The relationship between parent and child weight is complex as it is a consequence of both shared genetic and environmental factors [10]–[13]. Socio-economic status (SES) is an important determinant of the shared family environment. Numerous studies have demonstrated an association between SES and obesity [14]. SES can influence lifestyle choices and behaviours, area of residence and food affordability, all of which are factors that have been shown to be associated with obesity [15]–[18].

The inverse association between SES and obesity in adults is well established [19]. However, evidence of a relationship between childhood obesity and SES remains equivocal [13], [20]–[23]. Variation in the types and definition of SES indicators used in studies may partly explain this. A review by Shrewsbury and Wardle [20] suggested that the association between child weight and SES is dependent on the type of SES indicator assessed. Parental education appeared to be most consistently associated with childhood obesity [20]. However, evidence of an association between household class and household income with child obesity remained less consistent [20].

In total, 30% of girls were overweight or obese compared with 22% of boys (p=0.000). Within each of the SES indicators, there was an inverse relationship between SES and the prevalence of child overweight and obese. Those ranked lower within each of the socio-economic variables (household income p=0.013, maternal education p=0.000 & household class p=0.000) were significantly more likely to be overweight or obese than those ranked at a higher position. A higher prevalence of overweight and obesity was found among children whose mothers were either overweight or obese compared with children whose fathers were overweight or obese (p=0.000). Overall, 47.2% (95% CI, 45.7%–48.7%) of mothers were normal weight whilst 20.6% (95% CI, 19.4%–21.8%) of fathers were normal weight. Of children from two parent families, only 12% had 2 normal weight parents while 39.2% had at least one obese parent. In total, 11% (95% CI, 8.5%–14.1%) of children with 2 normal weight parents were overweight or obese. This increased to 24.7% (95% CI, 21.8%–28%) when one parent was obese and to 49.2% (95% CI, 43.3%–55.1%) when both parents were obese. Of children from single parent families, 49.2% (95% CI, 45.1%–53.3%) had a normal weight parent and 20% (95% CI, 16.7%–23.9%) had an obese parent. Overall, 18.1% (95% CI, 14.1%–23%) of children from single parent families with a normal weight parent were overweight or obese. This increased to 34.1% (95% CI, 27.7%–41.2%) when the parent was overweight and 41% (95% CI, 32%–50.6%) when the parent was obese.

Table 2 presents the results of the univariate multinomial logistic regression analyses. Univariate regression indicates that female gender, one parent family type, being an only child, lower household class, lower maternal education, lower household income and higher parental BMI (mother’s BMI, father’s BMI and parent weight status) were all associated with a child being in a higher BMI category. Having an overweight parent (within mother’s BMI, father’s BMI and the combined single index variable parent weight status) consistently increased the odds of childhood overweight and obesity. Parent weight status was most strongly associated with childhood overweight and obesity. The univariate regression also indicated that a lower household class and lower maternal education were associated with greater odds of childhood obesity than household income.

Results of the forward stepwise multinomial logistic regression are presented table 3. The social demographic variables, female gender (p=0.000) and one parent family type (p=0.000) were significantly associated with childhood obesity. One parent family type was no longer significantly associated with childhood obesity when the SES indicators were added to the model (model 3: p=0.173). When household income was added to model 3, household income was no longer significantly associated with the risk of a child being in a higher BMI category. However, the association between household class and maternal education with child BMI remained unchanged (when comparing model 3 to model 2). In the fully adjusted model (model 4), female gender, one parent family type, lower household class, lower maternal education and having overweight or obese parents significantly increased the risk of child obesity. Within model 4, children whose mothers were educated to less than a graduate level had at least double the odds of childhood obesity compared with those educated to a graduate level. A lower household class remained significantly associated with child obesity. Although not significant, lower levels of education and a lower household class were associated with a higher risk of childhood overweight. Parent weight status was most significantly associated with childhood overweight and obesity. Children with obese parents were at a significantly increased odds of overweight (OR 3.9, 95% CI, 2.8–5.6) when compared to children with normal weight parents. The odds of childhood obesity were 15.3 (95% CI, 8.4–27.7) when the single parent/both parents were obese. The odds of childhood obesity increased by nearly 3 fold when the single parent/both parents were obese compared to the single parent/both parents being overweight.

