hong Kong, as a low incidence country, reported increasing rates in children up to 18 years (4.3/100,000/year) detailing data from 2008 to 2017, almost double that of the previous decade although the rate of rise slowed towards the end of the study [11]. However, other centres, such as the Calabria region of Italy, maintained and increased their very high rates, with Calabria showing a marked increase in the last 2 years of their 3-year study (2019–2021) [9]. The German (DPV) registry which collects data on children up to 18 years of age found their IRs increased from 2011 to 2019 and this trend continued into 2020 as predicted [15, 16].

Fluctuation in the incidence of T1D within populations has long been recognised with some centres showing a periodicity in incidence [17]. Monitoring changes in the incidence of T1D internationally is important as it may help shed light on the elusive mechanism(s) triggering this poorly understood disease. The apparent stabilisation (or decline) of incidence rates in a number of high incidence countries, including Ireland, raised the possibility that a saturation point of diabetes had been reached in these populations [1, 3, 5, 14]. However, as Sweden with the second highest incidence of T1D in the world has demonstrated rising incidence rates following a previous plateau, it is important to examine recent data relating to the current status of T1D incidence, particularly in high incidence countries to ascertain if the incidence has remained stable, is falling or is increasing again.

A wide range of environmental triggers, including viruses, obesity and lack of vitamin D, have been postulated to account for these global differences and fluctuating rates [1, 18, 19]. A recent study from the University of Dublin, Trinity College, highlighted that up to one quarter of children living in Dublin (Ireland’s capital city) have low vitamin D levels [20]. Low levels of vitamin D have long been postulated to be associated with T1D progression/development and particularly the north–south gradient in T1D incidence. However, the vitamin D data are somewhat conflicting [21, 22].

A related factor is seasonal variation at diagnosis which is likely influenced by temperature and viruses. Harvey et al. in their 2021 study in Wales (UK) looked at the effect of seasonality and found a greater effect of seasonality at diagnosis with higher rates of diabetes in winter [5]. They noted that their meteorological data showed increasing hours of sunlight since 1980 and postulated that this was associated with stabilisation in their IR due to the effect of sunlight on vitamin D levels [5]. A Swedish study found a stronger association with low mean temperature than hours of sunshine with the development of T1D in children [23].

The primary aim of this study is to determine the incidence rate of T1D in children in Ireland aged under 15 years in the period 2019 to 2021 and compare it with the robust annual incidence data of the Irish Childhood Diabetes National Register (ICDNR) established in 2008, in order to confirm if the previously noted stabilisation of IR over recent years has persisted, reflecting a true stabilisation or whether the incidence is rising again in this high incidence country. Age, sex and season of diagnosis were also evaluated to ascertain if they had changed over the period.

New or incident cases of T1D for the period 2019 to 2021 were reported using the established methodology of the Irish Childhood Diabetes National Register (ICDNR), which continues to work closely with all participating centres nationally responsible for the diagnosis and treatment of children with T1D in Ireland. Details regarding the ICDNR have been published previously and are unchanged, apart from the merger of one of the smaller centres with a larger centre in 2019 [14, 24]. Briefly, this involves detailed prospective reporting of incident cases of T1D from all 19 paediatric centres caring for children nationally, case verification and the application of capture-recapture methodology (with the support of the Primary Care Eligibility and Reimbursement Service (PCERS)) [25]. The case definition employed is similar to that of the WHO Diabetes Mondial study and Eurodiab Tiger study group [26, 27]. Inclusion criteria, as in previous years, were physician diagnosed new onset T1D in a child aged under 15 years occurring in the period from January 1st to December 31st and resident in Ireland at the time of diagnosis. The date of diagnosis was taken as the time of first insulin injection. Cases were excluded if children were aged over 15 years or not resident in Ireland at diagnosis, if diabetes was diagnosed outside the study period or if diabetes was not T1D, such as monogenic or type 2 diabetes. Written informed consent was obtained from parent(s) of each child participant. Ethical approval for this study was granted by the SJH/AMNCH Joint Research Ethics Committee in accordance with the Declaration of Helsinki.

The age categories chosen for analysis are 0–4.99 years (under 5), 5–9.99 (5–10 years) and 10–14.99 years (under 15 years). Age and sex category specific incidence rates were calculated using population data from the Central Statistics Office (CSO) [28]. Rates are reported per 100,000 of population for children under 15-year-old (in 5 year groupings) per year. Standardised incidence rates (SIRs) for age and sex were calculated to permit comparison between populations internationally and over time. The direct method of standardisation was employed using a common standard population comprising equal numbers in the six age and sex categories [29]. The results are reported for males, females and all children combined.

