The issue of childhood obesity has emerged as a serious global epidemic of the 21st century [1
]. Recent trends show stabilization of obesity prevalence throughout the population in many nations [2
], including Ireland. However, specific strategies are necessary to reverse the tide towards healthy living [1
The World Health Organization (WHO) reported that the prevalence of overweight and obesity combined for children has increased from 4% to over 18% between 1975 and 2016 [9
]. It is estimated that 41 million children under the age of five and over 340 million between the ages of 5–19 are either overweight or obese [9
]. Ireland is ranked among the countries with high rates of childhood obesity [10
]. Current estimates show that about 7% of girls and 6% of boys aged 4–16 are obese in Ireland, putting the country at 58 out of 200 countries in the childhood obesity charts [11
]. Interestingly, almost half of the children in Irish households are exposed to secondhand smoke (SHS) [12
]. Taken together, both childhood obesity and childhood SHS exposure are important public health challenges for the Irish government.
Childhood obesity is associated with adverse health outcomes throughout the life course, including an increased risk for type 2 diabetes, heart diseases, cancer, as well as lifelong overweight and obesity [13
]. Therefore, governmental strategies, for instance, the Healthy Ireland Framework, have adopted a life-course perspective to chronic diseases associated with obesity and other lifestyle factors, such as tobacco smoke exposure [14
]. In 2005, the Bogalusa Heart Study showed that obese children between the ages of 6–13 are ten times as likely to become obese adults as those who are underweight or have lower Body Mass Index (BMI) [15
]. Worryingly, children living today could be the first generation to live shorter, less healthy lives than their parents [16
Obesity is a complex system, and requires a systems thinking approach to tackle childhood obesity [17
]. Clearly, it is evident that effective intervention strategies are necessary to address the modifiable risk factors contributing to childhood obesity [18
]. According to the WHO, the increase in childhood obesity can be largely attributed to the changing nature of the environment [19
]. Consequently, childhood obesity can only be successfully tackled if we focus on both the child and the child’s prevailing environment [13
]. One such environment is the setting where a child spends most of the time, namely, in a household or in a carers’ service. A 2006 survey by the Central Statistics Office in Ireland found that 60% of infants are looked after by a parent/guardian closely followed by 12% looked after by paid carers’ and 11.5 unpaid relatives [20
]. However, 98% of the carers’ in our study are biological mothers. Berman and colleagues [21
] also found that children who spend more time at home, in the presence of a smoker or living with a carer who smokes have an increased SHS exposure.
Furthermore, SHS exposure has been identified as one of such environmental hazards that are detrimental to the health of the pediatric population [22
]. Despite a decrease in the prevalence of smoking [24
], 40% of children aged 3–11 years had SHS exposure at home, and a large proportion of this exposure is recorded in Europe [26
]. A growing body of evidence indicates that there is a link between maternal smoking during pregnancy (prenatal) and childhood obesity [18
]. The mechanisms linking this association has been attributed to the presence of psychological mechanism and residual confounding [18
]. It has also been hypothesized that inhaling the chemicals in tobacco smoke may cause impaired metabolic and immune functions leading to an increase in the child’s susceptibility to obesity [30
However, there is limited evidence suggesting a link between post-natal or childhood SHS exposure and childhood obesity [32
]. Using the Danish Birth Cohort, Moller et al. did not find a statistical association as the group with exposure to smoking only postnatally was small (n
= 140). Furthermore, children being exposed to SHS during their childhood irrespective of their pre-natal or early post-natal SHS exposure has not been well-researched in settings where both childhood obesity and childhood SHS exposure within households are high [32
]. Ireland is one such unique population setting—one of the highest rates of childhood obesity [11
], and also half of the children being exposed to SHS in households [12
]. Moreover, the Irish national health strategy has set out targets and goals to tackle chronic diseases associated with lifestyle factors (obesity and tobacco smoke) through a life-course perspective [14
]. Therefore, this study is timely.
In summary, the carers’ in our study are 98% biological mothers, and it is probable that childhood SHS exposure would primarily be due to maternal smoking. Therefore, the current study has a two-fold rationale; (1) to examine the impact of childhood SHS exposure on childhood obesity risk in a private setting to provide further evidence in support of a healthy living environment, as actioned in the Irish Healthy Ireland strategy; (2) to reproduce similar findings in Ireland that were previously reported in other comparable population settings, using a nationally representative population-based cohort while controlling for potential confounders available to the dataset.
Given that 98% of primary carers’ in this study were biological mothers, this large cohort study in Ireland showed increased odds of obesity/overweight in children at both age three and five years, if exposed to maternal smoking in early childhood compared with children of non-smoking mothers. Our study findings also suggest that the risk of childhood overweight/obesity following childhood SHS exposure was independent of both low birth-weight and breastfeeding. Both these findings are in agreement with previous studies in Denmark and in Germany [32
]. However, the Danish Birth Cohort study did not show a statistical association as the group with exposure to smoking only postnatally was small (n
= 140), and also focused on both prenatal and early post-natal. Our exposed group was larger than the Danish cohort, and, thus, had more statistical power, and also showed that such an effect could continue for a longer period post-nasally through early childhood. Raum et al. also found a positive association between exposure to maternal smoking in the child’s first year and childhood overweight at age six [44
]. In short, our study findings are consistent with previous evidence, and were reproducible in an Irish setting, which has local policy implications.
