Article Text

Growing up before growing out: secular trends in height, weight and obesity in 5–6-year-old children born between 1970 and 2006
  1. Sarah M Smith1,
  2. Leone C A Craig2,
  3. Amalraj E Raja3,
  4. Geraldine McNeill2,
  5. Stephen W Turner1
  1. 1Department of Child Health, University of Aberdeen, Royal Aberdeen Children's Hospital, Aberdeen, UK
  2. 2Public Health Nutrition Research Group, University of Aberdeen, Aberdeen, UK
  3. 3Department of Medical Statistics, University of Aberdeen, Aberdeen, UK
  1. Correspondence to Dr Stephen W Turner, Department of Child Health, University of Aberdeen, Royal Aberdeen Children's Hospital, Aberdeen AB25 2ZD, UK; s.w.turner{at}abdn.ac.uk

Abstract

Background This was a whole-population study of height, weight and obesity prevalence in 5–6-year-old children born between 1970 and 2006 in the Grampian region, north east Scotland.

Methods Heights and weights collected as part of routine primary school medical entry were obtained from different sources. Obesity was defined as body mass index (BMI)≥98th centile.

Results Anthropometric measurements were made in 194391 children, mean age 5.6 years (SD 0.8). The mean height z scores rose for those born between 1970-2000 respectively and were static thereafter. Obesity prevalence was non-linear over time and initially fell for birth years 1970–1977, rising between 1977 and 1998 before falling for those born between 1998 and 2006. For the whole population, the prevalence of obesity rose from 1.3% for those born in 1976 to 6.9% for those born in 1998 and fell back to 5.7% for children born in 2006. Obesity was initially highest in girls and most affluent communities but became most prevalent among boys and least affluent communities.

Conclusions The secular increase in height at school entry in children born in 1970 and afterwards was followed by an increase in weight leading to an initial reduction in obesity prevalence. Whole-population obesity prevalence for children born in the 2000s is now falling but prevalence remains considerably higher compared with those born in the 1970s. Obesity prevalence continues to rise among less affluent communities.

  • Comm Child Health
  • Growth
View Full Text

Statistics from Altmetric.com

What is already known on this topic

  • Childhood obesity is a public health concern.

  • Obesity prevalence is usually assessed on small proportions of the population often sampled at erratic intervals.

What this study adds

  • We observed an increase in height preceding one in weight among children born in the 1970s leading to an initial decline in obesity prevalence.

  • There was a non-linear change in the prevalence of obesity for children born between 1970 and 2006.

  • The initial association between increased obesity and affluence and female gender was reversed between 1970 and 2006.

Introduction

Childhood obesity prevalence in the UK rose throughout the 1990s to reach epidemic proportions.1 Surveys from across Britain found that obesity prevalence in childhood rose from approximately 1% to between 4% and 6%.2–5 The rise in childhood obesity is of concern as childhood obesity may persist into adulthood6 and may be associated with type II diabetes,7 cancer8 and cardiovascular disease.9

In England, evidence has emerged of a plateau in prevalence of childhood obesity being reached in the 2000s10 and perhaps even a modest fall.11 ,12 For Scotland, there is also evidence that the rise in obesity and overweight prevalence among girls came to an end during the 2000s13 but not for boys. While obesity prevalence may remain static for the whole population, there is evidence to suggest that children from poorer communities may continue to experience increasing obesity prevalence3 ,10 ,14 and that obesity prevalence may be rising in boys compared with girls.10 ,13 Despite the inclusion of data from many thousands of children, the current understanding of obesity trends in the UK is limited to surveys carried out at intervals where samples of the population are measured and which may lack power for subgroup analysis. A study of year-on-year trends in childhood obesity for a whole population measured over the medium–long term would give important insight into patterns of obesity prevalence in children.

Here, we present the results from a whole-population study of height and weight measurements in 5–6-year-old children living in north east Scotland who were born between 1970 and 2006. Our aim was to describe trends in height, weight, body mass index (BMI) and overweight/obesity for the whole population from 1975 to the present day and to explore the differences between subgroups categorised by gender and socioeonomic deprivation.

Methods

Study design

This was an observational whole-population study of trends in height, weight, BMI and overweight/obesity in children. Data had been routinely collected by school nurses from children in their first year at primary school.

