Demographic, Clinical, Management, and Outcome Characteristics of 8,004 Young Children With Type 1 Diabetes Free

Management of type 1 diabetes in preschoolers has many challenges; young children are dependent on others for their care, have unpredictable diet and activity levels and a reduced ability to communicate needs and hypoglycemic symptoms, and often require very low total daily insulin doses (1). Early glycemic optimization is important in this population to reduce the risk of complications (1,2). Therefore, furthering our understanding of this vulnerable age group is essential in order to identify strategies to improve early type 1 diabetes management and optimize long-term outcomes.

The Diabetes Prospective Follow-Up Registry (DPV), the T1D Exchange (T1DX-QI) Quality Improvement Network, and the Australasian Diabetes Data Network (ADDN) are three large consortia of diabetes centers from different continents that were established with the objective of improving the care of children with diabetes. DPV and T1D Exchange previously reported data in children aged <6 years with type 1 diabetes from 2010 to 2012 (3), demonstrating that not all children met recommended contemporary glycemic targets (noting the different targets of the International Society of Pediatric and Adolescent Diabetes [ISPAD] and American Diabetes Association [ADA] at that time). In that report, the current ISPAD HbA_1c target (4) of <7.5% (58 mmol/mol) was met by only 56% in DPV and 22% in T1DX-QI. Mean HbA_1c was 8.2% in T1DX-QI compared with 7.4% in DPV. Continuous subcutaneous insulin infusion (insulin pump) use was higher in DPV (74% vs. 50% in T1DX-QI). Fewer than 10% of young children used continuous glucose monitoring (CGM) technology, and hybrid closed-loop (HCL) therapy was not commercially available. ADDN previously published data from 2015 on children of all ages with type 1 diabetes; that report showed that 34% of children age ≤6 years had HbA_1c <7.5% (58 mmol/mol) and that 43% used insulin pumps. CGM use was not reported in the study (5).

Over the last decade, the landscape of type 1 diabetes management has changed considerably, with more ambitious glycemic targets (HbA_1c <7.0% [53 mmol/L]) and an increase in the approval, availability, and use of diabetes technologies, including insulin pumps and CGM systems (1,6,7) There is clear evidence that use of diabetes technologies benefits young children with type 1 diabetes; this includes two recent randomized controlled trials showing that HCL systems reduced HbA_1c and hypoglycemia (8,9). Our study describes demographic, clinical, and management characteristics and outcomes in registries across three continents.

This cross-sectional study involved an analysis of three prospective registries that included 8,004 participants. Participants were included if they had a diagnosis of type 1 diabetes and were aged <6 years at the time of diagnosis. Data were included if collected between 1 January 2019 and 31 December 2021 (inclusive). Characteristics and methods of the three participating registries are summarized below, and more information detailing each registry has been provided in previous publications (3,10–12).

The DPV registry is a prospective longitudinal standardized computer-based documentation system for individuals with all types of diabetes and of all age groups including more than 500 pediatric and adult care centers in Austria, Germany, Luxembourg, and Switzerland. Data are documented locally by the participating centers in an electronic health record. Twice yearly, anonymized data are exported and transmitted for central analysis and external quality assurance. Missing and inconsistent data are reported back to the centers for verification (3,13).

The T1DX-QI clinic network includes >50 pediatric and adult endocrinology practices across the U.S. and collects electronic medical record data on all individuals with type 1 diabetes receiving care in participating institutions. Data are shared monthly with the central coordinating center, and T1DX-QI reviews these submitted files monthly for validation and quality assurance. T1DX-QI shares a data quality assurance report quarterly with each center. The report includes the percentage of missing variables, incorrect values, and benchmarked scorecards with other centers in the network. Data capture varies per center, but overall estimated data capture for key variables is 90–97% (14,15).

ADDN is a prospective longitudinal registry that includes clinical data from 25 collaborating pediatric and adult diabetes centers across Australia and New Zealand. Participating centers upload data to ADDN every 6 months; comprehensive data validation rules and error reports have been implemented to ensure data quality (16).

