Skip to main content

Table 2 Factors mediating the association between socioeconomic position and adiposity in youth in Ireland and the UK

From: Mediators of socioeconomic differences in overweight and obesity among youth in Ireland and the UK (2011–2021): a systematic review

Study Mediated relationship (direction of the association) Method used to Assess Mediation [name of model used] Mediation Results*
Cetateanu & Jones [46] Association btw deprivation and
(a) OB ( +) and (b) OW/OB ( +) for:
(1) 4–5 year olds (-)
(2) 10–11 year olds (-)
Preacher and Hayes indirect effect method (1) No mediating effect in the 4–5 year old group
(2) For the older cohort, availability of fast food outlets and other types of unhealthy food outlets partially mediated the association btw deprivation and OB and OW/OB by between 1 and 2%. No mediation was found for the availability of mixed food outlets
Goisis et al. [39] Association btw family income and risk of:
(1) OB at age 5 (-)
(2) OW at age 11 (-)
(3) OB at age 11 (0)
(4) Upward movement across weight categories from age 5 to age 11 (-)
Assessment of attenuation/reduction of regression coefficients upon inclusion of mediators (1 and 4) Physical activity, TV use, bedtime, fruit intake, sweet drink intake and maternal BMI skipping breakfast did most to attenuate inequalities. Other factors including maternal smoking during pregnancy, breastfeeding duration and time of weaning also played a role in mediation
(2 and 3) Fruit, sweet drink, and breakfast intake did most to attenuate inequalities, with other factors (see 1 and 4) playing a smaller role
Goisis et al. [31] Association btw family income and OW/OB Logistic regression models Poorer White children are at higher risk of OW/OB than higher-income White children (RRR 1.13; 95% CI: 1.02 to 1.25). This SEP differential is reversed for children from Black Caribbean/African backgrounds and non-existent for Indian and Pakistani/Bangladeshi backgrounds. In contrast to White children, lower income children from all other ethnic backgrounds are less likely to be OW/OB at age 7 than their more advantaged counterparts
Laverty et al. [40] Association btw household income group and occupational social class with: (1) BMI (2) % BF Longitudinal (panel) regression models (1) Switching to active travel was associated with a − 0.32 kg/m2 BMI (95% CI − 0.58 to − 0.06) among those in the lowest household income group compared with a -0.11 kg/m2 among the highest group (-0.24 to 0.03)
(2) Switching to active travel was associated with a − 0.71% BF (95% CI − 1.47% to 0.05%) among the lowest household income group compared with a -0.55% BF (-1.01 to -0.09%) among those in the highest income group
Layte et al. [17] Association btw social class (baseline professional class) and: (1) rapid growth from birth to 9 months (2) rapid grow from 9 months to 3 years (3) rapid OB at 3 years Assessment of attenuation/reduction of regression coefficients upon inclusion of mediators (1) Breastfeeding and age at weaning most important for non-manual class. Antenatal smoking and alcohol consumption most important for manual and unclassified classes. The model with all mediators reduced coefficients by an average of 76%
(2) Child diet, TV viewing and maternal BMI led to highest reductions in all classes. Lower maternal BMI and lower levels of TV viewing mediated lower odds of rapid weight gain
(3) Child diet, TV viewing and maternal BMI led to highest reductions in coefficients in all classes. All mediator groups had some contribution
Lu et al. [43] (1) Association btw poverty and higher BMI in children (2) Association btw maternal education and higher BMI in children Mixed-effects fractional polynomial and multinomial regression modelling (1) Poverty was associated with higher BMI in children of White and South Asian origins, widening with age to 0.75 kg/m2 (95% CI, 0.59–0.91) and 0.77 kg/m2 (0.26–1.27) at 14 years for the White and South Asian groups, respectively. A reverse income-BMI association in children of Black (African-Caribbean) origin was found with the poverty group having a lower BMI (− 0.37 kg/m2 [− 0.71 to -0.04] at 5 years and − 0.95 kg/m2 [− 1.79 to − 0.11] at 14 years
