1 INTRODUCTION

Autoimmune diseases are a group of complex immunological disorders, in which the immune system erroneously interprets normal tissues as foreign, generating unwanted attacks, and inflammations in different organs. This study focuses on four autoimmune diseases with partially to represent autoimmune diseases in individuals between the age of 11 and 19 years in general: type 1 diabetes (DM), autoimmune thyroiditis (AIT), juvenile idiopathic arthritis (JIA), and inflammatory bowel diseases (IBD). These diseases were chosen due to their potentially overlapping genetic pathways. In addition, their pathogeneses resemble each other, showing T-cell organ infiltrations with intermittent inflammations, yet without specific triggering factors. Pancreatic T-cell infiltration leads to depleted insulin secretion, hyperglycaemia, and eventually DM; thyroid gland infiltration leads to reduced thyroid hormone function and hypothyroidism seen in AIT; synovial membrane infiltration leads to joint inflammation and JIA; and gut mucosa infiltration leads to chronic autoimmune inflammations of the gastrointestinal tract seen in IBD.1-6

In recent decades, the numbers of both autoimmune diseases and obesity have been increasing.7, 8 This has led researchers to wonder about their potential relationship, especially when obesity is seen as a continuous state of chronic low-grade inflammation and has been associated with high disease activity in several autoimmune diseases.9, 10 While obesity may be related to the onset of hypothyroidism,11 its role in the pathogenesis of DM, JIA, and IBD is still unclear.

‘Western type’ diet (with high fat, protein, sugar, and salt intake), when compared with more fibre-rich diet, might be associated with inflammation-related mechanisms in autoimmune diseases.12, 13 This finding is supported by previous studies, indicating an inflammatory role of a particular diet, and relating consumption of pro-inflammatory foods with disease activities of autoimmune diseases.14-16 Furthermore, long-term dietary patterns have been shown to modify gut microbiota to a certain degree and, as a result, may contribute to the inflammation processes.17, 18 So far, most studies associating diet and autoimmune diseases have involved adults in a retrospective setting, or concerned the impact of diet on disease activities. Prospective studies to determine the relationship between dietary patterns, body composition, and the development of autoimmune diseases in children and adolescents have been deficient. This study aims to investigate whether dietary patterns (meal patterns, eating habits, and consumption frequencies of sugary products and fruits/vegetables), waist-to-height ratio (WTHR), and body mass index (BMI) in school-aged children are risk factors for developing autoimmune diseases (DM, AIT, JIA, and IBD) in adolescence.

2 METHODS 2.1 Data sources and study population

The study population was derived from the Finnish Health in Teens (Fin-HIT) cohort.19 The Fin-HIT cohort was assembled by school recruitment in vast, densely populated areas throughout Finland without specific exclusion criteria, comprising 11 407 school-aged children. At baseline, when the children's median age was 11.3 years, questionnaire was used to obtain data concerning food consumption (food frequency questionnaire, FFQ) and meal patterns. In addition, weight, height, and waist circumference were objectively measured. After approximately 2.6 years from baseline, parents were asked to report their children's body measures, which were obtained at home with a provided measuring tape according to given instructions. These measures were available for 50% of the children in the cohort.19

DM, AIT, JIA, and IBD (including Crohn's disease, ulcerative colitis, and IBD unclassified) diagnoses (from birth to 31 December 2018) were obtained by linking the participants' personal identifier number (presenting the date of birth and sex) with two national health registers—the Special Reimbursement Register and the Drug Purchase Register—maintained by the Finnish Social Insurance Institution.20 The excellent coverage of these registers is described elsewhere.21 Celiac disease has a known triggering factor (gluten) and requires no long-term medical prescriptions, hence this diagnosis was not available in the registers we used.

In the Fin-HIT cohort, 245 children received their primary diagnoses (DM, AIT, JIA, or IBD) by the end of follow-up. Of them, girls were more commonly affected by AIT and JIA than boys, while AIT was more common in adolescence than in preschool age.22 To limit the role of age, sex, and residential areas as profound confounders, a matched case–control design was chosen for this study. The case group consisted of 105 children with available FFQ, whose diagnoses were dated at least 1 month after recruitment. Of them, 34 (32.3%) developed DM, 39 (37.1%) AIT, 18 (17.1%) JIA, and 14 (13.3%) IBD during the follow-up period. For each child in the case group, four children with matching age, sex, and residential area were selected from the same cohort by a person outside the study group, generating the control group of 420 children (Figure S1).