Data collected

For all individuals, the following data fields were sought: decimal age at measurement, gender, whether height and weight were recorded, height (cm), weight (kg), school attended, year of birth and year of measurement. Height and weight measurements were converted from imperial to metric units if necessary. Quality control ensured that anomalous results were deleted, that is, wrong measurements entered (eg, height=23.3 kg, weight=115 cm) or measurements wrongly entered (eg, height=11.5 cm, weight=233 kg). Anonymity was maintained and the study was approved by the NHS Grampian Caldicott guardian. Ethnicity was not recorded, but the proportion of the population belonging to minority ethnic groups is very low in north east Scotland (in the 2001 census 6% of the population reported themselves to be not white in Aberdeen City, 1.6% in Aberdeenshire and 0.8% in Moray15); the proportion is likely to have been even lower in the 1970s and 1980s.

Socioeconomic status

Our primary index of socioeconomic status was the 2009 Scottish Index of Multiple Deprivation (SIMD), which was derived from the school postcode, and our secondary index was the percentage of children at a school who were entitled to free school meals. SIMD is a measure of neighbourhood deprivation based on census data, which incorporates seven domains, including income, employment, health, education, skills and training, housing, geographical access and crime. For data collected in the later part of the study, the child's home postcode was also recorded; for children whose address had been changed in the records, it was not possible to retrospectively determine their postcode at the time of measurements and home postcode was not collected.

Sources of data

The origins of data depended on the child's year of birth. (i) Years of birth 1970–1988: a data extract from the Scottish morbidity records (SMR)10 School Health Record was provided by the Child Health Information Team of the Information Services Division (ISD) Scotland and NHS National Services Scotland. (ii) Years of birth 1989–1993: measurements were retrieved from archived school medical records, which are stored after individuals leave school until their 25th birthday. (iii) Years of birth 1992–2003: measurements were retrieved from school medical records held by the school nurses. (iv) Years of birth 2003–2006: measurements were provided by NHS Grampian from electronically entered data.

Height and weight measurements

Measurements had been carried out by school nurses: no information was available on equipment for protocols used. BMI was calculated and expressed as a centile and z score of the UK 1990 reference data.16 ,17 BMI z scores of ±5 SDs were excluded on the assumption that these were wrongly entered data (n=293, of whom 248 had BMI z score less than −5 and 45 had BMI z score more than 5); corresponding heights and weights were also removed.

Definitions

Clinical thresholds were used, and overweight was defined as BMI≥91st centile and obesity as BMI≥98th centile; these values approximate to BMI z scores of 1.333 and 2, respectively.

Analysis

Chi-square  and Student's t test were used to compare characteristics of the children between those for whom data were and were not obtained. Chi-square was used to compare SIMD quintile derived from school and home postcode. A univariate multivariable linear model was created to relate height or weight z score to year of birth adjusting for socioeconomic status. Whole population and sex-specific prevalence curves for overweight and obese were plotted, and (piecewise) linear spline method with four equally spaced internal knots at the year of birth—1977, 1984, 1991 and 199818—was used to model the prevalence of obesity. Linear spline was used as opposed to cubic spline: although cubic spline gives a smooth curve, its regression coefficients are hard to interpret, but linear spline provides pattern in the prevalence as well as the regression coefficient, which can be meaningfully interpreted. A methodology described elsewhere10 was used to compare change in the prevalence of overweight and obesity over time between SIMD quintiles and then separately by gender. Standard statistical software was used (SPSS V.19.0.0 and STATA for the spline analysis), and significance was assumed at p<0.05.

Results

Study subjects

The records of 208 479 children were accessed, of whom BMI could be determined in 194 391 (93%), and there were no means to identify why measurements were missing for some children. The measurements were made between 1975 and 2011. Scottish government data (http://www.scotland.gov.uk/Topics/Statistics/Browse/School-Education/PupilTeacherHistoric) indicate that the median number of children where BMI was measured was 81% of the total children entering school during 1994–2011; this suggests that in addition to the 9% of children where BMI was not recorded, there was a similar proportion where records were missing/who attended private schools. Online supplementary tables S1 and S2 present the number of boys and girls assessed each year and their average age, height, weight, BMI and BMI SD score.

Table 1 shows how children where BMI data were not available were older and from more affluent communities (higher SIMD) compared to those where BMI data were available. There was a good agreement between SIMD quintile from school and home domestic postcode among the 17 485 where paired data were available (p<0.001).

