Abbreviations BMI body mass index DiOGenes diet, obesity and genes RCTs randomized controlled trials LCD low calorie diet GI glycaemic index CID clinical investigation days BIA bioelectrical impedance analysis DXA dual-energy X-ray absorptiometry HOMA-IR homeostatic model assessment for insulin resistance SD standard deviation 1 INTRODUCTION

Childhood obesity is a growing epidemic and has been recognized as one of the most serious public health challenges.1 In 2016, the prevalence increased by 9 million since 1990 in infants and young children (aged 0–5 years), whereas the prevalence had increased by 14% since 1975 in children and young adolescents (aged 5–19 years).2 Childhood obesity is often carried through into adolescence3 and adulthood4-7 as shown by longitudinal studies. Besides resulting in obesity in adulthood, childhood obesity increases the risk of many health problems, including dyslipidaemia, hypertension, fatty liver disease, type 2 diabetes mellitus, and psychosocial complications.8, 9 Notably, children of parents with overweight were found to be at greater risk of having overweight themselves.10 Parental obesity more than doubles the risk of obesity during adulthood among children with or without obesity under 10 years of age.11, 12 Especially maternal obesity was found to be a significant driver for childhood obesity, in which epigenetic transmission from mother to offspring may play a causal role in the predisposition of the child to become overweight or have obesity.13, 14 Parental involvement is the key mediator in driving the obesogenic environment at home. Besides parental obesity, the socioeconomic status, education level, smoking, alcohol consumption, dietary consumption, and physical activity behaviours of the parents also contribute to the obesogenic environment.13, 15 Moreover, the attitude, beliefs, and behaviour of the parents play an important role in the eating behaviour of the child and the decision-making regarding the types of food available at home and its preparation.16, 17 Therefore, parents are often involved and targeted in interventions for child weight management. Nevertheless, parental involvement and treatment components vary across childhood obesity interventions. Systematic reviews have shown greater effectiveness of family-based parent–child interventions and parent-only interventions compared to child-only interventions in improving a child's weight, body mass index (BMI), and body composition.18 This indicates the importance of parental involvement in weight management and lifestyle interventions targeting childhood obesity. Likewise, greater overall effectiveness was found in studies with greater parental involvement and multiple treatment components, including dietary monitoring, physical activity and behavioural modification, compared to single-component interventions.18 Nevertheless, parental involvement may be less predictive of successful long-term weight management in children and drop-out rates in family-based long-term weight management programs are often high.

In 2006, a multicentre, randomized, controlled dietary intervention study (Diet, Obesity and Genes [DiOGenes]), in families with at least one parent with overweight or obesity and their children, was conducted, focusing on lifestyle changes for weight maintenance of the whole family.19 Overweight and obesity were targeted from a dietary perspective and families received dietary and behavioural advice. Although evidence, based on a limited number of randomized controlled trials (RCTs), has shown that BMI changes in children with overweight or obesity and their parents with overweight or obesity were correlated during lifestyle-changing interventions in single-centre trials,20, 21 more research is needed to contribute to the existing evidence and gain more insight in managing children's weight by use of family-based approaches. Furthermore, it remains to be determined whether similar associations between changes in body weight of children and their parents can be found, regardless of their initial weight status, in the context of parental weight-maintenance intervention.

The main aim of this study was to investigate the relationship between children's BMI-for-age Z-score change and parental BMI change during the 6 months family-based dietary intervention focusing on weight maintenance for the whole family (DiOGenes), after the parents had lost weight by means of a dietary intervention. Furthermore, as a secondary aim, the relationship of changes in other metabolic parameters between adults and children was determined.

