Who Benefited Most? Identifying Parent–Child Dyads With Improved Weight Outcomes in the Feel4Diabetes Study

ABSTRACT

Background

Family‐based interventions offer a promising solution to the pressing global health concern of childhood obesity by addressing the interconnected behaviours of parents and children. Nevertheless, limited evidence identifies which families benefit most when outcomes are evaluated at the parent–child dyad level.

Aim

To identify which parent–child dyads experienced clinically meaningful improvements in weight status over 1‐ and 2‐year periods within the Feel4Diabetes study, a school‐ and community‐based intervention targeting European families at risk for Type 2 diabetes.

Methods

Data were drawn from 2402 high‐risk families across six European countries. For participants with overweight/obesity, improvement meant ≥ 5% parental weight loss and a ≥ 0.15‐unit child BMI‐z reduction. Multiple logistic regression models identified socio‐demographic and contextual predictors of dyadic improvement.

Results

After 2 years, 5.3% of families achieved improvements in both parent and child. Significant predictors of dyadic success included intervention group participation [OR = 2.56 (95% CI: 1.31–5.03)] and residing in high‐income countries under austerity measures [OR = 2.44 (1.00–6.02)]. Girls were more likely to improve than boys, while full‐time maternal employment was associated with lower odds of dyadic improvement.

Conclusions

Sustained benefits were observed in a subset of families, highlighting the need for tailored, context‐sensitive family‐centred obesity prevention strategies.

1. Introduction

Childhood obesity has become a global public health challenge, with alarming prevalence rates and serious long‐term consequences for cardiometabolic health, including elevated risks of Type 2 diabetes and cardiovascular disease [1]. In response, there is growing recognition among researchers and public health professionals that effective interventions must go beyond targeting the child alone. Instead, they should address the broader family context, given the influential role of the home environment and the interconnected nature of parent and child health behaviours [2, 3]. Within this framework, family‐based lifestyle interventions have gained popularity as a particularly promising strategy [4].

However, not all families benefit equally from family‐based lifestyle interventions [5, 6]. This heterogeneity in intervention response may reflect the multifactorial determinants of obesity, which—together with its associated health behaviours—is shaped by a complex interplay of child characteristics (e.g., age and sex), parental socio‐demographic characteristics (e.g., age, educational level and employment status), characteristics of the intervention received and the broader socioeconomic environment in which families live [7, 8, 9, 10, 11]. In the context of family‐based interventions, understanding how these factors relate to changes in both parent and child outcomes is of particular importance.

Evidence consistently demonstrates a strong association between parental and child weight status, shaped by shared genetics, behaviours and environmental exposures [12, 13]. Within the framework of the Feel4Diabetes study—a multi‐country school‐ and community‐based intervention targeting vulnerable families across six European countries—our previous analyses have shown that reductions in parental body mass index (BMI) were significantly associated with favourable changes in children's BMI z‐scores (BMIz) among families at elevated risk for Type 2 diabetes [14]. Simultaneous improvements in parents and children may reflect deeper, more cohesive changes in household behaviours—ranging from healthier food practices to more active lifestyles—and could indicate stronger engagement with the intervention and motivation to reduce health risks [15, 16]. Νevertheless, despite this evidence, relatively little is known about whether the factors associated with obesity risk and family engagement are also associated with concurrent, clinically meaningful improvements in weight status when parent and child outcomes are evaluated jointly as dyads, particularly over longer follow‐up periods. Identifying these characteristics may help optimize family‐based interventions by informing more targeted and equitable obesity prevention strategies.

Therefore, the aim of the present study was to identify the socio‐demographic and contextual characteristics associated with clinically meaningful improvements in weight status among parent–child dyads participating in the Feel4Diabetes intervention over 1‐ and 2‐year follow‐up periods. Specifically, we examined whether child characteristics (age and sex), parental characteristics (age, educational level and employment status), characteristics of the intervention received and country economic classification were associated with concurrent improvements in parental body weight (BW) and children's BMIz.