In our study, descriptive statistics were calculated for demographic, clinical, and management data based on the most recent available data from the most recent clinic visit in which a child was aged <6 years, except for those calculated based on the most recent year of data. Individual median HbA_1c, height, weight, BMI, age at assessment, and diabetes duration from the most recent 12 months were used for analyses, with mean and SD values calculated for each registry. z scores for BMI, height, and weight were calculated based on World Health Organization reference ranges (17). Means and SDs are provided for continuous variables, and percentages were calculated for categorical variables. Overweight and obesity were defined according to the World Health Organization as BMI SD scores of >1 and >2, respectively (17). Severe hypoglycemia in this report was defined per 2018 ISPAD guidelines as an event with severe cognitive impairment requiring external assistance to treat (18). However, registries vary in how they record hypoglycemia; DPV and T1DX-QI use similar terminology to document hypoglycemia requiring external assistance as severe hypoglycemia, whereas ADDN documents hypoglycemia with coma as severe and hypoglycemia requiring assistance as moderate. Diabetic ketoacidosis (DKA) was defined as pH <7.3 and/or serum bicarbonate <15 mmol/L. Incidence of DKA and hypoglycemic episodes were calculated per person-years, with episodes of DKA at diagnosis excluded. The HbA_1c target of <7.0% (53 mmol/mol) was used based on current ISPAD and ADA guidelines (1,19).

Differences among registries or treatment groups (multiple daily injections [MDI] vs. insulin pump and self-monitoring of blood glucose [SMBG] vs. CGM) were analyzed using ANOVAs or unpaired t tests for continuous variables and χ² tests for categorical variables. Regression analysis was used to adjust HbA_1c for age when comparing mean HbA_1c among technology/treatment groups. Ethical approval was obtained at the respective centers. DPV was approved by the ethics committee of Ulm University (Ulm, Germany; 314/21) and by local review boards of the participating centers. T1DX-QI registry data were approved by the Western Institutional Review Board (Puyallop, WA), and each center secured local approval. ADDN was approved by the ethics committee of Monash University (Melbourne, Victoria, Australia), and each center secured local approval. Because of registry data sharing restrictions, statistical analyses were conducted at each registry with no raw data sharing and no data pooling. Statistical analyses were performed using IBM SPSS Statistics for Windows (version 27; IBM Corp., Armonk, NY) and R (version 3.4.0; R Core Team, Vienna, Austria). A priori, in view of the large sample size and multiple comparisons, only P values <0.01 were considered statistically significant.

The data that support the findings of this study are available from respective registries (DPV, T1DX-QI, and ADDN), but restrictions apply to the availability of these data, which were used under license for the current study and therefore are not publicly available. However, data are available from the authors upon reasonable request and with permission of the respective registries.

Data were available for 8,004 children: 5,219 from DPV, 1,831 from T1DX-QI, and 954 from ADDN. Results are summarized in Table 1. Mean age at diabetes onset was lower in DPV and ADDN compared with that in T1DX-QI. Age at assessment was slightly higher in T1DX-QI compared with ages in the other registries. Young children from ADDN had higher BMI compared with those in DPV and T1DX-QI (P < 0.001), and those from T1DX-QI were shorter for their age.

Insulin pumps were more commonly used in DPV (86.6%) than in T1DX-QI (46.6%) and ADDN (39.2%). CGM use was also more common in DPV (85.1%) versus T1DX-QI (57.6%) and ADDN (75.6%). HCL system use was low (DPV 6.9%, T1DX-QI 3.6%, and ADDN 0.5%).

HbA_1c was highest in T1DX-QI (8.0 ± 1.4% [64 ± 16 mmol/mol]) versus ADDN (7.7 ± 1.2% [61 ± 13 mmol/mol]) and DPV (7.3 ± 0.9% [56 ± 10 mmol/mol]). Target HbA_1c <7% (53 mmol/mol) was achieved by more children in DPV (43.6%) versus ADDN (27.5%) and T1DX (25.5%; P < 0.001). The previous HbA_1c target of <7.5% (58 mmol/mol) was achieved in 65.3% in DPV, 30.5% in T1DX-QI, and 44.6% in ADDN (P < 0.001). HbA_1c <6.5% was achieved in 21.9% in DPV, 9.2% in T1DX-QI, and 11.6% in ADDN.

Incidence of severe hypoglycemia and hypoglycemia with coma was higher in ADDN compared with the other two registries. Incidence of DKA was low across the three registries (Table 1). Each registry compared incidence of DKA and severe hypoglycemia with use of technologies (Table 2); when comparing those using CGM versus SMBG, there was a significantly lower incidence of severe hypoglycemia in DPV and of hypoglycemia with coma in ADDN, but no other differences were observed.