2 METHODS 2.1 Study Design

This is a secondary analysis of the DiOGenes study, a multicentre, randomized, controlled, family-based dietary intervention study, executed between November 2005 and April 2007 at eight European centres: Maastricht (The Netherlands), Copenhagen (Denmark), Cambridge (United Kingdom), Heraklion (Greece), Potsdam (Germany), Pamplona (Spain), Sofia (Bulgaria) and Prague (The Czech Republic). Parents with overweight or obesity followed an 8-week low-calorie diet (LCD) after screening and inclusion in the study. During the LCD period, their children received no instructions with respect to their food intake. Families of which (one of) the parent(s) managed to have a > 8% weight loss were randomized to one of the five diet groups for 6 months. The simple block randomization procedure was used with the centre, the number of parents with overweight or obesity in the family and BMI (greater or less than 34 kg/m2, screening value) as stratification criteria. A third party performed the randomization, without any knowledge about the subjects, which might affect the outcome of the study. The five intervention diet groups were a high protein/high glycemic index (GI) group, a low protein/high GI group, a high protein/low GI group, a low protein/low GI group and a healthy diet control group. The control group followed the national dietary guidelines in each of the countries, with a medium protein content and without specific instructions on GI. During the 6-month dietary intervention, dietary counselling was arranged every 2–4 weeks. The children were encouraged to be involved in the dietary counselling together with their parents, otherwise, parents were instructed to aid their children with their randomized diets. During the intervention, the families were provided with recipes, together with cooking and behavioural advice. Parents with children below the age of 5 years were advised to feed these children according to local dietary guidelines. The children were requested to attend six counselling sessions, accompanied by their parents, during which intensive guidance was provided. Dietitians advised on weight control and reinforced the diet composition messages through food-choice and behaviour-modification advice. In addition, physical activity and general health issues were also covered by the dieticians. Dietary instructions on the ad libitum diets were provided by trained dietitians, as described previously.22 Participants were instructed to maintain their habitual physical activity. The adults and children attended clinical examination visits to measure anthropometric parameters and body composition and to collect blood samples at each clinical investigation day (CID). The adults underwent three CIDs, before (CID1) and at the end of the LCD period (CID2), and after the 6-month weight-maintenance phase (CID3). Furthermore, the children attended two CIDs corresponding in time with CIDs 2 and 3 of their parents (Figure 1).

The timeline of the study for children and their parents. Weight-loss phase for only parents. Weight-maintenance phase for both parents and children. CID, clinical investigation day

The local ethics committees in the respective countries approved that the study protocol was in accordance with the Declaration of Helsinki and the United Nations Convention on the Rights of the Child. All individuals participating in the study signed an informed consent document according to local legislation after verbal and written instructions. The informed consent document of the children was obtained from their parents or from the child itself from the age of 12 onwards. The study was registered with ClinicalTrials.gov, number NCT00390637. The design of the DiOGenes study has been described extensively elsewhere.19

2.2 Participants

Families of the DiOGenes trial, of which at least one of the parents was overweight or with obesity and completed the 8-week weight-loss phase successfully, were included in the 6-month dietary intervention study. Parents who participated in the study were healthy and below 65 years of age. Furthermore, at least one parent had a BMI between 27 and 45 kg/m2 at screening and at least one healthy child aged between 5 and 18 years living in their household. The children who were included in this study participated together with their families. Exclusion criteria for parents were BMI >45 kg/m2, heart or coronary diseases, kidney or liver diseases, psychiatric illnesses, systemic infections or endocrine diseases, weight change >3 kg within 2 months prior to first CID and using prescription medication that might interfere with the outcome of the study. Exclusion criteria for children were systemic infections, chronic disease, or special diets. Further details on recruitment strategies and inclusion and exclusion criteria of the parents and children are provided by Larsen et al..19

2.3 Anthropometric measurements and blood sampling

For all investigations, the same standard operating procedures were used in all centres and all blood samples were analysed in one laboratory to ensure standardization across centres. The children were fasted for 4 h and adults for at least 10 h prior to the investigations. Body weight, height, waist- and hip circumference, and fat percentage (%) were measured and blood samples were collected to determine fasting glucose and insulin as described in a previous paper.19 Subjects were measured wearing underwear and with an empty bladder. Weight was measured to the nearest 0.1 kg. Height was measured with a wall-mounted stadiometer to the nearest 0.5 cm without shoes. The circumference was measured to the nearest 0.5 cm with a tape, if possible, with the subject standing. Waist circumference was measured between the bottom of the ribs and the top of the hip bone. Hip circumference was measured as the largest circumference in the area around the buttocks. Body composition was assessed by bioelectrical impedance analysis (BIA; Quad-Scan 4000; Bodystat, Douglas, Isle of Man, United Kingdom) at five centres (Maastricht, Pamplona, Heraklion, Sofia and Prague), by dual-energy X-ray absorptiometry (DXA; Lunar Radiation, Madison, WI) at two centres (Copenhagen and Cambridge) and by both DXA and BIA at one centre (Potsdam).