2. Methods

2.1. Bioethics

The Feel4Diabetes study complied with the Declaration of Helsinki [17] and the conventions of the Council of Europe on human rights and biomedicine. The Feel4Diabetes‐study adhered to the Declaration of Helsinki and the conventions of the Council of Europe on human rights and biomedicine. All participating countries obtained ethical clearance from the relevant ethical committees and local authorities. More specifically, in Belgium the study was approved by the Medical Ethics Committee of the Ghent University Hospital (ethical approval code: B670201524437); in Bulgaria, by the Ethics Committee of the Medical University of Varna (ethical approval code: 52/10‐3‐2016r) and the Municipalities of Sofia and Varna, as well as the Ministry of Education and Science local representatives; in Finland, by the hospital district of Southwest Finland ethical committee (ethical approval code: 174/1801/2015); in Greece, by the Bioethics Committee of Harokopio University (ethical approval code: 46/3‐4‐2015) and the Greek Ministry of Education; in Hungary, by the National Committee for Scientific Research in Medicine (ethical approval code: 20095/2016/EKU) and in Spain, by the Clinical Research Ethics Committee and the Department of Consumers' Health of the Government of Aragón (ethical approval code: CP03/2016). Informed written consent was given by all participants before study commencement [18].

2.2. Study Design

The present study is a secondary analysis of data from the Feel4Diabetes study, a multicentre, school‐ and community‐based, cluster‐randomized intervention conducted between 2016 and 2018 in six European countries (Belgium, Bulgaria, Finland, Greece, Hungary and Spain). The intervention aimed to promote healthy lifestyle behaviours and prevent Type 2 diabetes among vulnerable individuals through a multicomponent approach combining school‐, community‐ and family‐based strategies.

Α standardized multistage sampling strategy was employed to recruit participants. Vulnerable families were defined as those living in low‐ and middle‐income countries (e.g., Bulgaria, Hungary), in socioeconomically disadvantaged areas within high‐income countries (e.g., Belgium, Finland) or in nations experiencing austerity measures (e.g., Greece, Spain). Following baseline assessments, schools and participating families were randomized at the munivipality level (1:1 ratio) to either the intervention or standard care group.

The intervention comprised two components. All participating families received a universal school‐ and community‐based intervention promoting healthy eating, increased physical activity, reduced sedentary behaviour and healthier family lifestyle habits. In addition, families identified as being at high risk for Type 2 diabetes using the Finnish Diabetes Risk Score (FINDRISC) [11] received an enhanced intervention consisting of seven lifestyle counselling sessions delivered by trained health professionals during the first year, followed by a text‐message support programme during the second year.

The baseline cohort, recruited in 2016, comprised 12 280 children and their parents across the six participating countries. The present analyses were conducted among the 2402 high‐risk families with available data for the outcomes of interest. Further methodological details regarding study design, recruitment procedures and intervention are provided in Manios et al. [19].

2.3. Anthropometric Measurements

At baseline and during two follow‐up periods, trained personnel performed standardized assessments of children's and parents' BW and height. To minimize measurement bias, validated instruments and consistent measurement protocols were used. Height was measured using telescopic stadiometers (SECA models 213, 214, 217 and 225), while BW was recorded using digital scales (SECA 813 and 877). Participants were measured barefoot and in light clothing. Height and weight were recorded to the nearest 0.1 cm and 0.1 kg, respectively. BMI was calculated as weight in kg divided by height in metres squared and its reporting followed the WHO classification for adults [20]. The BMIz was calculated and classified following the WHO standards [21]. The main outcome of interest in the present study was a simultaneous reduction of ≥ 5% in BW among parents with overweight or obesity and a ≥ 0.15 decrease in BMIz among children with overweight or obesity over the 1‐ and 2‐year periods.