Figure 1 shows differences in mean age-adjusted HbA_1c for different treatment groups (insulin pump vs. MDI and CGM vs. SMBG). There were no significant differences between adjusted and unadjusted values. In DPV, mean HbA_1c was similar across all treatment groups. In T1DX-QI, mean HbA_1c was lower in those using an insulin pump (7.7% [61 mmol/mol]) versus MDI therapy (8.3% [67 mmol/mol]; P < 0.001) and in those using CGM (7.8% [62 mmol/mol]) compared with SMBG (8.4% [68 mmol/mol]; P < 0.001). In ADDN, HbA_1c was lower in those using an insulin pump (7.4% [57 mmol/mol]) versus MDI (7.8% [62 mmol/mol]; P < 0.001), whereas mean HbA_1c level was similar in those using CGM and in those using SMBG (7.7% [61 mmol/mol] vs. 7.8% [62 mmol/mol]; P > 0.05).

In this analysis of 8,004 children aged <6 years with type 1 diabetes across three international registries, we report differences in HbA_1c, use of diabetes technologies, and anthropometric parameters.

ISPAD and ADA currently recommend the HbA_1c target of <7.0% (53 mmol/mol) (1,19); more than half (62.5%) of this cohort did not achieve this target. ISPAD guidelines go on to suggest that preschoolers with access to modern diabetes care can safely achieve HbA_1c <6.5% (53 mmol/mol), and the National Institute for Health and Care Excellence also states an ideal target of <6.5% (53 mmol/mol) reduces risk of long-term complications (1,19,20). Only a small minority of participants (∼20% from DPV and ∼10% from ADDN and T1DX-QI) achieved HbA_1c <6.5%. Additional research and resources are needed to identify areas of improvement to help bring clinical care into line with guideline expectations.

Compared with published data from DPV and T1DX-QI in 2011–2012 in this age group (3), mean HbA_1c was slightly lower; mean HbA_1c was 0.1% lower in DPV (7.4 vs. 7.3%; P < 0.001) and 0.2% lower in T1DX-QI (8.2 vs. 8.0%; P < 0.001). In addition, the proportion achieving HbA_1c <7.5% (58 mmol/L) was increased by almost 10% in both DPV and T1DX-QI compared with the previous study. Similarly, improvements were seen in the ADDN cohort when compared with 2015 data (5); the proportion of children aged <6 years with type 1 diabetes achieving HbA_1c <7.5% increased from 35 to 44.6%. The proportion achieving HbA_1c <7% (53 mmol/mol) in all three registries was lower than in 2018 Swedish registry data of children with type 1 diabetes aged <7 years, where 61% of children achieved this target, and a threefold improvement was seen when compared with their 2008 data (21). Therefore, although progress seems to have been made when compared with previous data, more rapid and substantial progress is possible and needed in the coming decades to improve outcomes for young children with type 1 diabetes.

Differences among registries in HbA_1c, with a higher proportion achieving targets in DPV followed by ADDN and then T1DX-QI, may in part be due to previous discrepancies in glycemic targets. Variability among countries in mean HbA_1c in pediatric type 1 diabetes has been observed in other studies. One international comparison observed the lowest HbA_1c in Sweden’s national registry, as compared with other European registries (including DPV), and data from the U.S. and U.K. recorded a higher mean HbA_1c in all age groups compared with European data sets (22). This study noted that Sweden had the lowest HbA_1c targets at the time. Although current ISPAD and ADA guidelines both suggest an HbA_1c target of <7% (53 mmol/mol), ISPAD has recommended this HbA_1c target since 2018, whereas the ADA has only recently recommended this target (from <7.5% [58 mmol/mol] in 2020) (1,19,23). There is evidence that lower glycemic targets are associated with achievement of lower HbA_1c (24,25); thus, the lower recommended targets for HbA_1c may have contributed to the modest reduction in mean HbA_1c in these present data compared with the previous study (3). Similarly, although there are likely many reasons contributing to the lower HbA_1c in DPV compared with in the other registries, these data likely also reflect the previous disparity between guidelines and time taken for this change to affect clinical practice (6,19,26,27).