Blood samples were drawn from a Venflon after an overnight fast of at least 10 h and plasma and serum were stored at −80°C until analysis.19, 23 Serum glucose was measured by a colorimetric assay (Ortho-Clinical Diagnostics) for the Vitros 950 analyser and serum insulin was measured by an immunoassay (Siemens Healthcare Diagnostics) for the ADVIA Centaur XP.23 Homeostatic model assessment (HOMA) was used to estimate insulin resistance (IR) from basal (fasting) glucose and insulin. The homeostatic model assessment for insulin resistance (HOMA-IR) index was calculated as follows: (fasting insulin (mIU/L) × fasting glucose (mmol/L))/22.5.24 For adults, BMI was calculated as the body weight/height2 (kg/m2). For children, gender- and age-specific Z-scores for BMI were calculated using WHO Anthro (0–5 y) and WHO AnthroPlus (5–19 y) software.25-28

2.4 Data analysis

Results are reported as mean ± standard deviation (SD) unless indicated otherwise. The normality of data was assessed with the Shapiro–Wilk test procedure and ln transformation was used if the assumption of normality was not met. Paired t-tests were implemented to determine the significance of changes of anthropometric measures, fasting glucose, fasting insulin and HOMA-IR in adults and children over the dietary intervention. Correlations between the changes in BMI-for-age of children and the changes in BMI of father or mother were determined using Pearson correlation coefficient analysis. Linear regression analyses were conducted to examine the relationship between the changes of anthropometric measures, fasting glucose, fasting insulin, and HOMA-IR in mothers, fathers, and children over the diet intervention period. Furthermore, multiple linear regression analyses were conducted with child measures as dependent variables and measures of both parents as independent variables. The analyses were adjusted for diet group, centre, child's sex and age. All data were analysed by using IBM SPSS Statistics 25 and significance was defined as p < 0.05.

3 RESULTS

Only the first and/or second children of families were included in this analysis, resulting in a total of 343 families, comprised of 426 adults and 468 children. Many children had no blood draw or measurement of fat percentage as children did not have to participate in these measurements when they did not want to. For the first children, blood drawing and BIA or DXA scan were performed in 65% ( n = 206) and 55% ( n = 174) of the children, respectively. For the 2nd children, blood drawing and BIA or DXA scan were performed in 53% ( n = 80) and 59% ( n = 89) of the children, respectively. Thus, the sample size varies for the different variables. The average age was 12.6 ± 3.5 years for the first (oldest) child and 10.7 ± 3.2 for the second (younger) child. The average age for the parents was 43.4 ± 5.7 for the fathers and 41.9 ± 5.8 for the mothers (Table 1). Before the start of the weight-maintenance phase, the father and/or the mother had successfully lost at least 8% of their initial body weight during the weight-loss phase. At the start of the weight-maintenance phase, the mean BMI-for-age Z-score for the first child was 1.2 ± 1.2 and for the second child, it was 1.1 ± 1.3. Of the first children, overweight or obesity was present in 58.7% of the children, whereas for the second children, this was 53.6%. The mean BMI of the father was 30.4 ± 4.1 kg/m2, whereas for the mother, the mean BMI was 30.7 ± 4.6 kg/m2. Furthermore, the study population showed a healthy average fasting glucose value at the start of the weight-maintenance phase in fathers, mothers, first children, and second children (4.9 ± 0.5, 4.7 ± 0.6, 4.6 ± 0.5, and 4.7 ± 0.4 mmol/L, respectively). HOMA-IR for fathers was 2.2 ± 2.9 and for mothers, it was 1.8 ± 1.2. For first children, HOMA-IR was 3.3 ± 2.6 and for second children, it was 2.8 ± 1.9.

