記住我
Evidence before this study
We searched PubMed for papers published between Jan 1, 2000, and Aug 1, 2020, using the terms (“obese” OR “overweight” OR “weight status”) AND (“prevalence”) AND (“trend”) AND (“child”). We reviewed the evidence with a focus on early childhood (children younger than 5 years), time trends with end dates after 2010, and high-income countries.
There is high-quality observational evidence from large samples showing stabilised or decreasing trends in early childhood overweight and obesity in high-income countries in the past two decades. However, this trend has followed significant increases before the year 2000. The evidence on how these trends have differed by socioeconomic status is scarce, with some studies showing a widening of socioeconomic disparities. Only a few studies have examined metropolitan regional differences.
Added value of this study
We analysed measurements from more than 1·3 million children aged 1–3·5 years in the state of Victoria (Australia), which were routinely collected during a 15 year period from 2003 to 2017. Data included weight status, age, sex, indicators for area-level socioeconomic status, remoteness, and mother's country of birth. We found that body-mass index Z scores (BMIz) and prevalence of high BMIz decreased during the period of this study in each of the six age–sex groups. When stratified according to socioeconomic status, BMIz significantly decreased across all socioeconomic status levels, although these decreases were more pronounced in higher socioeconomic status groups. When analysed according to location (major cities compared with regional or remote areas), there was a significant difference in BMIz trend, with scores in inner regional and in outer regional or remote areas generally increasing over time, in contrast with falling BMIz in major cities, in all six age–sex analyses. The overall decrease in BMIz across the study population might have been driven by an increasing proportion of children in metropolitan areas with mothers born overseas. Among children of Australian-born mothers, BMIz did not significantly change for most age–sex groups.
Implications of all the available evidence
Results from this study show persistent socioeconomic inequalities in early childhood weight status, and are consistent with previous studies. For the first time, to our knowledge, we showed that residing in a regional area, compared with major cities, was an important factor associated with inequalities in BMIz. Although there is growing evidence internationally for plateauing or decreasing trends in childhood overweight at population level in high-income countries, this study highlights that trends might vary across population subgroups, leading to increased inequalities.
More needs to be done to understand why these trends are different and how children in rural and remote settings can be supported to improve health. Programmes and policies focused on opportunities for improving nutrition, activity, and weight status among children should explicitly consider how they might contribute to, or address, rural–metropolitan inequalities, in addition to the continued focus on reducing health inequalities for those with a socioeconomic disadvantage.
Victoria has a track record of substantial investment in policies and programmes during the past two decades to promote healthy weight among children. These programmes have included numerous community-wide obesity prevention demonstration projects.13Bell AC Simmons A Sanigorski AM Kremer PJ Swinburn BA Preventing childhood obesity: the sentinel site for obesity prevention in Victoria, Australia., 14Waters E Gibbs L Tadic M et al.Cluster randomised trial of a school-community child health promotion and obesity prevention intervention: findings from the evaluation of fun ‘n healthy in Moreland!. The Romp & Chomp project (2004–08) focused on early childhood (5 years or younger), and was among the first to show effective prevention of obesity through community-level interventions for this age group.15de Silva-Sanigorski AM Bell AC Kremer P et al.Reducing obesity in early childhood: results from Romp & Chomp, an Australian community-wide intervention program. The lessons from these projects informed several state-level initiatives, including Kids–Go for Your Life, a setting-based health promotion initiative across Victoria.16Honisett S Woolcock S Porter C Hughes I Developing an award program for children's settings to support healthy eating and physical activity and reduce the risk of overweight and obesity. Comprehensive approaches to preventing unhealthy weight gain were brought together in the Healthy Together Victoria (HTV) initiative, established in 2011, which was implemented state-wide with additional focused investment for 12 communities.17Australian Bureau of StatisticsThis study aimed to identify trends in body-mass index Z score (BMIz) among children aged 1–3·5 years in Victoria (Australia) by age, sex, socioeconomic status, and geographical location.
ResultsElectronic data collected between Jan 1, 2003, and Dec 31, 2017, were available for 48 local government areas, representing approximately 63% of the Victorian population. Overall, 1 329 520 measurements from 675 991 children were included in the analysis (table 1). Slightly more than half (51·5%) were boys. There were more observations for children aged 1 year (n=559 800) than those aged 2 years (n=433 514) or 3·5 years (n=336 206), consistent with higher participation rates in the key age and stage consultations at younger ages, and children from high socioeconomic status areas were over-represented compared with the state-wide distribution.Table 1Overall characteristics, by age group from 2003 to 2017
Data are mean (SD) or n (%). BMIz=body-mass index Z score. SEIFA=Socio-Economic Indexes for Areas. IRSAD=Index of Relative Socio-economic Advantage and Disadvantage. T=tertile.
