Childhood obesity has escalated worldwide over the past two decades(Reference Ogden, Carroll and Curtin1–Reference Wang and Lobstein3), with obesity complications, previously thought of as adult conditions, now affecting children(4–Reference Lobstein, Baur and Uauy7). In addition excess body weight in childhood and particularly in adolescence has the tendency to track into adulthood(Reference Whitaker, Wright and Pepe8, Reference Power, Lake and Cole9). Thus, from a public health perspective, this is an important time to monitor weight trends.
Ireland is no exception to the obesity epidemic, with 25 % of young children(Reference McMaster, Cullen and Raymond10, Reference O'Neill, McCarthy and Burke11) and adolescents(Reference O'Neill, McCarthy and Burke11) overweight and obese compared with 19 % in Great Britain(Reference Jebb, Rennie and Cole12). Alarmingly, an international survey identified adolescents from Ireland among the most obese in the world(Reference Lissau, Overpeck and Ruan13). Epidemiological evidence from Northern Ireland (NI) is sparse, with 25 % of 4–16-year-olds overweight or obese in 2002(Reference Whelton, Harrington and Crowley14); similar to that reported in the Republic of Ireland in 1998(Reference Lissau, Overpeck and Ruan13). Trend data from the Young Hearts Project reported overall increases in height and weight in 12–15-year-olds between 1990 and 2000, with overweight and obesity increasing from 15·0 % to 19·6 %(Reference Watkins, Murray and McCarron15). However no trend data from NI have been published since. It would be of interest to see if recent trend data replicate those in England showing marked increases in obese young people between 1995 and 2005, but a plateau or even a decrease towards 2007(Reference Stamatakis, Zaninotto and Falaschetti16). Moreover, those from lower socio-economic circumstances appear to be at greater risk(Reference Jebb, Rennie and Cole12, Reference Stamatakis, Zaninotto and Falaschetti16, Reference Howe, Galobardes and Sattar17), particularly girls(Reference Howe, Galobardes and Sattar17). Future projections highlight this socio-economic divide to worsen in England(Reference Stamatakis, Zaninotto and Falaschetti16) but the magnitude of this gap within NI adolescents is unknown.
In the recent years of the ‘obesity epidemic’, research examining trends of underweight among children has been neglected. In Scotland 3·3 % of pre-school children were identified as underweight, which was related to social deprivation(Reference Armstrong, Dorosty and Reilly18), while trend data in 9–10-year-olds from 1998 to 2006 observed a decline(Reference Boddy, Hackett and Stratton19). Few data are available on the prevalence of underweight adolescents and given the media emphasis of body image(Reference Hogan and Strasburger20) this is an important time to examine both ends of the body mass spectrum.
It would also be of interest to examine whether the dieting behaviour of adolescents has changed over time. An association between overweight children and an increased risk of disordered eating symptoms including weight concerns, dieting and other unhealthy weight-control methods has been established(Reference Goldschmidt, Aspen and Sinton21). There is the perception in the Western world that thinness is the ideal body shape, particularly for females. Therefore, investigating whether adherence to weight-reduction diets has increased in parallel with increasing obesity would be of interest, particularly as disordered eating patterns can predict future weight gain(Reference Shisslak, Crago and McKnight22).
Earlier cross-sectional data from our group collected in 1996 from NI adolescents aged 12–15 years observed that 16 % were overweight and obese and low levels of concern or awareness manifested as attempts to modify diet(Reference Yarnell, McCrum and Patterson23). By repeating this representative cross-sectional study in the same age group, we had the opportunity to examine more recent trends. In the present paper we investigate 11-year trends in underweight, overweight and obesity in 12–15-year-old adolescents in NI and examine changes in dieting behaviour.
Subjects and methods
Survey and sample population
These surveys were conducted using data from the International Study of Asthma and Allergies (ISAAC), which is a series of worldwide repeat cross-sectional studies in schoolchildren. Investigations in NI have been conducted in 12–15-year-olds, following ethical approval granted by the Queen's University Research Ethics Committee in 1996 and the Office of Research Ethics Committees Northern Ireland in 2007. The sampling procedure for each survey aimed to obtain a random sample of schoolchildren, stratified by management type and approximating to religious affiliation and geographical area. The five Education and Library Boards in NI were contacted and lists of secondary and grammar schools were obtained. Letters detailing the study design and objectives were sent to a predefined number of school principals within each board and management sector; if they agreed to participate in the study, consent was obtained on a class basis unless a parent or child opted out. If the principal was unable to accommodate the researchers, the school was replaced at random by another in the same stratum.