Only 18.9% of children were from families (either single parent or two parent families) with normal weight parents. Having normal weight parents appears to have a protective effect against the risk of childhood obesity. Only 14.4% of children from such families were overweight or obese whereas 46.2% of children with obese parents were overweight or obese. After adjustment for household socio-economic characteristics, children from obese parent families remained at greater than 15 (95% CI, 8.44–27.65) times the odds of obesity when compared to children from families with normal weight parents. This suggests that SES alone cannot explain the association between parent obesity and child obesity. SES indicators appear to only capture some shared familial environmental factors which result in weight gain. We suggest that these results highlight that the shared family environment is a multi-dimensional contributor to the obesity epidemic with both genetic and environmental origins.

Within this present study, children who were more deprived were at a higher risk of overweight and obesity, which is similar to results found in adults [19]. Children from one parent families were found to be at significantly higher odds of overweight and obesity than children from two parent families. Some research suggests that one parent families may have greater levels of social deprivation and this may play a role in explaining this [29]. However, our results indicate that parental weight was more predictive overweight and obesity in children from single parent families than SES. There was an inverse association between household class and maternal education with childhood obesity. The association between household class and childhood obesity was more graded. Within the final adjusted model, children from a lower household class were at higher odds of obesity than children with lesser educated mothers. Research indicates that parental education is the SES indicator most consistently associated with childhood obesity [20], [21]. This may be because maternal education is a more stable indicator of SES over time than household income or household class. Maternal education is likely to influence factors including literacy as well as knowledge of healthy versus unhealthy behaviours which impact on weight status [14], [30]. As a higher level of education appears protective against child obesity, this suggests that education may be crucial in tackling the obesity epidemic. Overall, variations in odds of obesity by each indicator of SES suggest that household class, household income and maternal education may all influence different behaviours and choices that impact weight gain. Further research is required to fully understand how each SES characteristic predicts behaviours which result in weight gain. In addition, efforts are necessary to standardise SES indicators and definitions used across studies.

In this study SES indicators do not explain all the association between parent and child weight. Therefore, other causal pathways for childhood obesity need to be considered. Research from other studies of childhood obesity indicate that the weight status of parents from 2 parent families may interact [4], [31]. Mechanisms resulting in a positive energy balance in both parents appear to be more predictive of childhood obesity than such mechanisms in one parent. In this current research having 2 obese parents compared with one obese parent resulted in a 2 fold increase in the odds of childhood obesity.

A study by Wardle et al. [32] compared food, physical activity and lifestyle patterns in children from lean and obese families. This study found that children from obese families had higher preferences for fatty foods and sedentary activities and a lower preference for fruit and vegetable consumption. Such food and physical activity patterns may have a negative impact on energy balance resulting in weight gain. Such diet and activity patterns may potentially explain the lack of significance for household income in this present study. Parent weight status may be a better predictor of food types purchased rather than income or other measures of household SES. More affluent families with obese parents may have a preference for energy dense food regardless of income available to spend on good quality foods. Grunert at al. [33] suggest that habitual behaviour is difficult to change even if an individual is aware of the negative consequences of their behaviours. Grunert et al. suggest that obese individuals have a greater response to external cues (sight, smell) for food intake whilst normal weight individuals respond to internal cues (hungry). Children may acquire habitual behaviours and responses to dietary and physical activity patterns from that of their parents. Another possible explanation is that genotypes including the FTO gene which impacts appetite may influence control over food intake and choices resulting in children from obese families having a greater predisposition for obesity [34]–[37].

Similar to other findings [38], [39], maternal obesity was more predictive of a child being in a higher BMI category than paternal obesity. There are a number of possible explanations for this. Mothers were nominated as the primary caregiver (the person who spent most time with the study child) for 98% of children who took part in this study. This indicates that children spend more time in their mother’s environment and thus may acquire more behaviour’s from their mother. A study by Hannon et al. [40] found that the eating habits of the family food preparer, 84% of whom were mothers, predicted the eating habits of their child. Birth factors including the role of the intra-uterine environment on subsequent risk of childhood obesity is a second possible explanation [41], [42].