2.4. Statistical Analysis

Categorical variables are presented as frequencies (n) and percentages (%), while continuous variables are summarized using means and standard deviations (SDs). Several baseline characteristics reflecting the families' socio‐economic background were investigated as potential predictors of BW/BMIz improvement. These included parental education (categorized as: < 9, 9–14 and > 14 years), employment status (unemployed/other, full‐time, part‐time) and age (< 45 vs. ≥ 45 years). Additionally, country's economic classification, grouped into three categories: low‐income (Bulgaria and Hungary), high‐income under economic austerity (Greece and Spain) and high‐income without austerity (Belgium and Finland), as defined in Moschonis et al. [22] Other covariates included group assignment (intervention vs. standard care) and child's sex (boy or girl), treated as categorical variables, while child's age was used as a continuous variable. Improvement in parental BW was defined as a ≥ 5% reduction from baseline among those classified as having overweight or obesity, assessed over 1 year and again over the two‐year follow‐up. For children, improvement in BMIz was defined as a decrease of ≥ 0.15 units from baseline among those who were living with overweight or obesity, also evaluated at 1 and 2 years. Parent–child pair improvement was defined as simultaneous achievement of both aforementioned criteria. Thus, the outcome was a two‐level categorical variable (0: no improvement; 1: improvement). χ ² test was applied to investigate the impact of the intervention on the improvement of both parent's BW and children's BMIz as a pair, in HR‐families. Also, χ ² tests or Fisher's exact test and t‐test were used to identify if the economic classification of countries and if the various socio‐demographic collected at baseline were associated with an improvement in both parental and children's BW/BMIz. Multiple logistic regression models were fitted to evaluate predictors of BW/BMIz improvement over 1 and 2 years, adjusting for relevant study design variables. A p‐value less than 0.05 was considered statistically significant, while less than 0.10 as indicative. All analyses were performed in STATA (StataCorp. 2015. Stata: Release 15. Statistical Software).

3. Results

At baseline, 30% of mothers and 47.3% of fathers in high‐risk families (N = 2402) were classified as having overweight, while 21.9% of mothers and 33.4% of fathers met criteria for having obesity. Among children, 42.1% had a BMIz greater than +1 SD, indicating overweight or obesity. Additional baseline characteristics of the study population are presented in Table 1. Clinically meaningful improvement (≥ 5% reduction in parental BW or ≥ 0.15‐unit decrease in child BMIz) was assessed only among parents and children with overweight or obesity at baseline; the resulting proportions, expressed relative to the total number of high‐risk families (N = 2402), are presented in Table 2. Overall, a modest yet notable increase in success rates was observed over time. By the second‐year follow‐up, 8.8% of mothers, 12.8% of fathers and 18.6% of children had achieved the defined thresholds, with children consistently showing the highest rate of improvement. However, only in 5.3% of families both a parent and child succeeded in achieving weight reduction.

TABLE 1.

Baseline anthropometric and socio‐demographic characteristics of children and parents in High‐Risk families.

	Total population (N = 2402)	Intervention group (N = 1297)	Control group (N = 1105)	p
Children
Age, years	8.2 ± 1.0	8.2 ± 1.0	8.1 ± 1.0	0.679 a
Sex				0.196 b
% girls	50.9%	49.8%	52.3%
BMI z‐score	0.702 ± 1.103	0.721 ± 1.120	0.679 ± 1.082	0.351 a
BMI z‐score > +1 SD, %	42.1%	43.3%	41.5%	0.584 b
Region by income status c				0.093 b
High‐income countries (Belgium, Finland)	31.6%	29.9%	33.7%
High‐income countries under economic crisis (Greece, Spain)	39.6%	40.3%	38.7%
Low‐income (Bulgaria, Hungary)	28.9%	29.9%	27.6%
Mother
Age				0.401 b
%< 45 years	84.6%	84.0%	85.2%
Weight category				0.285 b
Normal weight, %	48.1%	47.8%	48.6%
Overweight, %	30.0%	31.1%	28.6%
Obesity, %	21.9%	21.1%	22.8%
Education				0.887 b
% < 9 years	6.5%	6.6%	6.4%
%9–14 years	32.5%	30.9%	34.5%
% > 14 years	61.0%	62.5%	59.1%
Father
Age				0.418 b
%< 45 years	67.5%	66.8%	68.3%
Weight category				0.098 b
Normal weight, %	19.3%	18.8%	19.7%
Overweight, %	47.3%	49.3%	45.0%
Obesity, %	33.4%	31.9%	35.3%
Education				0.631 b
% < 9 years	8.7%	8.1%	9.3%
%9–14 years	41.2%	41.9%	40.4%
% > 14 years	50.1%	50.0%	50.3%
Family income status d
%Difficult	49.3%	48.0%	50.9%	0.133 b

Note: Values are means ± SD or percentages (%).

^a t‐Test.

^b Pearson's χ ² test for the comparison between intervention and control groups.

^c Country income status according to Word Bank. High income: Belgium and Finland; under economic crisis: Greece and Spain; low income: Bulgaria and Hungary.

^d Based on the question: How easy or difficult do you find to cover the household expenditure? (six potential responses, merged into two).

TABLE 2.