2.2 Anthropometric measures: Waist-to-height ratio (WHTR) and body mass index (BMI)

Based on the WHTR, the children were categorized as having a normal waist (WHTR <0.5) or being centrally obese (WHTR ≥0.5).23 BMI (=weight [kg]/height [m]2) was categorized based on age and sex according to the International Obesity Task Force cutoffs as ‘underweight’ (comprised thinness grades II and III only, with standard deviation [SD] score equivalent <−1.88 for boys and <−1.79 for girls), ‘normal’ (SD score equivalent −1.88 to 1.31 for boys and −1.79 to 1.24 for girls), and ‘overweight’ (including both overweight and obese with SD score equivalent >1.31 for boys and >1.24 for girls).24

2.3 Meal patterns, eating habits, sweet treat index (STI), plant consumption index (PCI)

Food consumption in the preceding month prior to baseline was analysed based on obtained FFQ data, which included 16 food items to make it simple enough for 9- to 12-year-old children to apprehend, yet sufficiently detailed to cover mandatory key indicators used to determine healthy and unhealthy dietary habits in children according to the Health Behaviour in School-Aged Children study protocol.25 These food items were: (1) fruits/berries, (2) fresh vegetables, (3) cooked vegetables, (4) sugary soft drinks. The additional food items for healthy eating behaviours were (5) dark grain bread, (6) milk or buttermilk, (7) fresh juice (no added sugar), and (8) water. For unhealthy eating behaviours, the additional food items were (9) pizza, (10) hamburger or hot dog, (11) biscuits/cookies, (12) ice cream, (13) chocolate/sweets, (14) sweet pastries, (15) salty snacks, and (16) sugary juice drinks.26 The participants estimated their weekly consumption of these food items using a 7-point scale ranging from 0 (not consumed) to 6 (consumed several times per day).

Eating habits and meal patterns at baseline have been assessed previously using FFQ and a questionnaire on the regularity of breakfast, lunch, and dinner during school days.27 Children with unhealthy eating habits frequently consumed unhealthy food items such as pizza, hamburger/hot dog, sugary products, and salty snacks. The fruit and vegetable avoiders rarely consumed unhealthy food items, but they scarcely consumed fruits/berries, fresh juice, and fresh or cooked vegetables as well. Healthy eaters regularly consumed dark grain bread, milk, fruits/berries, fresh juice, and fresh grated or cooked vegetables, while they consumed unhealthy food items less often. Furthermore, children with regular meal patterns were those who had lunch and dinner every weekday, while the rest had irregular meal patterns. Children who ate breakfast every school day were classified as having a regular breakfast pattern, while the rest had an irregular breakfast pattern.

To estimate weekly consumption frequencies of sugary products and plants at baseline, FFQ answers were recoded into a times/week continuous variable: ‘not at all’ as 0, ‘less than once a week’ as 0.5 (assuming an average consumption of twice per month), ‘once a week’ as 1, ‘2–4 times a week’ as 3, ‘5–6 times a week’ as 5.5, ‘once a day’ as 7, and ‘several times a day’ as 14 times a week (assuming an average consumption of twice daily). The weekly consumption frequency of sugary products was evaluated using a sum variable referred to as a sweet treat index (STI).28 It was calculated by adding up continuous weekly consumption frequencies of food and drink items representing sugary products: (1) chocolate/sweets, (2) biscuits/cookies, (3) ice cream, (4) sweet pastry, (5) sugary juice drinks, and (6) sugary soft drinks.

Plant consumption index (PCI)—a sum variable analogous to the STI, combining continuous weekly consumption frequencies of (1) fruits/berries, (2) fresh vegetables, and (3) cooked vegetables. It was used to estimate the weekly use of plants as a crude indicator for anti-inflammatory food components.14, 29

2.4 Statistical methods

Each child who developed a primary autoimmune disease was compared with four matching controls. The data were presented with mean and standard deviation (SD); median and interquartile range (IQR); and numbers/proportion (%). Pearson's chi-squared test, the independent samples t-test, and the Kruskal–Wallis test were used as appropriate when comparing background information of case and control groups—to inquire how well the control group were matched, and to observe anthropometric characteristics (BMI SD score and WHTR) of both groups. Risk factor variables (central obesity, BMI categories, eating habits, meal patterns, STI, and PCI) were presented with odds ratio (OR) and 95% confidence interval (CI) using conditional logistic regression tests. A 5% statistical significance level was adopted. The software used was IBM SPSS Statistics 26.0.

