The multifactorial aetiology of diabetes mellitus in children and adolescents: A systematic review

Protocol and guidelines

This systematic review was designed and executed in strict adherence to the PRISMA 2020 statement. The objective was to identify, evaluate, and synthesise evidence on the factors contributing to the development of all major forms of DM in paediatric populations, defined as individuals with disease onset before 18 years of age. The review encompasses T1DM, T2DM, monogenic diabetes, and syndromic forms of diabetes.

Eligibility criteria

Studies were included if they met the following criteria: (1) Peer-reviewed publication type, including systematic reviews, meta-analyses, cohort studies, or authoritative narrative reviews; (2) Primary focus on risk factors, aetiology, or epidemiology of paediatric DM; (3) Publication date between 1 January 2010 and 31 December 2025, to capture contemporary evidence; (4) Published in the English language; and (5) Sourced from authentic biomedical databases and repositories (e.g., PubMed/NCBI).

Exclusion criteria were: (1) Studies focused exclusively on adult populations (age ≥18 years at diagnosis or study); (2) Case reports, editorials, commentaries, or non-research articles; (3) non-human or in vitro studies; and (4) Studies not accessible in full text for eligibility assessment.

Information sources and search strategy

A systematic electronic search was conducted in November 2025 using the PubMed database and the NCBI Bookshelf repository. The search strategy employed a combination of targeted keywords and site-specific operators to ensure depth and relevance. The primary search queries were:

• ‘Factors contributing to paediatric DM systematic review’

• ‘Risk factors paediatric T1DM systematic review’

• ‘Risk factors paediatric T2DM systematic review’

• ‘PRISMA flow diagram paediatric diabetes factors.’

The site: Pubmed.ncbi.nlm.nih.gov operator was utilised to restrict searches to the primary database. Search parameters were set to return 20–30 results per query to balance comprehensiveness with manageability. This process initially identified 90 unique records.

Detailed Search String Example (PubMed): (‘paediatric DM ’[MeSH Terms] OR ‘child’[MeSH Terms] OR ‘adolescent’[MeSH Terms]) AND (‘risk factors’[MeSH Terms] OR ‘aetiology’[MeSH Terms]) AND (‘type 1 diabetes’[MeSH Terms] OR ‘type 2 diabetes’[MeSH Terms]) AND (‘systematic review’[Publication Type] OR ‘meta-analysis’[Publication Type]) AND (’01 January 2010’[Date - Publication]: ’12 December 2025’[Date - Publication]).

Study selection process

The study selection followed a two-phase screening process managed by the primary reviewer. First, all 90 identified records were imported, and 20 duplicate entries were removed using automated and manual checks. The remaining 70 unique records underwent title and abstract screening against the eligibility criteria. At this stage, 20 records were excluded for clear irrelevance (e.g., adult-focused or with the wrong outcome).

Subsequently, 50 full-text reports were retrieved and assessed in detail for eligibility. A further 15 reports were excluded with reasons: 10 were not primarily focused on paediatric populations on detailed examination, and 5 were deemed low quality (e.g., lacking rigorous methodology, as assessed by a measurement tool to assess systematic reviews 2 (AMSTAR-2). Consequently, 35 studies met all inclusion criteria and were selected for data extraction and synthesis. This process is depicted in the PRISMA flow diagram below [Figure 1].

Close

Figure 1:

PRISMA flow diagram.

Export to PPT

Data extraction and quality assessment

Data from the 35 included studies were extracted into a standardised template. Key extracted information included: Study identifier/design, population characteristics (age, diabetes subtype), primary risk factors investigated, key findings (including odds ratios [ORs], hazard ratios [HR], and mean differences [MDs], where reported), and reported clinical outcomes. For systematic reviews and meta-analyses, methodological quality was assessed using the AMSTAR-2 checklist. Only studies of moderate to high quality were included in the final synthesis to ensure robustness of conclusions.

Data synthesis and presentation

Given the anticipated heterogeneity in study designs, populations, and reported outcomes across the included literature, a quantitative meta-analysis was deemed neither feasible nor appropriate. Therefore, a narrative synthesis approach was employed. Extracted data were thematically grouped into major domains of contributing factors: (1) Genetic and familial, (2) environmental and perinatal, (3) lifestyle and behavioural, and (4) socioeconomic and ethnic. Findings are presented narratively within the results section. They are further summarised in two tables: Table 1 provides a synthesised overview of key risk factors by diabetes subtype, and Table 2 presents the characteristics and key findings of a representative sample (10 of 35) of the included studies.