Distribution of parents and children with overweight or obesity at baseline who achieved clinically meaningful reductions in body weight (≥ 5%) and BMI‐for‐age z‐score (BMIz ≥ 0.15 units), respectively, in the first follow‐up (FUP) and at the second FUP, in high‐risk‐ families.

High‐risk families	Baseline to 1st FUP	Baseline to 2nd FUP
Mother	7.8%	8.8%
Father	10.8%	12.8%
Child	16.6%	18.6%
At least one parent	19.9%	22.9%
At least one parent and child	4.6%	5.3%

Note: Percentages are expressed as a proportion of the total study population.

Findings related to changes in parental BW and child BMIz as a dyadic outcome, along with associated baseline socio‐demographic factors at baseline, are summarized in Table 3.

TABLE 3.

Association between parent–child pair body‐weight and body mass index‐for‐age z‐score (BMIz) prospective improvement in HR‐families and group, economic classification of countries and children's and parental socio‐demographic characteristics at baseline.

Characteristics	Improved BW‐BMIz (%) over the 1‐year period in HR‐families			Improved BW‐BMIz (%) over the 2‐year period in HR‐families
	No change	Improved	p	No change	Improved	p
Group
Intervention	95.2	4.8	0.785 c	92.6	7.4	0.003 c , *
Standard care	95.5	4.5		97.1	2.9
Child's gender (%)
Boy	96.3	3.7	0.158 c	96.2	3.8	0.066 c
Girl	94.5	5.5		93.4	6.6
Child's age (years; mean ± SD)	8.1 (0.99)	8.1 (0.97)	0.850 d	8.0 (0.97)	7.9 (1.04)	0.376 d
Country economic classification a (%)
High income	97.7	2.3	0.123 c	97.4	2.3	0.078 c
Under economic crisis	94.6	5.5		93.2	6.8
Low income	94.8	5.2		94.8	5.2
Age of mother (%)
< 45 years	95.4	4.6	0.821 c	94.7	5.3	0.758 c
≥ 45 years	95.0	5.0		95.3	4.7
Age of father (%)
< 45 years	95.8	4.2	0.596 c	94.5	5.5	0.517 c
≥ 45 years	95.1	4.9		95.5	4.5
Education of mother (%)
< 9 years	100	0	0.907 e , *	100	0	0.234 e
9–14 years	95.2	4.8		92.6	7.4
> 14 years	95.5	4.5		95.6	4.4
Education of father (%)
< 9 years	91.7	8.2	0.223 c , *	90.5	9.5	0.236 c
9–14 years	95.5	4.5		95.0	5.0
> 14 years	96.2	3.8		95.5	4.5
Occupation of mother (%)
Unemployed/other b	94.4	5.6	0.326 c	92.3	7.7	0.063 c
Employed full‐time	95.4	4.6		95.7	4.3
Employed part‐time	97.5	2.5		97.4	2.6
Occupation of father (%)
Unemployed/other b	96.1	3.9	0.855 c	92.2	7.8	0.059 c
Employed full‐time	95.5	4.5		95.6	4.4
Employed part‐time	94.2	5.8		87.9	12.1

Abbreviation: SD: standard deviation.

^a Low‐income countries (Bulgaria, Hungary), high‐income countries (Belgium, Finland) and countries under economic crisis (Greece, Spain).

^b Other: stay at home parent or full‐time student or retired or else.

^c χ ² test.

^d t‐Test.

^e Fisher's exact test.

^*Statistically significant at α = 5%; marked in bold significant results at α = 10%.

While no significant difference was observed between intervention and standard care groups after 1 year, a significantly higher proportion of parent–children dyads in the intervention group achieved BW–BMIz improvement over the 2‐year period compared to the standard care group (7.4% vs. 2.9%, p = 0.003), indicating a potential long‐term benefit of the intervention. Although not statistically significant at the conventional 5% level, gender differences approached significance, with girls showing slightly higher BMIz improvement than boys over both follow‐up periods. Families from countries under economic crisis or with lower income had numerically higher improvement rates, although these differences did not reach statistical significance. Parental age, education and employment status were not significantly associated with outcomes, although families of part‐time employed fathers showed a notably higher rate of improvement at 2 years compared to other paternal occupational categories.