Anthropometry and questionnaire data
Previous work conducted by our group investigated the cross-sectional data from the 1996 cohort(Reference Yarnell, McCrum and Patterson23). Briefly, heights and weights were measured by two trained researchers while participants were wearing lightweight clothing and no shoes, using equipment which was available in each school and after calibration with standard measures. Questionnaires were completed by the selected participants under classroom conditions. The further data collection in 2007 has now provided the opportunity to examine trends in overweight and obesity in NI adolescents over the past 11 years. Data collection was carried out in the same manner as before, only this time using the same calibrated equipment for every school. Both surveys incorporated questions to assess dieting behaviour. Two questions were of relevance to the present paper. First, the participants were asked the occupation of their father and/or mother, which was categorised into manual or non-manual socio-economic status (SES) according to the Registrar General's Classification(Reference Simon and Szreter24). The participant's SES was based on the highest socio-economic occupation within the household. Second the participants were asked if they were following any special diet and, if so, they were asked to select one of the following: (i) a weight-reducing diet devised by yourself or a parent; (ii) a weight-reducing diet proposed by a dietitian or doctor; (iii) a diet prescribed for a medical reason such as diabetes or food allergy; (iv) a vegetarian diet; or (v) other (please specify).
Definition of underweight, overweight and obesity
The International Obesity Taskforce (IOTF) age (to nearest 6 months) and sex-specific BMI thresholds for defining underweight (thinness), overweight and obesity classification between 2- and 18-year-olds were used(Reference Cole, Flegal and Nicholls25, Reference Cole, Bellizzi and Flegal26).
Statistical analysis
Statistical tests were carried out using the SPSS statistical software package version 17·0 (SPSS UK Ltd, Chertsey, UK). Differences in the proportions of underweight, overweight and obese children and those on specific diets between the two surveys were analysed using χ 2 tests and χ 2 tests for trend as appropriate. The mean and standard deviation were calculated for height, weight and BMI, in addition to BMI across specific diet groups, between the surveys. Differences were then assessed using independent-sample t tests for boys and girls adjusting for age. Confidence intervals for small proportions (i.e. those on weight-reducing/medical diets within each of the weight categories) were calculated using Wilson's Procedure with continuity correction(Reference Wilson27). Due to the number of significance tests applied and to reduce the number of spurious significant results that could ensue, we applied a significance level of 0·01 rather than the customary 0·05.
Results
Response rate
A total sample of 2484 adolescents (53 % male) completed the survey in 1996 from twenty-six schools across NI between the months of April to June. The overall response rate was high with less than 0·5 % of parents refusing consent to their child's participation in the study. Approximately 10 % (258 adolescents) of those adolescents initially selected for participation in the survey did not complete the questionnaire due to absenteeism at the time of researchers visiting the schools. The 2007 survey consisted of 2647 adolescents (48 % male) from twenty-four schools across the same geographical area between the months of May to November. Approximately 1 % of parents refused their child's participation in the study, and 14 % (427 adolescents) were absent on the day of survey completion. There was a small trend in socio-economic distribution of families from semi/unskilled in 1996 towards professional in 2007 reflecting a societal change over the 11-year period, with a small increase in single-parent families.
Prevalence of underweight, overweight and obesity
The prevalence of overweight and obesity increased significantly over the decade between the two surveys (Table 1), with rates of overweight and obesity increasing from 13·6 % to 20·4 % and from 3·0 % to 6·2 % respectively, in contrast to a decrease from 8·4 % to 5·0 % in prevalence of underweight in all adolescents (P < 0·001). Overall the largest trend towards overweight and obesity was seen in girls, with the prevalence of obesity increasing from 2·4 % to 6·4 % and overweight from 14·5 % to 22·8 %. As with girls, a significant trend towards increasing overweight and obesity was also observed in boys, increasing from 12·8 % to 17·9 % and from 3·4 % to 6·0 % respectively. Table 2 shows that mean weight and BMI, but not height, increased significantly (both P ≤ 0·001) in both boys and girls between the two surveys. These analyses were adjusted for age, as there was a 3-month mean age difference between the two cohorts. A further adjustment for SES did not significantly alter the results.