There were small, but statistically significant, and consistently decreasing trends in BMIz across all six age–sex groups (figure 1; table 2). The same general decreasing pattern was seen in the prevalence of high BMIz across all age groups (proportion of children with BMIz >+1 and BMIz >+2; figure 1; table 3; appendix p 8).Figure 1Mean BMIz and prevalence of high BMIz by age and sex from 2003 to 2017
Table 2Change in mean BMIz from 2003 to 2015, by socioeconomic status and remoteness in each group defined by age and sex
Mean change, CIs, and p values were estimated for each age–sex group with linear models including time as a continuous variable and the effect modifiers—ie, socioeconomic status or remoteness (three levels), or both socioeconomic status and remoteness (two levels), and their interaction. Estimated differences between levels of effect modifiers are shown in the appendix (p 9). BMIz=body-mass index Z score. T=tertile.Table 3Percentage point change in prevalence of high BMIz (>+1) from 2003 to 2017, by socioeconomic status and remoteness in each group defined by age and sex
Trends in prevalence, CIs, and p values for each age–sex group were estimated with generalised linear models (identity link, binary distribution) including time as a continuous variable and the effect modifiers—ie, socioeconomic status or remoteness (three levels), or both socioeconomic status and remoteness (two levels), and their interaction. Estimated differences between levels of effect modifiers are shown in the appendix (p 10). BMIz=body-mass index Z score. T=tertile.BMIz trends decreased for all socioeconomic status tertile levels across all six age–sex groups (table 2; appendix p 15), although not all reductions were statistically significant. There were few significant differences in trends between socioeconomic status tertiles within age–sex groups (table 2; appendix p 9), and the point estimates for the decreasing trends were largest in the high socioeconomic status (most advantaged) tertile in all cases. There were consistent differences in the intercepts by socioeconomic status tertile, such that the children from more advantaged areas had lower mean BMIz (appendix p 15). Similar trends were observed for the prevalence of high BMIz; however, for both boys and girls aged 2 years, these reductions were statistically insignificant (table 3; appendix p 10).Analysis by remoteness showed that across all six age–sex groups, children living in major cities had significantly decreasing mean BMIz and prevalence of BMIz of more than 1 (figure 2; table 2–3). The estimated mean BMIz increased in inner regional areas and outer regional or remote areas for all six age–sex strata, although not all groups were statistically significant. Across all six age–sex strata, trends in BMIz were significantly different between children living in major cities and children in inner regional areas and outer regional or remote areas. There were no significant differences in trends between children living in inner regional areas and those in outer regional or remote areas (data not shown). Similar patterns were observed for prevalence of high BMIz (table 3).Figure 2Mean BMIz by age and sex from 2003 to 2017, stratified by remoteness
When socioeconomic status was analysed within the level of remoteness, the results showed that among children living in major cities, the mean BMIz significantly declined across all three levels of socioeconomic status in all six age–sex strata (table 2; figure 3). Within regional and remote areas, the mean BMIz trends increased for all but one of the 18 subgroups defined by age, sex, and socioeconomic status, although differences were significant only in seven of the subgroups. This analysis also showed that within remoteness, age, and sex strata, there were few differences in BMIz trends across socioeconomic status despite clear cross-sectional differences. By contrast, there were consistent and significant differences in mean BMIz trends between children living in major cities and those in regional areas within every socioeconomic status level (table 2; figure 3; appendix p 9), meaning that, when socioeconomic status tertile remains constant, the reductions are significantly less favourable in regional areas compared with major cities across all six strata. Similar patterns were observed for prevalence of high BMIz (table 3; appendix p 10).Figure 3Mean BMIz by age and sex from 2003 to 2017, stratified by remoteness and tertile of socioeconomic status
The proportion of children with mothers born overseas increased throughout the study period, primarily in major cities (example for children aged 1 year shown in the appendix p 16). When BMIz trends were examined according to mother's country of birth (born in Australia or overseas), the overall decreasing trends in mean BMIz were only apparent among children with mothers born overseas (appendix p 17). In an analysis of the subset of children with mothers born in Australia, the overall trends across age–sex groups were mixed (appendix p 11), with three of the six groups showing no change over time, a decreasing trend in boys aged 1 year, and an increasing trend in boys and girls aged 2 years. However, the differences in trends by remoteness persisted in this subset of children with generally increasing trends observed in regional areas in all age–sex groups, most of which were significantly higher than the unchanged trends over time in major cities for this subset of children (appendix pp 11, 18, 19).DiscussionBy use of repeated, cross-sectional analyses of children aged 1, 2, and 3·5 years in Victoria, this study showed that at a state level, mean BMIz and prevalence of high BMIz decreased between 2003 and 2017. These findings are similar to the trends observed internationally.4Olds T Maher C Zumin S et al.Evidence that the prevalence of childhood overweight is plateauing: data from nine countries., 6Chung A Backholer K Wong E Palermo C Keating C Peeters A Trends in child and adolescent obesity prevalence in economically advanced countries according to socioeconomic position: a systematic review., 7Shackleton N Milne BJ Audas R et al.Improving rates of overweight, obesity and extreme obesity in New Zealand 4-year-old children in 2010–2016.The decreasing trends observed overall and in major cities could be explained by the changing demographic profile of families over time, specifically, an overall increase in the proportion of mothers born overseas, which occurred primarily in city areas. In the past 10 years, an increasing proportion of mothers who have young children in Victoria were born overseas in countries that have a lower population average BMI. At the 2016 census, the largest cohort of mothers born overseas and living in Victoria were from England, India, and China.21Australian Bureau of StatisticsSeparate examination of standardised scores for weight and height or length for age (data not shown) indicated that the observed trends over time in BMIz have been accompanied by decreases in weight for age and height or length for age at differential rates. What could have driven these patterns remains unclear, and no data are available in this study that would provide further insights. This issue warrants further investigation and might explain the mechanisms driving divergent population trends in BMIz, including shifts in the demographic composition of the population over time or changes in child and maternal nutrition.