Use of diabetes technologies in this age group has increased in the last decade (3,21) but is still far from universal. Insulin pumps and CGM are beneficial for the management of type 1 diabetes in young children (7,9,28,29), and the use of diabetes technologies is recommended in the preschool age group (1,6,7,20). However, access to these technologies varies significantly worldwide. Insulin pump use was more common in DPV compared with in the other registries and previous data (87 from 74%), whereas insulin pump use in T1DX-QI and ADDN was unchanged compared with in previous registry studies (T1DX-QI 50 to 47% and ADDN 43 to 39%). Also consistent with other research (10,11), our data showed an increase in CGM use over time; a majority of children in all three registries used CGM, as compared with <10% between 2010 and 2012 in T1DX-QI/DPV (3). CGM use was higher in DPV (80%) and ADDN (76%) compared with in T1DX-QI (58%). Differences in diabetes technology uptake among registries may be due to variations in clinical practice and ability to access these devices. Many European countries (including those involved in DPV) have government-funded insulin pumps and CGM systems for all children with type 1 diabetes; however, children in the U.S. have variable access to technologies, dependent on insurance or geographical location. Previous research has shown an association between low socioeconomic status and lower use of diabetes technologies and higher HbA_1c in T1DX-QI, with this association being weaker in DPV (30). In Australia, insulin pumps are dependent on private health insurance or compassionate access via limited government programs and charities, whereas CGM has been government funded for all children with type 1 diabetes since 2017; research shows there was a high uptake of CGM after the subsidy was introduced and reduced HbA_1c associated with use (31). In New Zealand, insulin pumps are publicly funded, but CGM is not. These variations in practice and access likely contribute to the discrepancies seen among registries, and local advocacy is essential to improve access to diabetes technologies for all young children with diabetes.

Recent randomized controlled trials in children with type 1 diabetes aged <6 years have shown that automated insulin delivery systems lower HbA_1c and hypoglycemia incidence, increase time in range, and improve quality of life (8,9). Despite this evidence supporting the use of HCL systems, our report shows that use of these systems was uncommon across all three registries. Notably, these systems were not approved for use in very young children during the study period. Use of HCL systems in this age group is likely to increase with time and may lead to more young children achieving glycemic targets. Given the significant benefits in this vulnerable age group and an estimated time from research to clinical translation estimated at 17 years (32), local and international advocacy focusing on HCL systems are important to improve timeliness of access and more widespread uptake of these systems in young children with type 1 diabetes.

Consistent with an increase in global numbers of obese/overweight children with and without type 1 diabetes (33–35), there was a high rate of children with BMI above the recommended healthy range; BMI was >2 SD scores in 9% of those from DPV, 12.8% of those from T1DX-QI, and 15% of those from ADDN. This discrepancy may be in part due to a lower background rate of childhood obesity in Europe compared with in the U.S. and Australasia, although data for children aged <6 years are lacking (33–35). Of concern, the rate of obese and overweight children aged <6 years has increased compared with 2015 ADDN data; the proportion of children who are obese increased from 8 to 15% and that of those who are overweight from 31 to 50% (5). However, mean BMI z score in this data set for DPV and T1DX-QI was similar to data from 2011–2012 (3). Higher BMI, along with higher HbA_1c, increases cardiovascular disease risk (36,37); these data therefore highlight the importance of integrating healthy lifestyle education into clinical care for young children with type 1 diabetes, and the increase over time in obesity seen in children registered with ADDN warrants further investigation.

Hyperglycemia and hypoglycemia have negative impacts on the developing brain of young children (38,39). Therefore, the gradual improvement in HbA_1c, along with the low rate of severe hypoglycemia, is reassuring. ADDN reported a higher rate of severe hypoglycemia (defined as requiring assistance), a finding that warrants further review. However, it is difficult to compare registries directly, because definitions of hypoglycemia and methods of data collection differ. Although both DPV and T1DX-QI use similar terminology, where severe hypoglycemia was defined as requiring external assistance, ADDN labels hypoglycemia with coma as severe and hypoglycemia requiring assistance as moderate. This difference in language used may have biased reporting clinicians to over-report such episodes. Given that young children almost always require assistance to manage a hypoglycemic episode, the use of age-appropriate definitions for hypoglycemia in future registries may improve the accuracy of data in the future.

We report on a very large sample size of children with type 1 diabetes aged <6 years; however, all countries included in these three registries are high-income countries. This is therefore not a global representation of all young children with type 1 diabetes, and future studies should consider including data from middle- and low-income countries. Another limitation is that we did not have adequate data to report time in range, an important measure of glycemic control.