Eleven-year trends in underweight, overweight and obesity based on BMI in Northern Ireland adolescents aged 12–15 years: data from the International Study of Asthma and Allergies (ISAAC)Footnote *
Differences in the proportions of underweight, normal weight, overweight and obese adolescents in each cohort were analysed using the χ 2 test for trend.
* Data missing for n 193 from ISAAC 1996 and n 26 from ISAAC 2007.
Eleven-year trends in height, weight and BMI in Northern Ireland adolescents aged 12–15 years following age adjustment: data from the International Study of Asthma and Allergies (ISAAC)
Differences between the two cohorts were analysed using one-way analysis of covariance controlling for age, significant when P ≤ 0·01.
Dieting behaviour
Table 3 highlights an overall reduced level of dietary concern in 2007 compared with 1996, particularly in girls; 19·5 % of girls in 1996 were following some form of special diet compared with 10·7 % in 2007. The percentage of girls following a self-proposed weight-reduction diet decreased from 10·7 % to 5·8 % although this remained stable for boys (3·6 % to 3·2 %). The reduction in girls following a self-proposed weight-reducing diet occurred even though there was an increase in mean BMI over the same period. The proportion of adolescents following a prescribed weight-reduction or medical diet (1996, 2·0 % and 1·4 %; 2007, 0·9 % and 0·9 %, girls and boys respectively) remained relatively stable despite the overall increasing prevalence of overweight and obesity; all those who were on prescribed weight-reducing diets had a BMI >25 kg/m2. The percentage following a vegetarian diet decreased threefold in girls and twofold in boys over the 11-year period.
Number, percentage and BMI of adolescents aged 12–15 years on particular diets in Northern Ireland: data from the International Study of Asthma and Allergies (ISAAC)
*Data unavailable for eighty-nine adolescents.
†Prescribed weight-reducing (1996, n 2 and n 5; 2007, n 1 and n 3, boys and girls respectively) or medical diet (1996, n 9 and n 16; 2007, n 9 and n 16 boys and girls respectively). Differences between the number of children for specific diet category in each cohort analysed using the χ 2 test, significant when P ≤ 0·01.
Examining those adolescents who were following a self-proposed or prescribed weight-reduction diet by BMI category (Table 4) revealed a smaller proportion of overweight/obese adolescents following weight-reduction diets in 2007 compared with 1996. Overweight and obese boys following these particular diets decreased from 12·7 % to 7·0 % and girls from 18·5 % to 12·9 %. Also we hypothesised that more girls in the normal weight/underweight category would be following self-proposed weight-reduction diets in 2007 due to media-generated peer pressure; however this was not the case, and only 3·1 % of these girls were following weight-reduction diets in 2007 compared with 9·8 % in the previous survey.
Comparison of number and percentage of Northern Ireland adolescents aged 12–15 years on weight-reduction diets (self-proposed or prescribed) over an 11-year period: data from the International Study of Asthma and Allergies (ISAAC)
CI calculated using Wilson's Procedure with continuity correction(Reference Wilson27).
Change in weight and dieting behaviour by socio-economic status
Figure 1 indicates that the relative risk of being overweight and obese as a proportion of all young people generally increased with a manual socio-economic background, being female and being from the later cohort. Boys showed little difference across the socio-economic groups, although they did show significant increases over time. Girls, whether from a manual or non-manual background, showed the greatest increased relative risk over time. On investigating the proportion of overweight/obese adolescents following weight-reduction diets (either self-proposed or prescribed), girls from a manual socio-economic background had the greatest decrease over time from 24·6 % to 10·6 % (relative risk 0·48, 95 % CI 0·25, 0·93). Proportionally smaller decreases were observed in boys from both manual and non-manual backgrounds whereas a slight increase was noted in girls from a non-manual background.