To better understand which factors predicted improvement, multiple logistic regression analyses were conducted, as detailed in Table 4. Notably, participation in the intervention group was significantly associated with greater odds of improvement at 2‐year period (OR = 2.56, 95% CI: 1.31–5.03), but not at 1 year, suggesting that the intervention effects became more pronounced over time. Girls had higher odds of improvement over 2 years (OR = 1.77, 95% CI: 1.00–3.29). Families residing in countries experiencing economic crisis had significantly higher odds of improvement at both time points (OR = 2.44, 95% CI: 1.00–6.01 at 1 year and OR = 2.44, 95% CI: 1.00–6.02 at 2 years) relative to those in high‐income countries. Interestingly, full‐time employment of the mother was associated with reduced odds of improvement at 2 years (OR = 0.54, 95% CI: 0.28–0.99). Other variables, including parental age, education and father's occupation, did not demonstrate statistically significant associations.

TABLE 4.

Multiple logistic regression results (odds ratios and corresponding 95% confidence intervals) for the association between parent–child body‐weight and body mass index‐for‐age z‐score (BMIz) prospective improvement, as a pair in HR‐families, adjusting for group, country's economic classification and socio‐demographic characteristics.

Independent variables	Dependent variable: improved BW‐BMIz 1‐year	Dependent variable: improved BW‐BMIz 2‐year
	OR (95% CI)	OR (95% CI)
Group
Standard care	Reference group	Reference group
Intervention	1.02 (0.58–1.79)	2.56 (1.31–5.03)*
Child's gender
Boy	Reference group	Reference group
Girl	1.50 (0.85–2.68)	1.77 (1.00–3.29)*
Child's age (years)	1.08 (0.80–1.45)	0.92 (0.67–1.27)
Country economic classification a
High income	Reference group	Reference group
Under economic crisis	2.44 (1.00–6.01)*	2.44 (1.00–6.02)*
Low income	2.33 (0.92–5.90)	1.71 (0.63–4.60)
Age of mother
< 45 years	Reference group	Reference group
≥ 45 years	1.07 (0.49–2.34)	0.77 (0.33–1.77)
Age of father
< 45 years	Reference group	Reference group
≥ 45 years	1.11 (0.60–2.04)	0.74 (0.38–1.43)
Education of mother
< 9 years	Reference group	Reference group
9–14 years	1.02 (0.51–2.03)	0.62 (0.32–1.22)
> 14 years	1.02 (0.51–2.04)	1.71 (0.87–3.34)
Education of father
< 9 years	Reference group	Reference group
9–14 years	0.61 (0.23–1.58)	0.58 (0.22–1.55)
> 14 years	0.48 (0.19–1.25)	0.47 (0.18–1.24)
Occupation of mother
Unemployed/other b	Reference group	Reference group
Employed full‐time	0.87 (0.47–1.62)	0.54 (0.28–0.99)*
Employed part‐time	0.50 (0.16–1.53)	0.32 (0.09–1.12)
Occupation of father
Unemployed/other b	Reference group	Reference group
Employed full‐time	1.19 (0.46–3.10)	0.54 (0.24–1.23)
Employed part‐time	1.52 (0.35–6.63)	1.67 (0.46–6.03)

Abbreviations: 95% CI: 95% confidence interval; OR: odds ratio.

^a Low‐income countries (Bulgaria, Hungary), high‐income countries (Belgium, Finland) and countries under economic crisis (Greece, Spain).

^b Other: stay at home parent or full‐time student or retired or else.

^*Statistically significant at α = 5%.

4. Discussion

This study explored predictors of meaningful, concurrent weight improvements in parent–child dyads over a two‐year follow‐up period in the Feel4Diabetes study. Overall, only a small proportion of families met both criteria, highlighting the challenge of achieving simultaneous improvements in parents and children. While a reduction of 0.15 in BMIz is often considered sufficient to elicit meaningful health benefits in children [22], even smaller decreases may still yield positive outcomes at the population level when widely implemented [23]. In adults, although a weight loss of at least 5% is commonly used as a target in weight management interventions, it has been shown that weight losses of less than 5% are also associated with improvements in cardiometabolic, inflammatory, psychosocial and other health outcomes [24].

Notably, the findings indicate that participation in the intervention group was associated with a significantly greater likelihood of dyadic improvement at 2 years, suggesting that the intervention had a delayed but positive impact on families at higher metabolic risk. It is often observed that parents underestimate their children's weight status [25, 26] and it is possible that the nature of the intervention helped increase parental awareness of their child's actual weight, making them more willing to adopt behaviour changes. The observed time‐dependent effect of the intervention aligns with previous research [27] and underscores the importance of long‐term intervention design and follow‐up in evaluating public health strategies for obesity prevention.

