Main findings of this study
This current study found a strong association between child BMI z-score at Year 1 and Year 4. Mean BMI z-score increased between these time points for both boys and girls. Children from more deprived areas had a higher BMI z-score than those from less deprived areas. Increased parental BMI was associated with increased child BMI z-score in both Year 1 and Year 4, though this effect was small. Further exploratory analyses demonstrated that this association was observed between both maternal BMI and child BMI z-score and paternal BMI and child BMI z-score. Finally, regression analyses showed that there was no statistically significant association between change in parental BMI between Year 1 and Year 4 and Year 4 child BMI z-score.
Implications of findings in context of existing literature
It has not been possible to conduct analyses tracking a child’s weight status from the national NCMP database yet, as up until 2013, all local data was anonymised prior to being uploaded which has prevented data linkage [
30]. Although the Health Survey for England provides data on childhood obesity prevalence, different children are selected each year thereby preventing tracking analyses [
31]. Studies using local NCMP data (with participant identifiable data) have shown a rise in child BMI z-score [
6] and the prevalence of obesity [
32] between Reception and Year 6. Our study uses longitudinal data and is novel in providing insight into the changes occurring between NCMP measurements and determines whether child BMI z-score is already increasing by Year 4 or whether this rise takes place after Year 4.
Parental obesity is a known risk factor for childhood obesity [
9‐
13]. The literature is less clear regarding the significance of parental gender on offspring obesity risk. Some studies suggest maternal obesity to be of greater importance to offspring (of either gender) than paternal obesity [
14,
15] whereas Freeman et al. reported that having an overweight or obese father, but a healthy weight mother, increased the odds of a child becoming obese, but not the reverse scenario (i.e. having an overweight or obese mother with a healthy weight father was not a significant predictor of childhood obesity) [
16]. More recently, Aris et al. have shown paternal overweight status to be a significant risk factor for childhood obesity [
17]. Some studies report gender assortative relationships between parental and offspring weight status (i.e. mother-daughter and father-son associations in BMI / BMI z-score) [
18] whereas other studies have found minimal or no compelling evidence of these associations [
13,
33]. The Health Survey for England 2017 found child BMI status to be independently associated with both their mother’s and father’s BMI status [
34]. Similarly, our study found both paternal and maternal BMI to be important predictors of a child BMI z-score in both Year 1 and Year 4 of Primary School. Higher IMD scores were also associated with higher child BMI z-scores, supporting previous findings in the literature that social deprivation increases obesity risk [
34‐
36].
The current study observed how parental BMI change between Year 1 and Year 4 is associated with a child’s BMI z-score in Year 4. Whilst there are numerous studies reporting on correlations between parental weight loss and child BMI z-score reductions in family-based treatments for childhood obesity [
20,
21], there are fewer studies simply observing the effect of parental BMI change over time on a child’s BMI z-score when no intervention is provided. This is important as it may provide justification towards allocating scarce public health funding [
37] to ‘parent-based programmes’ targeted at preventing parental weight gain (even where the child is normal weight) and thereby child weight gain. This in turn may generate a viable alternative cost-effective strategy to parent-only or parent-child interventions aimed specifically at overweight or obese children. There is already emerging evidence that parent-only interventions (where only the parent attends the intervention) may be as effective [
22,
23,
38] as the current gold-standard family-based interventions (where both parent and child attends) [
19] for the treatment of childhood overweight/obesity,. However, there is less extensive literature looking at parent-only interventions (focused specifically on preventing parental weight gain) for the prevention and treatment of childhood obesity.
Our study showed that there was no significant association between change in parental BMI between Year 1 and Year 4 and Year 4 child BMI z-score. These findings need to be interpreted with caution as the sample size was relatively small, and missing data prevented analysis of all participants and their parents taking part in the B-ProAct1v study. An Indonesian observational study reported a father’s weight gain or weight loss to be correlated to their daughters’ weight but not their sons, whereas a mother’s weight gain or loss was correlated to both the weight of sons and daughters [
39]. Further sufficiently powered observational studies are needed to explore this.
Limitations of this study
The Year 1, B-ProAct1v child weight data (measured in 2012/2013), showed a slightly lower percentage of children classified as overweight (12%) or obese (8%) compared to Bristol NCMP data from Reception (measured in 2011/2012, i.e. the children who would be moving into Year 1 in 2012/2013) where there were 14% classified as overweight and 10% classified as obese. It is unlikely that the prevalence of obesity reduces between Reception and Year 1 and so this discrepancy may indicate consent bias in the B-ProAct1v project (i.e. overweight or obese children and their parents may have been less likely to agree to participate in the study than their normal weight counterparts).
For recruitment to the B-ProAct1v study, a child and at least one parent (designated as parent 1) needed to consent to participate in data collection. This led to the main analyses in this current study focusing on child BMI z-score and Parent 1 BMI data and limited analyses of child BMI z-score with gender of parent or child. Parent 1 was usually female, so this also limited the power of the study to detect associations between maternal and paternal BMI and child BMI z-score as the number of male parents designated as Parent 1 was considerably smaller than the number of female parents. There was limited BMI data available for the same parent 1 in Year 1 and Year 4 which restricted analyses investigating how parental change between Year 1 and Year 4 influences Year 4 child BMI z-score to a relatively small sample of Year 4 children. Due to the limited data, it was also not possible to undertake separate analyses according to parental gender or child gender.
There was no assessment of pubertal stage for children participating in the study, though this is unlikely to impact the results given that few children would be in puberty in Year 4 of school. Finally, parental BMI was calculated from self-reported height and weight data. Self-report data has previously been shown to slightly overestimate height and underestimate weight [
40,
41]. In this study, this could result in underestimation of the observed associations. However, previous large validation studies have supported the accuracy of self-reported adult height and weight measurements to calculate BMI [
42,
43]. Children’s height and weight were measured at their primary school by B-ProAct1v fieldworkers.