TABLE 1. Characteristics of the study population at randomization (at the beginning of the weight-maintenance phase) Family ( n = 343) Adults ( n = 436) Children ( n = 468) Father ( n = 160) Mother ( n = 276) First child ( n = 317) Second child ( n = 151) Age (year) 43.4 ± 5.7 41.9 ± 5.8 12.6 ± 3.5 10.7 ± 3.2 Height (cm) 1.78 ± 0.08 1.66 ± 0.07 1.57 ± 0.17 1.48 ± 0.17 Weight (kg) 95.8 ± 15.4 84.6 ± 14.6 58.0 ± 22.2 47.6 ± 18.8 BMI (kg/m2) 30.0 ± 4.1 30.7 ± 4.6 22.7 ± 5.3 20.7 ± 4.7 BMI (for-age Z-score) 1.2 ± 1.2 1.1 ± 1.3 Waist circumference (cm) 101.9 ± 11.8 94.8 ± 11.7 75.4 ± 13.7 70.0 ± 13.2 Hip circumference (cm) 105.9 ± 9.5 110.9 ± 10.7 88.6 ± 15.8 82.0 ± 15.7 Fat (%) 27.8 ± 6.3 40.2 ± 6.1 28.6 ± 9.0 29.2 ± 9.3 HOMA-IR 2.2 ± 2.9 1.8 ± 1.2 3.3 ± 2.6 2.8 ± 1.9 Fasting glucose (mmol/L) 4.9 ± 0.5 4.7 ± 0.6 4.6 ± 0.5 4.7 ± 0.4 Fasting insulin (mIU/L) 8.6 ± 10.4 7.3 ± 4.5 16.0 ± 13.1 13.2 ± 8.8 Centres ( n (%)) The Netherlands 39 (24.4) 59 (21.5) 72 (22.7) 41 (27.2) Denmark 42 (26.3) 58 (21.2) 60 (18.9) 52 (34.4) United Kingdom 12 (7.5) 34 (12.4) 37 (11.7) 19 (12.6) Greece 8 (5.0) 17 (6.2) 16 (5.0) 7 (4.6) Germany 15 (9.4) 21 (7.7) 28 (8.8) 6 (4.0) Spain 20 (12.5) 27 (9.9) 37 (11.7) 23 (15.2) Bulgaria 13 (8.1) 34 (12.4) 40 (12.6) 0 (0) Czech Republic 11 (6.9) 26 (9.5) 27 (8.5) 3 (2.0) Diet ( n (%)) LP/LGI 30 (18.8) 57 (20.8) 65 (20.5) 30 (19.9) LP/HGI 29 (18.1) 48 (17.5) 55 (17.4) 27 (17.9) HP/LGI 39 (24.4) 59 (21.5) 71 (22.4) 33 (21.9) HP/HGI 31 (19.4) 53 (19.3) 59 (18.6) 35 (23.2) Healthy diet 31 (19.4) 59 (21.5) 67 (21.1) 26 (17.2) Note: Values are presented in mean ± SD, unless otherwise specified. Adult's BMI is presented kg/m2, whereas children's BMI is presented in BMI-for-age Z-score. Missing values: Father: waist circumference, hip circumference ( n = 3); fat percentage ( n = 22); fasting glucose ( n = 4); HOMA-IR, fasting insulin (n = 15). Mother: waist circumference, hip circumference ( n = 5); fat percentage ( n = 44); fasting glucose ( n = 7); HOMA-IR ( n = 29); fasting insulin ( n = 24). First child: waist circumference ( n = 32); hip circumference ( n = 33); fat percentage ( n = 144); fasting glucose ( n = 111); HOMA-IR, fasting insulin ( n = 124); Second child: waist circumference ( n = 2); hip circumference ( n = 3); fat percentage ( n = 62); fasting glucose ( n = 71); HOMA-IR, fasting insulin ( n = 78). Abbreviations: BMI, body mass index; HGI, high glycemic index; HOMA-IR, homeostatic model assessment for insulin resistance; HP, high protein; LGI, low glycemic index; LP, low protein. 3.1 Changes in anthropometry and metabolic parameters

The BMI and waist circumference of the fathers increased significantly over the 6-month intervention period, whereas no change was found in the mothers (Table 2). The BMI Z-score was decreased significantly in the first child and second child. Moreover, the fat percentage decreased significantly in mothers and first and second children. Furthermore, HOMA-IR, fasting glucose, and fasting insulin were increased significantly in both fathers and mothers, whereas no significant changes were found in children.

TABLE 2. Changes in anthropometry and metabolic parameters before and after the weight-maintenance phase in adults and children Adults Children Father Mother First child Second child Change p Change p Change p Change p BMI (kg/m2 or Z-score) 0.5 ± 1.6 0.000 0.0 ± 2.0 1.000 −0.1 ± 0.4 0.000 −0.1 ± 0.6 0.044 Waist circumference (cm) 1.2 ± 6.5 0.032 0.0 ± 7.0 0.925 −0.5 ± 5.1 0.101 −0.2 ± 4.3 0.568 Hip circumference (cm) −0.2 ± 4.6 0.656 −0.5 ± 6.7 0.243 0.6 ± 5.3 0.054 0.5 ± 3.9 0.165 Fat (%) −0.6 ± 3.5 0.069 −1.3 ± 4.6 0.000 −1.0 ± 4.9 0.011 −1.0 ± 3.6 0.012 HOMA-IR 0.5 ± 2.0 0.008 0.3 ± 1.1 0.000 −0.1 ± 2.0 0.698 0.1 ± 2.2 0.787 Fasting glucose (mmol/L) 0.2 ± 0.5 0.000 0.1 ± 0.5 0.000 0.0 ± 0.5 0.720 0.0 ± 0.6 0.687 Fasting insulin (mIU/L) 1.5 ± 6.3 0.007 0.9 ± 3.7 0.001 −0.5 ± 9.5 0.575 0.0 ± 9.0 0.974 Note: Values of change are presented in mean change ± SD. The normality of data was assessed and In transformation was used for fasting glucose of the second child. Bold values denote significant change p < 0.05. Adult BMI is presented in kg/m2, whereas children's BMI is presented in BMI-for-age Z-score. Missing values: Father: waist circumference ( n = 4); hip circumference ( n = 3); fat percentage ( n = 26); fasting glucose ( n = 7); HOMA-IR ( n = 23); fasting insulin ( n = 22). Mother: BMI ( n = 2); waist circumference ( n = 23); hip circumference ( n = 13); fat percentage ( n = 60); fasting glucose ( n = 18); HOMA-IR ( n = 4); fasting insulin ( n = 50). First child: BMI ( n = 1); waist circumference, hip circumference ( n = 34); fat percentage ( n = 148); fasting glucose ( n = 159); HOMA- IR, fasting insulin ( n = 171); second child: waist circumference ( n = 2); hip circumference ( n = 3); fat percentage ( n = 67); fasting glucose ( n = 85); HOMA-IR, fasting insulin ( n = 91). Abbreviations: BMI, body mass index; HOMA-IR, homeostatic model assessment for insulin resistance. 3.2 Association between changes in anthropometry in parents and children