Considering that the carers in this study were overwhelmingly biological mothers, we can be prudent in stating that the effects, thus observed can broadly be associated with parental smoking. However, in settings where the primary carers are not overwhelmingly biological mothers, there may be variations in the effect estimates because of potentially different underlying mechanistic pathways. Therefore, it is clearly important that similar studies are being undertaken where the carers’ profile varies. However, a growing number of studies have examined the impact of parental smoking postnatally on overweight/obesity in childhood across different population settings, and our observations are in agreement with the majority of these studies [28
]. For instance, childhood SHS exposure had an increased BMI in these children compared with children who were not exposed to SHS from parents during early childhood [31
]. Another study also found a dose-dependent association between exposure to SHS and obesity [47
]. A systematic review further validated the relationship between SHS exposure during childhood and increased BMI [18
]. The prospective PIAMA Study in the Netherlands showed that ‘smoking in the parental house” was a significant independent predictor of childhood overweight rather than maternal smoking during pregnancy [48
Not surprisingly, our study failed to show a significant association between paternal smoking and childhood overweight/obesity because of fewer numbers. However, hour study findings are consistent with the Nurses’ Health Study II, where an association was observed for maternal smoking and childhood obesity in the daughter, but not for paternal smoking [49
]. Similarly, in the generation R study, in comparison with non-smoking mothers, children whose mothers smoked during pregnancy had increased risk of overweight and obesity at four years old [50
]. However, no association was observed for paternal smoking and childhood obesity among non-smoking mothers [50
]. Furthermore, the overall significant, but relatively low estimates observed in this study could be attributed to the low prevalence of smoking among carers’ which were much smaller compared to other studies (an average of 24.7% among primary carers’ across all three waves). This can be explained in terms of the increase in smoking cessation campaigns, as well as the comprehensive smoke-free policy in Ireland since 2004 [51
]. In addition, parents and carers are more aware of the detrimental effects of exposure to tobacco smoke on children, who are the captive audience.
To the best of our knowledge, this study is among the first in Ireland to examine the impact of SHS exposure during early childhood through carers’ smoking status (mostly biological mothers smoking postnatally) on childhood obesity using prospectively collected information from a large nationally representative cohort of Irish children. This makes the study generalizable to other comparable population settings. The use of a prospective cohort study minimized the possibility of recall bias from carers’ report of smoking. Additionally, weight and height measurements in the first three waves were taken by trained interviewers, thereby reducing the possibility of recall and interviewer biases (sub-types of measurement bias). Furthermore, important covariates that are not often available in other studies, such as dietary patterns of the children were collected in the GUI Cohort study. We also showed that low birth weight and breastfeeding as individual covariates did not influence the association between childhood SHS exposure and the development of childhood obesity. In other words, low birth weight and breastfeeding can act both as an effect modifier, and a mediator of the association studied. Moreover, the influence may depend on the timing of smoking while breastfeeding and whether the mother inhales nicotine or not. However, for carers who are not biological mothers, the underlying mechanism may be far more complex.
A primary limitation of this study is that the exposure was based on self-report and no biomarker, such as cotinine measurement was used to verify the exposure. Although some studies have validated the accuracy of self-reported smoking in large populations as an indicator of current smoking behavior [52
], there is still a possibility of under-reporting of smoking status, due to social desirability. This may have led to non-differential misclassification bias, thereby attenuating observed estimates. Similarly, this study did not adjust for exposure from other household smokers. Secondly, this study did not control for maternal smoking during pregnancy and other prenatal exposures because the information was not available, introducing residual confounding. However, studies have shown increased smoking cessation during pregnancy, due to smoking cessation campaigns [55
], which has encouraged mothers to quit during pregnancy. Additionally, low birth weight was used instead as a proxy measure for maternal smoking [42
]. A similar limitation is that environmental smoke exposure by household smokers and smoke exposure outside the home (e.g., in cars) was not considered as a source of SHS exposure in this study. Nevertheless, the estimates in this study are conservative.
Although this study set out to examine primary and secondary carers smoking status (this includes biological parents, adoptive parents, caregivers, non-resident parents, grandparents, relatives and unrelated guardian where applicable, of the study child), in the GUI data, 98% of respondents were the biological mother/primary carer), therefore only information on the primary carer was presented. In other words, the study exposure is primarily, due to maternal smoking post-natally.
Additionally, the observed associations might be as a result of residual confounding by unmeasured confounders. For example, this study did not measure factors, such as physical activity which is strongly correlated with childhood obesity [49
]. This is an important limitation of this study also identified in previous studies.
The loss to follow up bias (<5%) is another inevitable limitation of a cohort study that was present in our study. Finally, the possibility of either residual or unmeasured confounding, due to lifestyle, genetic or familial factors [18
] and chance should not be ruled out as this may have influenced the findings.
4.3. Future Recommendations and Policy Implications
Future studies should use a reliable biomarker, such as cotinine levels in hair or urine for objective measurement of second-hand smoking, and also capture trajectories across the life-course, with better study designs. Similarly, fat distribution or visceral obesity should be used to determine body fat/obesity in children to account for differences that exist between BMI and body fat distribution, employing objective measurements [56
Nevertheless, our study findings are crucial for supporting the existing body of evidence. The study findings also inform policymakers, government and other stakeholders to refocus on cost-effective strategies (anti-smoking interventions and smoking cessation programs) to tackle childhood obesity in Ireland. Such rethinking among the policymakers will certainly contribute further to the design and evaluation of appropriate health policies and community-based childhood obesity prevention programs in Ireland.