Table 1

Comparison of details of children where height and weight data were and were not obtained

Whole-population trends in height and weight

There was a linear increase in mean height z score for children born between 1970 and 2000, inclusive (figure 1) and a linear increase in mean weight z score for children born between 1976 and 2000 (figure 1). Table 2 presents mean z scores for height and BMI for children born in the 1970s, 1980s, 1990s and 2000s.

Table 2

Mean z score for height and body mass index (BMI) for children born in the 1970s, 1980s, 1990s and 2000s

Figure 1

Mean difference and 95% CIs for changes in height z score (red square) and weight z score (black diamonds) in 6-year-old children born between 1970 and 2006 with reference to those born in 1970. Values are adjusted for deprivation index.

Prevalence of overweight and obesity

Over the whole period, the proportion of children who were overweight or obese rose by an average of 0.24%/year (95% CI 0.18% to 0.30%) p<0.001, and the prevalence of obesity rose by 0.15%/year (0.12% to 0.18%) p<0.001. Figure 2 demonstrates how the prevalence of obesity and overweight changed over time, and tables 3 and 4 presents the output of linear spline analysis of these trends. The absolute numbers of overweight and obese children for each birth year are presented in online supplementary table S3. The prevalence of overweight and obesity using the International Obesity Task Force criteria is presented in text and in the online supplementary figure S1. The interaction term between year of birth and SIMD quintile was significant for overweight (p=0.002) and obesity (p<0.001); the prevalence of obesity was initially highest in the most affluent group but became persistently highest in the least affluent group among those born during 1999–2006 (figure 3). Similar results were seen when free school meals were used as the index of socioeconomic deprivation (p<0.001 for interactions with free school meals and both obesity and overweight). The interaction term between year of birth and gender was significant for overweight (p=0.001) and obesity (p<0.001). The prevalence of overweight and obesity was initially greater in girls than in boys but this trend reversed in those born after 1988 (see online supplementary figure S2).

Table 3

The regression coefficient from the linear spline regression of prevalence of obesity for boys, girls and all children born between 1970 and 2006

Table 4

The regression coefficient from the linear spline regression of prevalence of overweight for boys, girls and all children born between 1970 and 2006

Figure 2

Overall prevalence of overweight/obese plotted against year of birth with fitted linear spline. Overweight was defined as body mass index (BMI)≥91st centile, and obesity was defined as BMI≥98th centile.

Figure 3

Obesity prevalence for children born between 1970 and 2006 stratified by highest, middle and lowest quintiles of affluence. Data from the second and the fourth quintiles are not presented to allow easier viewing of the figure. Data are presented in 5-year periods (7 years, 2000–2006) to allow smoothing of lines since the numbers of obese children in the least affluent quintile were relatively small for individual birth years.

Discussion

This observational study describes trends in weight, height, BMI and obesity for at least 80% of all 5–6-year-old children in north east Scotland who were born between 1970 and 2006. The first finding was of a difference in timing between the onset of rises in mean population height and weight, suggesting different drivers for these measurements. A second finding was that the prevalences of overweight and obesity have progressively fallen in those born after 1998. Additionally, we observed how the prevalence of obesity was initially highest among girls and the most affluent but became highest in boys and those living in the least affluent communities. These observations give insight into the dynamic secular trends in anthropometric measurements within a whole population of young children between 1970 and 2006.

The changing prevalence of obesity we report here is mostly consistent with other surveys of British children who were measured up to the mid-2000s and our results extend these other findings to 2011. Children became taller and more obese between 1972 and 199419 and we observed similar trends (figure 1). Surveys of older British children between 1997 and 2005, which applied BMI centile cut-offs of 95th centile12 and 98th centile,14 report obesity prevalence between 10% and 14%, which are higher than in our population for this period, but this difference may be at least partly explained by differences in age. Obesity prevalence in British children in one study rose fourfold between 1984 and 2003,3 and among Scottish children, the obesity prevalence approximately doubled between 1984 and 1994 but had been static between 1974 and 1984;4 our findings differ slightly to these patterns as we report an abrupt rise in obesity in children born between 1974 and 1994 (measured between 1980 and 2000) but a respective plateau and decline thereafter. One study of 6-year-old children in Aberdeen reported falling obesity prevalence for those born between 1995 and 1998;12 this trend for falling obesity prevalence is slightly earlier than that seen in this larger study but is consistent with a trend towards a falling obesity prevalence in more affluent communities.