The change in BMI of the mother was found to be a significant predictor for the change in BMI-for-age Z-score of the first and second child (first child unadjusted standardized [std] = 0.248, p = 0.000, second child unadjusted std = 0.326, p = 0.000). The change in BMI of the father was only found to be a significant predictor for the change in BMI-for-age Z-score of the first child (unadjusted std = 0.186, p = 0.031) (Figure 2 and Table 3). Adjustment for the sex and age of the child, diet and centre did not change these results. However, when BMI changes of both parents were included in the model, only the change in BMI of the mother was a significant predictor for the change in BMI-for-age Z-score of the first child and second child, with and without adjustment for child's age and sex, diet and centre (mother unadjusted std = 0.322, p = 0.011, father unadjusted std = 0.098, p = 0.433 for first child; mother unadjusted std = 0.480, p = 0.058, father unadjusted std = −0.086, p = 0.626 for second child; Table 4). Adjustment for the sex and age of the child, diet and centre did not change these results. Subgroup analyses revealed that 57.1% of the children were overweight or with obesity. Stronger parental associations for changes in BMI-for-age Z-score in first child with overweight or obesity were found compared to lean children. However, in the second child, stronger associations for lean children were found compared to children with overweight or obesity. Partial analysis revealed that the adiposity status, determined using the WHO cut-points for BMI-for-age Z-score, only had a significant impact on the association between the change in BMI Z-score of the first child and the change in BMI of the mother (r = 0.245, p = 0.000). Furthermore, the change in waist circumference of the mother was found to be a significant predictor for the change in waist circumference of the first child, with or without adjustment for child's age and sex, diet and centre (mother unadjusted std = 0.245, p = 0.000; father unadjusted std = −0.008, p = 0.933; Table 3). Associations were not significant for the second child (Table 3). When both parents were included in the regression model, a significant association was found between changes in waist circumference of both father and mother with changes in waist circumference of the first child, with or without adjustment for child's age and sex, diet and centre (father unadjusted std = 0.280, p = 0.012; mother unadjusted std = 0.432, p = 0.000; Table 4).

Correlation analysis of changes in BMI-for-age of first or second child and changes in BMI of father or mother. (A) Changes in BMI of the father or (B) changes in BMI of the mother were positively associated with the changes in BMI-for-age of the first child. There was no significant association between the (C) changes in BMI of the father or (D) changes in BMI of the mother and changes in BMI-for-age of the second child

TABLE 3. Linear regression analysis of the associations of changes in anthropometry and metabolic parameters over the weight-maintenance phase between father or mother and first or second child PREDICTORS First child Second child n std p std adj p n std p std adj p Father BMI (kg/m2 or Z-score) 136

0.186 ± 0.019

0.031

0.203 ± 0.021

0.033 69

0.051 ± 0.040

0.675

0.134 ± 0.042

0.296 Waist circumference (cm) 123

−0.008 ± 0.058

0.933

0.019 ± 0.063

0.850 70

0.038 ± 0.060

0.753

0.040 ± 0.065

0.758 Hip circumference (cm) 124

0.047 ± 0.118

0.606

0.140 ± 0.117

0.121 70

−0.078 ± 0.108

0.522

0.107 ± 0.110

0.385 Fat (%) 63

0.191 ± 0.163

0.134

0.187 ± 0.171

0.163 33

0.253 ± 0.213

0.163

0.252 ± 0.268

0.271 HOMA-IR 50 −0.195 ± 0.093 0.174

−0.227 ± 0.112

0.193