Seasons were defined using the meteorological classification of spring (March, April and May); summer (June, July and August); autumn (September, October and November) and winter (December, January and February). This is in line with Europe and the World Meteorological Organisation. Of note the Celtic Irish cultural classification is slightly different wherein each season begins a month before the corresponding meteorological season, i.e. winter commences in November and spring in February [14, 30]. Seasonality data were calculated using the number and percentage of cases diagnosed per season for each year.

Data were analysed using IBM Statistical Package for Social Sciences (SPSS) version 27.0 [31]. Exact confidence intervals for category specific and standardised rates were obtained using the freely available programming language R version 4.2.2 [32]. Chi-square tests and F-tests were used where appropriate to compare groups. A Poisson regression was run on the category specific rates as the dependent variable with the log (population) included as an offset to examine the relationship between age category, year and gender. Each variable was treated as a categorical variable.

In the 3-year period, 2019 to 2021, a total of 1027 children (542 male) aged under 15 years were notified and verified as incident cases of T1D to the ICDNR for calculation of incidence rates. An additional case was identified but subsequently reclassified as having monogenic diabetes and excluded. All paediatric centres nationally notified and offered registration to their eligible children. A number of children, 28 (2.7%), did not complete full registration for a variety of reasons, either they declined or failed to complete the full registration process post-discharge from hospital admission despite multiple reminders. However, anonymous baseline data for gender age and date of diagnosis was provided for all notified cases to enable calculation of IRs and seasonality.

Annual ascertainment rates for the ICDNR were at least 93% in 2019; this is an underestimate, as secondary ascertainment could not be undertaken on 6% of cases where only anonymous baseline data were provided. Thus, the ascertainment rate for 2019 is likely at least 95% and could be as high as 99%. Ascertainment for 2020 and 2021 are incomplete but also will likely be in excess of 93%.

This study demonstrates a significantly increased incidence of T1D during the years 2019–2021. The previously observed stabilisation and subsequent minor reduction in the incidence of T1D in the Irish childhood population, suggesting a saturation in childhood diagnoses, has not been maintained [14, 24]. In the period (2019–2021), the direct standardised rate increased by 21% with an average percentage increase of 10%. The trend in the direct standardised incidence rate in the Irish population for the last 14 years (2008–2021) shows a significant increase in the incidence rate (Fig. 1). Of note this increase in the Irish population occurred in 2019 and preceded the COVID-19 pandemic. A similar pattern of apparent stabilisation or fall in incidence rates followed by a subsequent rise in incidence has also been noted by some other high incidence countries, particularly Sweden and more recently reported for Finland [6, 7, 12, 33]. These findings suggest that the environmental pressure that promotes the development of T1D has been maintained, if not increased. However, the variability in T1D incidence continues and not every country or region shows the same pattern. Australia, another high incidence country, has not shown an overall increasing trend thus far [3, 34].

In our population, the increase in incidence has been found in both males and females, although slightly more marked in males (22%) than females (20%). This is unusual for autoimmune conditions which tend to be more prevalent in females. A number of high incidence countries, including Ireland, have long reported a slight excess of males in those diagnosed with T1D in childhood, although not reaching statistical significance [14].

The mean age of diagnosis fell over the 3 years. Since the establishment of the ICDNR and systematic monitoring of incidence rates of T1D in the Irish population, the highest incidence rates have consistently been in the 10–14-year age category. However, in 2021 for the first time this has changed, and the incidence rate has been highest in the 5–9-year age category. Others noting peaks in the 5–9-year age category were Finland, during 2006–2011, Algeria during 2015–2018 and the Calabrian region of Italy during 2019–2021 [9, 10, 12]. This raises a suggestion of a shift to an earlier age at diagnosis in Ireland, which has also been reported in other high incidence countries. Further monitoring is required to see if this effect will be maintained. An earlier age at presentation may also reflect increased environmental pressure and a more aggressive type of diabetes, applying the endotype concept with implications for future therapies, glycaemic control and potentially diabetes-related complications [35].

There have been interesting environmental changes from 2020 onwards with the COVID-19 pandemic that could possibly have an influence on the change in age at diagnosis, for example, in Ireland, children 12 years and older wore face coverings at school from September 2020 whilst younger children did not do so until the end of 2021. This potentially protected older children from viral infections precipitating decompensation of the struggling pancreas with progression to clinical symptoms and presentation whilst leaving younger children more exposed.