An important finding of this study is that families residing in countries experiencing economic austerity (i.e., Greece and Spain) demonstrated significantly higher odds of dyadic improvement in weight status. While these are high‐income countries, the recruitment strategy focused exclusively on low socioeconomic status (SES) areas, in contrast to the broader recruitment across SES levels in the low‐income countries (i.e., Bulgaria and Hungary). This difference likely contributed to the observed outcomes. Lower SES has been consistently associated with a higher prevalence of obesogenic behaviours, including poor dietary patterns and physical inactivity as well as with barriers to accessing healthcare [11, 28, 29, 30]. Economic austerity measures have also been linked to reductions in healthcare access and increased social and health inequalities [31]. In this context, families affected by austerity may have perceived the community‐based support provided by the intervention as a rare and valuable resource, increasing their motivation and engagement. Therefore, rather than economic hardship alone, it may be the intersection of low SES and austerity‐related constraints that shaped a more favourable response to the intervention in these settings.

Gender emerged as a potential moderator of change, with girls showing a trend towards greater BMIz improvements than boys at the 2‐year follow‐up. Although a recent meta‐analysis found no overall sex differences in response to lifestyle interventions, it suggested that boys may experience greater short‐term reductions in adiposity, whereas girls may benefit more in the long term [32]. A recent study reported no sex differences in prepubertal children over a 3‐month intervention, but greater BMIz reductions in pubertal boys [33]. These findings highlight the influence of developmental stage and intervention duration on gender‐specific outcomes and call for further research on the topic.

Interestingly, full‐time maternal employment was negatively associated with the likelihood of simultaneous parent–child weight improvement. This may be explained by time constraints among full‐time working mothers for health‐related routines such as meal planning, physical activity and family engagement in lifestyle changes [34]. These findings underscore the importance of delivering interventions in more flexible formats—such as digital platforms to better accommodate the schedules of working caregivers and improve programme adherence.

Contrary to some expectations, parental education and age did not significantly predict dyadic weight improvement. This may be due to the relatively narrow range of educational variation within the high‐risk sample. Nonetheless, these null findings highlight the complex and multidimensional nature of behaviour change in vulnerable populations.

Limitations of the study include reliance on weight and BMIz as sole indicators of adiposity and the small number of dyadic improvements. Additionally, the analysis did not account for important potential confounding variables such as family dietary intake, physical activity patterns, parenting practices, screen time or psychosocial stress—factors that may have influenced both parental and child BW/BMIz outcomes. Nevertheless, the study's strengths lie in its large, multi‐centre design, objective anthropometric measurements and focus on dyadic outcomes—an area often underexplored in obesity prevention research.

5. Conclusion

In conclusion, this study contributes to the growing evidence that targeted, family‐based interventions can yield significant benefits, particularly, over longer time frames and in socioeconomically vulnerable populations. Identifying families most likely to benefit from such programmes—and understanding the contextual factors that enable success—can enhance the precision and equity of future public health strategies aimed at reducing the intergenerational transmission of obesity.

Author Contributions

Dimitra‐Irinna Vitoratou conceived the idea for the manuscript, contributed to the statistical analyses and drafted the manuscript. Eva Karaglani and Matzourana Argyropoulou contributed to the literature review, interpretation of findings and critical revision of the manuscript. Konstantina Dimakopoulou contributed to the statistical analyses, interpretation of findings and critical revision of the manuscript. Violeta Iotova, Yuliya Bazdarska, Tsvetalina Tankova, Greet Cardon, Luis A. Moreno, Imre Rurik, Péter Torzsa, Jaana Lindström and Konstantinos Makrilakis contributed to the study design, implementation and data collection. Yannis Manios coordinated the Feel4Diabetes study and supervised the project. All authors read, revised and approved the final manuscript.

Funding

The Feel4Diabetes study received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 643708. The content of this article reflects only the authors' views and the European Community is not liable for any use that may be made of the information contained therein.

Ethics Statement

Researchers in participating countries obtained ethics approval from local authorities.

Consent

Participants were presented with a detailed description of the study and asked to fill in and sign consent forms for their participation and were given the chance to withdraw from the study at any point.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.