Overview of included studies and epidemiological context

The final synthesis incorporated 35 studies, comprising systematic reviews, meta-analyses, cohort studies, and authoritative narrative reviews. The collective evidence highlights a persistent upwards trend in the incidence of both major forms of paediatric diabetes. Data indicate a 21% increase in T1DM incidence between 2001 and 2009, with a continued annual rise of 2–5%.^[15,16] Concurrently, the incidence of T2DM in youth has surged, predominantly attributed to the global increase in childhood obesity.^[10] Adverse outcomes remain prevalent across settings; for instance, a meta-analysis focusing on East Africa reported suboptimal glycaemic control in 39–99% of youth with diabetes and an overall mortality rate of 6%.^[13] The following sections detail the contributing factors identified across the included literature.

Genetic and familial factors

Genetic predisposition forms the bedrock of risk for paediatric diabetes, though its role varies significantly by subtype.

In T1DM, the heritable component is substantial, with the HLA region on chromosome 6 accounting for approximately 50% of the genetic risk. Specific haplotypes, notably HLADR3 and HLA-DR4, are strongly associated with disease development, with individuals carrying both haplotypes exhibiting an OR >6.^[2,3,15,17] Beyond the HLA region, polymorphisms in genes such as insulin and PTPN22 are consistently implicated, with more modest effect sizes (OR ~2).^[15] A positive family history of T1DM not only increases disease risk but also appears to confer protection against the most severe presentation; knowledge of symptoms in families with affected members reduces the risk of DKA at diagnosis (OR 0.46).^[18]

In contrast, T2DM has a strong polygenic and familial background, in which risk is heavily influenced by shared genetics and family environments. It is estimated that 40– 60% of the risk for youth-onset T2DM can be attributed to parental transmission, particularly when a parent has T2DM.^[8,9,19] This genetic susceptibility interacts potently with modifiable lifestyle factors.

Monogenic diabetes, such as MODY, is caused by highly penetrant autosomal dominant mutations in single genes, such as Glucokinase (GCK) and Hepatocyte nuclear factor 1-alpha (HNF1A).^[19] These forms, though accounting for only 1–6% of paediatric diabetes cases, are clinically significant as they often require different management (e.g., sulfonylureas for HNF1A-MODY versus diet for GCK-MODY) than T1DM or typical T2DM.^[11] Finally, syndromic diabetes arises as part of complex genetic disorders involving multi-organ defects, such as mutations in WFS1 in Wolfram syndrome or mitochondrial DNA mutations.^[11]

Environmental and perinatal factors

Environmental exposures, particularly in early life, play a critical modifying role in the pathogenesis of T1DM and may influence T2DM risk.

A robust body of evidence implicates viral infections as potential triggers for islet autoimmunity. Enteroviruses (e.g., Coxsackie B) and other respiratory infections have been associated with an increased risk of T1DM, with reported ORs ranging from 1.5 to 3.^[6,7,14] Paradoxically, broader early-life exposure to common pathogens in group settings may be protective; a meta-analysis found that day care attendance was associated with a reduced risk of developing T1DM.^[4,20]

Perinatal and early nutritional factors are also influential. Birth by caesarean section has been linked to a modestly increased T1DM risk (OR ~1.2), possibly through alterations in the infant gut microbiome.^[15,21] Similarly, preterm birth (OR 1.18) and maternal obesity during pregnancy are associated with higher diabetes risk in offspring.^[15] Early infant diet is a key area of investigation; systematic reviews suggest that early exposure to cow’s milk proteins and gluten may be associated with an increased risk of islet autoimmunity, while longer breastfeeding duration appears protective.^[5,21]

For T2DM, the broader environment is significant. Urbanisation, with its associated changes in physical activity and food environment, is a major driver.^[12,22] Emerging evidence also points to ambient air pollution as a potential risk factor for insulin resistance and diabetes in children.^[23]

Lifestyle and behavioural factors

Lifestyle factors are the principal modifiable determinants for T2DM and significantly impact management outcomes in T1DM.