(2) Similar patterns (see (1)) presented with maternal education and obesity at 14 years
Martinson et al. [37] Association btw SEP and child OW Multivariate logistic regression models Low SEP children with non-White native- and foreign-born mothers are at lower risk of OW. Low SEP children with white immigrant mothers are at an increased risk of OW
Massion et al. [25] Association btw maternal education and childhood OW at age 11 Assessment of attenuation/reduction of RR on inclusion of mediators (Barron and Kenny) Early life risk factors (maternal pre-pregnancy OW, maternal smoking during pregnancy) reduced the RR from 1.72 (95%CI 1.48 to 2.01) to 1.47 (1.26–1.71)
Mireku & Rodriguez [45] Association btw deprivation and
(a) OW, OB and (b) %BF
Linear regression models and log-binomial models When stratified by geographic-level deprivation, the family income gradient in the risk of OB for moderately affluent (2nd, 3rd or 4th quintile deprivation) neighbourhoods was non-significant. However, family income gradient in the risk of OW/OB persisted for the most (RR 5.5 (95% CI 1.0–17.2, p < 0.05) and least (2.4 (1.0–5.8, p < 0.05) deprived quintiles of geographic-level deprivation
Noonan et al. [41] Association btw area deprivation and child BMI and waist circumference Linear regression analyses A significant inverse association was seen between neighbourhood aesthetics and high deprivation group's BMI (β =  − 0.29, p <  = 0.01) and waist circumferences (β =  − 0.27, p <  = 0.01)
Noonan [28] Association btw poverty and childhood OW/OB Adjusted logistic and multinomial logistic regression analyses Adolescents living in poverty compared to those not living in poverty reported more frequent consumption of sweetened drinks and fast food, and less frequent consumption of fruits and vegetables (OR = 1.92–3.61; p < 0.001). Difference in weight status and dietary intake outcomes for girls in poverty were greater (OR = 1.55–3.62; p < 0.001) compared to boys (OR = 1.39–3.60; p < 0.001)
Noonan & Fairclough [30] (1) Association btw individual-level SEP (maternal education) and childhood OW/OB
(2) Association btw area-level SEP and childhood OW/OB
Adjusted linear and multinomial logistic regression analysis Individual-level and area-level SEP were independently related to OW/OB. Higher rates of OW/OB among deprived children were not due to physical inactivity
Oude Groeniger et al. [27] Association between maternal education and childhood OB Causal mediation analysis At age 14, between 13 and 18% of relative inequalities in childhood OB were reduced if maternal education differences in screen media exposure at age 7 and 11 were eliminated
Parkes et al. [34] Association btw maternal education and child BMI trajectory slope (-) Path analysis Indirect effect of SEP via all mediators (0.16) comprised 89% of the total effect of SEP level on BMI outcome. Pathway to BMI slope from maternal education through parenting (informal meal setting) and then unhealthy diet accounted for 68% of the significant indirect pathways. The main indirect pathway involving parenting was via the effect of child bedroom TV, with smaller effects of informal meal setting and less positive mealtime interaction. An effect of unhealthy diet (which in turn affected BMI) which did not got through parenting was also seen
Samani-Radia & McCarthy [29] Association btw geographical location deprivation, family income and childhood OB and % BF Chi-square tests Children from lower incomes were significantly shorter, heavier and higher % BF, with a higher BMI for their age compared with those from a higher income background. A shorter height-for-age of the ‘lower income’ group children may contribute to the income group divisions
Schalkwijk [38] Association btw parental education, family income and childhood OW/OB Logistic regression models Among low SEP households, lack of garden access and less green space was associated with OW/OB; among higher SEP, poor neighbourhood condition influenced the probability of OW and OB: OR, 95% CI 1.38 (1.12–1.70), 1.38 (1.21–1.70) respectively