2.5 Ethical considerations

The Fin-HIT study protocol was approved by the Ethics Committee of the Hospital District of Helsinki and Uusimaa (decision number 169/13/03/00/10). The children and one of their guardians provided written informed consent.

3 RESULTS

Data on anthropometric measures and dietary patterns were collected approximately 3 years prior to established diagnoses. The matched case–control study population comprised 525 children (105 cases and 420 matching controls) with similar background characteristics (Table 1). The mean age of the study population was 11.3 years at baseline, and the children were followed-up for a median of 5.1 (IQR 4.7–5.6) years. Approximately 58.1% of the study population were girls and 41.9% were boys. Of them, 28.7% were from the densely populated capital region. Moreover, anthropometric measures of the matched controls as one group well represented the measures of 11 162 children in the cohort, who did not develop autoimmune diseases.

TABLE 1. Characteristics of children who developed primary autoimmune diseases, their matching controls, and children in the Fin-Hit cohort who did not develop autoimmune diseases Cases (N = 105)a Matched controls (N = 420)b p-value Children without diagnoses (N = 11 162) p-value Mean age, years ± SD At baseline 11.3 ± 0.8 11.3 ± 0.8 0.988c,f 11.2 ± 0.8 0.025c,g Missing (%) 0 0 At the end of follow-up 16.8 ± 1.3 16.5 ± 1.2 0.126c,f 16.5 ± 1.5 0.069c,g Missing (%) 0 0 419 (3.8) Sex, n (%) 0.979d,f 0.172d,g Boy 44 (41.9) 176 (41.9) 5074 (47.6) Girl 61 (58.1) 244 (58.1) 5576 (52.4) Missing 0 0 512 (4.6) Residential area, n (%) 1.000d,f 0.162d,g Capital (South) 30 (28.6) 121 (28.8) 3651 (32.8) Inner South 18 (17.1) 69 (16.4) 1232 (11.1) West 12 (11.4) 50 (11.9) 1128 (10.1) East 26 (24.8) 106 (25.2) 2373 (21.3) North 19 (18.1) 74 (17.6) 2737 (24.6) Missing 0 0 41 (0.4) Available FFQ, n (%) 105 (100.0) 419 (99.8) 10 214 (91.5) Missing 0 1 (0.2) 948 (8.5) Anthropometric measures at baseline: measured, n (%) 103 (98.1) 416 (99.0) 10 420 (93.4) BMI, median (IQR) 17.7 (16.1–19.5) 17.4 (15.9–19.2) 0.447e,f 17.3 (15.8–19.2) 0.269e,g Missing 2 (1.9) 0 WHTR, mean ± SD 0.44 ± 0.05 0.43 ± 0.04 0.101c,f 0.43 ± 0.05 0.070c,g Missing (%) 2 (1.9) 2 (0.5) 742 (6.6) Diagnosis, n (%) DM 34 (32.4) AIT 39 (37.1) JIA 18 (17.1) IBD 14 (13.3) Median age at diagnoses (IQR) 13.8 (12.3–15.5) a Of the 11 407 school-aged children in the background cohort, 105 children with primary diagnosis (AIT, autoimmune thyroiditis; DM, type 1 diabetes mellitus; IBD, inflammatory bowel diseases; JIA, juvenile idiopathic arthritis) at least 1 month after baseline and available Food Frequency Questionnaire (FFQ) generated the case group. SD=Standard Deviation, IQR = Interquartile Range, BMI=Body Mass index (kg/m2), WHTR = Waist to height Ratio. b Four children with matching age, sex, and residential areas were chosen for each child in the case group, generating the control group. c Independent samples t-test. d Pearson's chi-square test. e Kruskall-Wallis test. f Between case and controls. g Between case and all children without studied autoimmune diagnoses in the cohort (including the controls).