Our findings can also be compared with studies of secular trends for height and weight in children. The 0.8 mm year-on-year increase in height for children born between 1970 and 1992 will cumulatively be similar to the 2 cm increase in height for Scottish children measured annually or biannually between 1972 and 199419 and also to a 3 cm increase in final adult height in the Netherlands between 1981 and 2000, but no increase in men thereafter.20 The present study of children and that of adults in the Netherlands is consistent with the possibility that height gain in a population is due to increasing limb rather than truncal length occurring before the age of 2 years.21 In Japan, there were increases in children's standing height during a period of increasing affluence during the 1960s and 1970s.22 Children have also become taller and heavier over time in Turkey,23 China24 and Poland,25 with evidence for a plateau being reached in Poland.25

There are a number of strengths and limitations to this study. Strengths include the capture of measurements from more than 80% of a whole population born between 1970 and 2006. We were also able to ascertain that incomplete data only minimally affected the population demographics. One limitation was that there was no standardised methodology for collection of data, and although the curves in figure 1 are mostly smooth, there are some curious bumps (eg, height in 1980 and 1981), which might represent minor variations in methodology; community child health records have been validated for research purposes.26 A second limitation is that our index of deprivation was derived from the school and we assumed that deprivation in 2009 reflected that over the previous 40 years. A third limitation is that we excluded 293 children whose measurements were more than 5 SDs from the mean; this approach ensured that incorrect data entry did not bias the results, but some genuinely very underweight/overweight individuals may have been excluded. Finally, despite the large number included in the study, there were relatively few children in the least affluent quintiles as north east Scotland is relatively affluent compared with the rest of Scotland; we can therefore be more certain of trends in the more affluent communities.

This study was not designed to explain why height, weight and obesity trajectories changed over time, but the underlying mechanisms are complex and include environmental, genetic and behavioural factors.27 Tanner stated that ‘growth is a mirror of the conditions of society’28 and, consistent with this concept, a Japanese study observed increases in children's standing height during a period of increasing affluence during the 1960s and 1970s.22 The discovery of oil in the North Sea in the early 1970s resulted in increasing affluence and high levels of employment across north east Scotland, which persists to this day, and this may have had implications for changes in height and weight of the population for children born in the latter 1970s and thereafter. The reduction in weight and BMI in children born after 1998 may be in part consequent to public health initiatives promoting healthy eating and active lifestyle that were introduced in Scotland, for example, ‘Hungry for Success’, which was introduced in 2003.29 Genetic factors are important to growth and obesity, but the changes in height, weight and obesity seen in our cohort occurred too quickly for genetic factors to be primarily responsible, although gene–environment interactions may be important.30

In summary, we report the whole-population study of height, weight and BMI in young children born between 1970 and 2006, where we observed first a rise in height and then in weight. What remains uncertain is what factors drive these secular changes and how these might be modified to reverse and prevent obesity.

Acknowledgments

We would like to thank Mr Alistair Cross and Mr John Macleod for their assistance in accessing the archived data stored in Aberdeen. We are also grateful to the following individuals (listed alphabetically) for their assistance in retrieving archived data and assisting with database creation: Natalie Alexander Anna Celnik, David Devereux, Nigel Hessing, Caren Macleod, Tiffany Martin, Sarah Milliken, Natasha Moir, Emma Ross, Alison Shipp and Janice Todd. We are also grateful for funding support from NHS Grampian, which has made this project possible, and, in particular, we acknowledge Kim Penman (Aberdeenshire CHP), Caroline Comerford (Aberdeen City CHP) and Tracey Gervaise (Moray CHP). Finally, we would like to thank the staff at ISD, and, in particular, Judith Tate for providing archived data.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Correction notice This paper has been amended since it was published Online First. On the first page in the “What this study adds” box the first bullet point did not make sense. This has now been rewritten by the authors.

  • Contributors SMS collected data, LCAC and GMcN provided advice on obesity, SWT conceived the idea, analysed the data and wrote the first draft of the manuscript. AER undertook the spline analysis. All authors made important contributions to the final manuscript.

  • Funding NHS Grampian funded the study. The researchers were independent of the funders.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Atoms
    R Mark Beattie