The observational study design is another consideration. In addition, data-sharing restrictions meant that we were unable to data pool, and therefore, we could not perform multivariable analysis, and we did not adjust HbA_1c for factors such as duration of diabetes. Analyses were performed separately by representatives from each registry and compared. This is particularly relevant when examining HbA_1c variations and use of diabetes technologies, where a complex interplay of many contributing factors is likely. Information on underlying disparities in socioeconomic status and differences in ethnicity and language among registries was not available; however, these are potential contributing factors. Consistent with the wider literature, studies in both DPV and T1DX-QI have shown associations between socioeconomic factors, such as parental education level and household income, and differences in HbA_1c (30,40). Additionally, although DPV is estimated to cover >80% of children with type 1 diabetes in the countries it represents and ADDN 60%, T1DX-QI is a clinic-based registry and covers ∼2.5% of the population of the U.S.; this introduces potential bias and limits extrapolation of these findings (14). This is particularly relevant when reviewing use of diabetes technologies, which is variable across centers and often clinic or region specific in the U.S. It may also have influenced other differences seen in the registries’ data, such as small differences in age at diagnosis or anthropometry. Future research is required to determine how differences in mean HbA_1c and CGM metrics among registries and treatment groups are affected by clinical practice, government funding of diabetes technologies, health care systems, and cultural and socioeconomic factors.

These data are useful in identifying areas of further research and generating research questions, as well as in recognizing areas where management of type 1 diabetes in young children could be improved. Research into factors contributing to the discrepancies among registries in glycemia and use of diabetes technologies may help to identify ways to optimize care in this age group. Integrating healthy lifestyle education is also essential in reducing the rate of obesity and subsequent complications or comorbidities. These data highlight a need to identify potential barriers to diabetes technology uptake and initiate strategies to address these at a local level and on a wider community, national, or international scale. This youngest age group of children has been the last to benefit from technological advances, such as HCL systems, with uptake slower than in adults and older children. However, the rate of uptake is expected to continue to increase with time, and future studies may focus on confirming whether an increase in use of HCL systems leads to more young children achieving glycemic targets and improved outcomes.

In conclusion, in an analysis of 8,004 children with type 1 diabetes aged <6 years, HbA_1c was lower when compared with data from a decade earlier (3), but only 37.5% (n = 3,004) achieved the glycemic target of HbA_1c <7.0% (53 mmol/mol), with glycemic targets met by more children participating in DPV compared with in ADDN and T1DX-QI. Although most of the children used CGM, use of insulin pumps was variable across the three international registries. Despite the high-level evidence support provided in ISPAD and ADA guidelines (1,6–8,19,29), HCL systems were used uncommonly in the study population. Future research on differences in HbA_1c and use of diabetes technologies among registries may help define important areas of improvement to optimize care for young children with type 1 diabetes and their families.

Acknowledgments. The authors acknowledge the input and expertise of Professor Reinhard W. Holl (DPV) and Nudrat Noor (T1DX-QI). DPV, T1DX-QI, and ADDN acknowledge the participating centers, clinicians, patients, and families, as well as Albert Chan for statistical support.

Funding. T1DX-QI is funded by the Leona M. and Harry B. Helmsley Charitable Trust. Financial support for DPV was provided by the German Center for Diabetes Research (grant 82DZD14E03) and the Robert Koch Institute (grant 1368-1711). ADDN was previously supported by the Australian Type 1 Diabetes Clinical Research Network, led by JDRF Australia, the recipient of Australian Government funding from the Australian Research Council (through a special research initiative) and the Department of Health and Ageing. D.M.M. has had research support from the National Institutes of Health, JDRF, the National Service Framework for Diabetes, and the Helmsley Charitable Trust.

Duality of Interest. D.M.M. reports research support to institution from Medtronic, Dexcom, Insulet, Bigfoot Biomedical, Tandem, and Roche and consulting for Abbott, Aditxt, the Helmsley Charitable Trust, LifeScan, MannKind, Sanofi, Novo Nordisk, Eli Lilly, Medtronic, Insulet, Dompe, Biospex, Provention Bio, and Bayer. O.E. is an adviser for Medtronic, Vertex, and Sanofi and reports research support to T1D Exchange from Medtronic, MannKind, Vertex, Dexcom, and Eli Lilly. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.L.S., S.R.T., and S.R. reviewed and analyzed the data. J.L.S. and M.E.C. wrote the first draft of the manuscript. D.M.M. and M.E.C. were involved in the conception and design of the study and supervised and interpreted data analysis. All authors reviewed, revised, and edited the manuscript and approved the final version of the manuscript. J.L.S. and S.R.T. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. An abstract of the research was presented as a poster at the American Diabetes Association 83rd Scientific Sessions, San Diego, CA, 23–26 June 2023.