Central obesity was more prevalent in the 105 cases than in the 420 matched controls (16.2% vs. 9.8%) (Table 2). In fact, children with central obesity at baseline were 2.11 (95% CI 1.11–3.98) times more likely to develop an autoimmune disease during the follow-up period than those without central obesity (Figure 1). This finding was not related to any particular diagnosis, hence supported the theory that central obesity could be a potential risk factor for autoimmune diseases in general. In addition, 19.0% of children in the case group were overweight compared with 14.3% in the control group. However, being overweight was not related to developing an autoimmune disease (OR 1.60, 95% CI 0.89–2.87). Approximately 2.6 years after baseline, 253 (48%) children (of which 52 developed an autoimmune disease) had body measures available. Of them, 92.8% remained in their previous WTHR category and 89.9% in their previous BMI category. No children in this study were underweight.

TABLE 2. Associations of central obesity and being overweight in school-aged children with the onset of paediatric autoimmune diseases (DM, AIT, JIA, and IBD) Baseline anthropometric measures Casesa Controlsb Odds ratio (95% CI) Unadjusted Adjusted Autoimmune diseases Central obesity, n (%) N = 105 N = 420 No (WHTR<0.5) 86 (81.9) 373 (88.8) Reference Reference Yes (WHTR ≥0.5) 17 (16.2) 41 (9.8) 1.93 (1.04–3.57) 2.11 (1.11–3.98) Missing 2 (1.9) 6 (1.4) BMI categories, n (%) Normal weight 83 (79.0) 356 (84.8) Reference Reference Overweight 20 (19.0) 60 (14.3) 1.51 (0.86–2.67) 1.60 (0.89–2.87) Missing 2 (1.9) 4 (1.0) DM Central obesity, n (%) N = 34 N = 136 No (WHTR<0.5) 26 (76.5) 122 (89.7) Reference Reference Yes (WHTR ≥0.5) 6 (17.6) 12 (8.8) 2.97 (0.97–9.05) 3.20 (0.97–10.5) Missing 2 (5.9) 2 (1.5) BMI categories, n (%) Normal weight 28 (82.4) 113 (83.1) Reference Reference Overweight 4 (11.7) 22 (16.2) 0.80 (0.24–2.71) 0.86 (0.24–3.04) Missing 2 (5.9) 1 (0.7) AIT Central obesity, n (%) N = 39 N = 156 No (WHTR<0.5) 34 (87.2) 139 (89.1) Reference Reference Yes (WHTR ≥0.5) 5 (12.8) 16 (10.3) 1.28 (0.44–3.70) 1.36 (0.45–4.10) Missing 0 1 (0.6) BMI categories, n (%) Normal weight 30 (76.9) 133 (85.3) Reference Reference Overweight 9 (23.1) 23 (14.7) 1.66 (0.73–2.80) 1.70 (0.72–4.02) Missing 0 0 JIA Central obesity, n (%) N = 18 N = 72 No (WHTR <0.5) 15 (83.3) 61 (84.7) Reference Reference Yes (WHTR ≥0.5) 3 (16.7) 8 (11.1) 1.61 (0.38–6.81) 1.55 (0.28–8.63) Missing 0 3 (4.2) BMI categories, n (%) Normal weight 14 (77.8) 60 (83.3) Reference Reference Overweight 4 (22.2) 9 (12.5) 2.01 (0.54–7.51) 1.76 (0.36–8.59) Missing 2 (11.1) 3 (4.2) IBD Central obesity, n (%) N = 14 N = 56 No (WHTR<0.5) 11 (78.6) 51 (91.1) Reference Reference Yes (WHTR ≥0.5) 3 (21.4) 5 (8.9) 2.90 (0.56–15.0) 2.55 (0.45–14.6) Missing 0 0 BMI categories, n (%) Normal weight 11 (78.6) 50 (89.3) Reference Reference Overweight 3 (21.4) 6 (10.7) 1.66 (0.73–2.80) 2.16 (0.34–13.7) Missing 0 0 a Data were collected approximately 2 years prior to diagnosis. Median age at the time of the diagnosis was 13.75 (IQR 12.25–15.54).Of the 11 407 school-aged children in the background cohort, 105 children who obtained primary diagnosis (AIT, autoimmune thyroiditis; DM, type 1 diabetes mellitus; IBD, inflammatory bowel diseases; JIA, juvenile idiopathic arthritis) at least 1 month after baseline and had available Food Frequency Questionnaire generated the case group. OR, odds ratio; CI, confidence interval; WHTR, waist to height ratio; BMI, body mass index, weight (kg)/height2 (m2). Categorization was based on IOTF cut-offs.24 No children were underweight in this study. b Each child in the case group were compared with four childre