The large population sample analysed represents a substantial proportion of all children born in Victoria, which was a strength of this study. Data availability was determined by area-level administrative factors (because of timing of the specific electronic database system uptake) and unrelated to the data being examined. The height and weight measures were routinely collected by trained MCH nurses with qualifications in midwifery and child and family health. The size and quality of this comprehensive dataset allowed us for the first time, to our knowledge, to examine trends in smaller Victorian population groups, such as those living in regional areas, with sufficient power to identify small but consistent trends.
Although participation rates in MCH consultations were high, one limitation was that our sample contained an over-representation of children from higher socioeconomic status areas relative to the state-wide distribution. This imbalance might reflect characteristics of the local government areas with available data, and that children from higher socioeconomic status families were more likely to attend scheduled check-ups. This factor should be considered when interpreting point estimates of prevalence or mean BMIz; however, the sample size enabled examination of trends within socioeconomic status tertiles stratified by age, sex, and remoteness, which showed that the observed trends were generally consistent across socioeconomic status levels.
There are also limitations in measures available in the dataset for effect modifiers. The measure of socioeconomic status used in this study represents average socioeconomic characteristics of an area (based on home postcode) and is a proxy for family socioeconomic status. The various components of socioeconomic status that contribute to health and wellbeing include factors such as education, health literacy, neighbourhood environments, and family financial resources. Such differences would be better captured with family-level measures of socioeconomic status. Mother's country of birth is a crude indicator but has been used as a proxy for child ethnicity, which might represent a range of genetic and sociocultural influences on maternal and child health and growth. Furthermore, there are limitations in the use of this variable in this dataset because of the high rate of missing data, which appears to be non-random.
Finally, although height and weight measurements are routine practice for nurses who collected these data, measurements are only taken once at each visit and human errors in measurements and recording are possible, although these are unlikely to systematically influence trends or the interpretation of results.
This study raises important policy-relevant questions about the environments influencing early childhood weight and whether the observed metropolitan–rural differences occur elsewhere. In Australia, average socioeconomic status is lower outside of major cities. However, the difference between major cities and regional areas observed in this study, appears not to be merely a function of general socioeconomic disadvantage, but rather a distinct additional form of disadvantage and health inequality.
Further research is needed to understand why remoteness appears to be a strong driver of increasing BMIz in young children during the study period. The inequality in trends between children living in regional and remote areas and those in metropolitan areas was consistent across age, sex, and socioeconomic status; and persisted even when restricted to children of Australian born mothers. Previous obesity prevention efforts have focused on areas of socioeconomic disadvantage, with good justification, given the persistent inequalities in weight status between children within high and low socioeconomic status. Our analysis suggests that the differences between major cities and regional areas might be stronger drivers of inequalities than socioeconomic status alone, supporting the case for substantial investment in research, policy, and programmes to improve health outcomes outside major cities.
More international evidence, from countries with varying socioeconomic status and demographic characteristics within the non-metropolitan population, would contribute to our understanding of potential mechanisms for the observed results and to the generalisability of our findings to other contexts. Amid an emerging international evidence base that suggests a so-called flattening of trends in overweight and obesity, our results suggest a need to consider a range of potentially widening inequalities, including rurality, ethnicity, and other components of socioeconomic status.
Future research should include strong evaluation of natural experiments and existing large-scale intervention programmes, and the ways they might have affected early childhood environments. This evaluation will require Australia-wide data, detailed geographical analyses to identify areas with differing trajectories, and strong connections with local and state decision makers to ensure that this evidence can inform timely policy and programme responses. Future work should also seek to understand the variation in longitudinal (within each child) growth trajectories across socioeconomic and remoteness areas in Victoria, and the extent to which these have changed over time. A nationally standardised system for collecting and reporting routine early childhood health data would allow Australia to identify and respond to population shifts and inequalities in early childhood weight status in a timely and tailored way.
MN led the design of the study, data collection, data cleaning, and writing of the manuscript, including data interpretation and literature searching, and was responsible for the final decision to submit the manuscript. SA contributed to the design of the study, data collection, and interpretation of results. BS contributed to the design of the study and interpretation of results. LO led the design and did the statistical analysis, preparation of figures, and interpretation of results. MN and LO have accessed and verified all the data in the study. All authors contributed to the preparation of the manuscript and reviewed and approved the submitted version. All authors had full access to all the data in the study and accept responsibility to submit for publication.
留言 (0)