Silverwood et al. [36] Association btw maternal education, area deprivation and weekly household income with OW/OB Traditional (Baron & Kenny) and counterfactual-based mediation analyses (bootstrapping to assess significance) Higher BW in low SEP is associated with increased inequalities in OW
Straatmann et al. [26] Association btw maternal education and OW/OB at age 14 Counterfactual mediation analyses For OW/OB, 19% of the total effect of socioeconomic conditions was mediated through all ACEs investigated
Strugnell et al. [44] Association btw income deprivation and childhood OW/OB
(1) 4–5 year olds
(2) 10–11 year olds
Multivariable logistic regression models (1 and 2) Ethnicity has an independent influence on OW/OB for both groups (4–5 and 10–11 year olds), with the distribution between most and least advantaged widening for most ethnic groups between ages 4–5 and 10–11 years
(2) For the 10–11 year olds, SEP differentials were found to differ by sex and by ethnicity with the largest disparity reported for White children, and the smallest seen in Black African children. Comparing boys in the least deprived and most deprived groups, the difference was 12% among White British boys and 18% for Any Other White Backgrounds, compared with 11% for Indian boys, 5% for Pakistani boys and 2% for Black African boys
Stuart & Panico [35] Association btw parental income, parental education and a persistent poverty indicator with
(1) OW
(2) OB
Multinomial logistic regression models High BW (RRR: 2.16, p < 0.05), not being breastfed (RRR: 1.33, p < 0.05) and mother smoking during pregnancy (RRR: 1.96, p < 0.001) mediated some educational gradient (but not income gradient) between the OB and normal weight group. SEP does not uniformly impact BMI trajectories, and different indicators of disadvantage capture different trajectories. For SEP inequalities, the OW group was mostly characterized by low parental income, whereas the OB group was mostly characterized by low parental education
Townsend et al. [32] Association btw area deprivation and BMI Cross-classified multilevel regression models Longer time spent in school with a high percentage of children receiving FSM (poorer schools) affected the association found between BMI and low SEP. Deprivation explains a greater proportion of the variance in BMI for older compared with younger children, perhaps reflecting the impact of deprivation as children age, highlighting the widening of health inequalities through childhood
Walsh & Cullinan [42] (1) Association btw household income and childhood OB and OW/OB
(2) Inequality in OW/OB (based on concentration indices)
Prediction of inequality gradient using regression Parental occupation and education contributed to OB (41.16%) and OW/OB (44.18%) inequalities; parental health (maternal BMI and maternal smoking during pregnancy) contributed OB (3.7%) and OW/OB (84.1%) inequalities. Child variables had a low impact on observed inequalities—mainly via TV viewings and bedroom TV
Wijlaars et al. [33] (1) Association btw parental occupation (NS-SEC Index) and 3-month weight (-), weight gain btw birth and 3 months (-) and rapid weight gain (-)
(2) Association btw maternal education and 3-month weight (-), weight gain between birth and 3 months (-) and rapid weight gain (-)
Assessment of attenuation/reduction of regression coefficients upon inclusion of mediators (bootstrapping to assess significance) (1) Breastfeeding duration attenuated the association btw parental occupation and: 3-month weight by 68%; weight gain by 62%; and odds of rapid growth by 53%
(2) Breastfeeding duration attenuated the association btw maternal education and: 3-month weight by 88%; weight gain by 82% and odds of rapid growth by 64%. No mediating effect was found for smoking during pregnancy, maternal and paternal BMI
  1. Abbreviations: ACE Adverse Childhood Event, BF Body fat, BMI Body Mass Index, BW Birthweight, Btw between, CI Confidence Interval, FSM Free school meals, NS-SEC Index National Statistics Socioeconomic Class index (UK), OB Obesity, OR Odds Ratio, OW Overweight, RR Relative Risk, RRR Relative Risk Ratio, SEP Socioeconomic position, TV Television