Obesity is the single most powerful risk factor for youth-onset T2DM. The condition is often preceded by prediabetes, which is highly prevalent in adolescents with obesity.^[24,25] Furthermore, obesity complicates T1DM management; children and adolescents (particularly ages 10–14) with T1DM and a BMI ≥95^th percentile are more likely to experience suboptimal glycaemic control.^[26]

Sedentary behaviour and poor dietary patterns directly contribute to T2DM risk. High intake of sugar-sweetened beverages and snacks, coupled with low levels of physical activity, are consistently identified as risk factors.^[20,22] Importantly, these factors are modifiable through intervention. Family-based behavioural interventions targeting diet and physical activity have demonstrated efficacy in reducing BMI (MD −0.18–−0.21) and improving activity-related self-efficacy (standardised MD 0.73) in children at risk for T2DM.^[27] Stress is another behavioural factor, with evidence linking significant life stress to poorer metabolic control in T1DM (HR 1.67–3).^[15] As with other subtypes, breastfeeding is associated with a protective effect against the development of T2DM.^[22]

Socioeconomic and ethnic factors

Significant disparities in diabetes risk and outcomes exist across ethnic and socioeconomic groups.

Ethnic minority status is a strong risk marker. In many regions, children of non-European ancestry (e.g., Asian, Black, Hispanic) are at disproportionately higher risk for developing T2DM.^[10,19,28] Furthermore, ethnic minority children with T1DM face a higher risk of presenting with DKA at diagnosis, indicating disparities in early symptom recognition or healthcare access.^[15] Conversely, being part of the ethnic majority in a given region can be protective against DKA (OR 0.40).^[15]

SES is a fundamental driver of health inequities. Lower SES, characterised by lower household income and parental education levels, is strongly associated with both the incidence of T2DM and poorer glycaemic control and outcomes in all diabetes types.^[3,12,13] The association of higher maternal age and education with diabetes risk can be nonlinear, reflecting complex interactions with other environmental and behavioural factors.^[15] In low-resource settings such as East Africa, these socioeconomic challenges contribute to alarmingly high rates of adverse outcomes.^[13]

Complications and associated factors

Early and aggressive management is crucial to mitigate complications, which can begin in childhood. The risk of diabetic peripheral neuropathy (DPN) and cardiovascular autonomic neuropathy (CAN) in T1DM is driven by traditional factors such as prolonged hyperglycaemia, longer diabetes duration, and higher BMI, with reported prevalence rates of 13– 88% for DPN and 4–39% for CAN.^[1] Acute complications are also prevalent; younger age at diagnosis (particularly <2 years) is a powerful risk factor for DKA at onset (OR 3.51).^[21]

Strengths and limitations

The primary strength of this review lies in its adherence to PRISMA guidelines, its focus on authentic peer-reviewed sources, and its synthesis of factors across the spectrum of paediatric diabetes. The inclusion of 35 references from the past 15 years provides a contemporary and comprehensive overview.

Several limitations must be acknowledged. The significant heterogeneity in study designs, populations, and measured outcomes precluded a formal meta-analysis, limiting the ability to provide pooled quantitative estimates. The search was restricted to English-language articles and primarily to PubMed/NCBI, which may introduce language and database bias, potentially omitting relevant studies from other sources or in other languages. Furthermore, while quality assessment was performed, the reliance on secondary data (reviews and meta-analyses) means the synthesis is dependent on the accuracy and quality of the included primary studies.

Implications and future directions

The findings have clear implications for practice and policy. For clinicians: A high index of suspicion for monogenic diabetes is warranted in atypical cases. Screening for T2DM should be intensified in high-risk groups (e.g., adolescents with obesity, especially from high-risk ethnicities). For public health: Primary prevention efforts must dual-track: Investigating environmental modulators of autoimmunity for T1DM and decisively tackling childhood obesity through policies promoting healthy diets, physical activity, and breastfeeding for T2DM. For researchers: Future studies should prioritise longitudinal cohorts to understand the temporal sequence of exposures better, incorporate ‘omics’ technologies (e.g., genomics, metabolomics, microbiome analysis) to refine risk prediction, and design and evaluate interventions tailored for low-resource settings where the burden of adverse outcomes is highest.^[15,26]

In conclusion, the epidemic of paediatric diabetes is not inevitable. It is the result of a predictable collision between genetic risk and modifiable environmental, lifestyle, and socioeconomic factors. A nuanced understanding of this multifactorial aetiology, as presented in this review, is the essential first step towards developing the precise, equitable, and multifaceted strategies needed to reverse these concerning trends.