Evidence of social disparities over time in dieting behaviour among overweight and obese (OW/OB) boys (a) and girls (b) aged 12–15 years in Northern Ireland: - -▪- -, % OW/OB from manual socio-economic status (SES); —▪—, % OW/OB from non-manual SES; - -●- -, % OW/OB on diet from manual SES; —●—, % OW/OB on diet from non-manual SES. Data from the International Study of Asthma and Allergies (ISAAC)
Discussion
The current study observed marked increases in the prevalence of overweight and obesity over an 11-year period in 12–15-year-old adolescents from NI. Approximately 30 % of the girls surveyed were classified as overweight or obese using international definitions(Reference Cole, Bellizzi and Flegal26) with the prevalence of obesity increasing threefold. Increases over time were more modest for boys; however 25 % were still classified as overweight or obese. These increases occurred predominantly in body weight since increases in height were smaller and non-significant. It was hypothesised that, along with increases in overweight and obesity, the prevalence of underweight adolescents, particularly girls, would increase as a result of increased awareness of body image due to a media focus on fashion and celebrities. On the contrary, there appeared to be a reduced dietary concern, particularly in girls, as less reported being on a self-prescribed or medical weight-reduction diet, despite the trend for increasing BMI. Of particular concern was the evidence of social disparity between girls from manual and non-manual socio-economic backgrounds. Even though there was an upward trend in overweight/obesity in both groups, the proportion of girls from a manual background actually following weight-reduction diets decreased from 25 % to 11 % compared with a small increase in the non-manual girls.
This research provides the most recent estimates of underweight, overweight and obesity in adolescents of this age group in NI. Approximately 25 % of boys, but only 21 % of girls aged 12–15 years were overweight or obese in data collected in NI from the North South Ireland Survey(Reference Whelton, Harrington and Crowley14) conducted in 2001. However, it must be noted that the NI subset for that previous survey was smaller (n 860) and with a lower response rate at 53 %(Reference Whelton, Harrington and Crowley14) than the current study. In agreement with our data, significant increased trends in the number of adolescents who were overweight and obese over a 10-year period in NI has previously been reported(Reference Watkins, Murray and McCarron15). However the Young Hearts Project also reported significant increases in height over time(Reference Watkins, Murray and McCarron15). The Young Hearts data(Reference Watkins, Murray and McCarron15) estimated increases in overweight and obesity from 15·0 % to 19·6 % from 1990 to 2000 whereas we observed a larger increase from 16·6 % to 26·6 % from 1996 to 2007, suggesting that the obesity epidemic is still a major and increasing public health issue in adolescents from NI. However some recent research(Reference Stamatakis, Zaninotto and Falaschetti16, Reference Ogden, Carroll and Flegal28) has shown that the trend in childhood obesity may have reached a plateau. In both of these studies(Reference Stamatakis, Zaninotto and Falaschetti16, Reference Ogden, Carroll and Flegal28), trend data were collected at more time points than in the current survey. Therefore, it is possible that the prevalence rates in NI adolescents increased greatly over the earlier years and have slowed down or reached a plateau more recently.
A novel aspect of the present research was the collection of information on adolescents’ dieting behaviour. A recent qualitative investigation from the Republic of Ireland on adolescent girls observed a high level of body dissatisfaction significantly influenced by media celebrities, together with a high prevalence of dieting(Reference Mooney, Farley and Strugnell29). Given these perceptions we had hypothesised that more girls, particularly in the normal weight and underweight category, would be following weight-reduction diets, as observed in a study of Dublin schoolgirls(Reference Ryan, Gibney and Flynn30). However this was not the case and fewer girls were following self-proposed weight-reduction diets, despite significant increases in BMI over time, particularly those girls from a manual socio-economic background. This widening of social disparities in dietary awareness in adolescents from NI mirrors that observed in England(Reference Stamatakis, Zaninotto and Falaschetti16, Reference Howe, Galobardes and Sattar17) and suggests that the current obesity strategies are either ineffective for those from a manual background, particularly girls, or are more actively followed by those from a higher socio-economic background. A potential cause for this widening in social disparities in girls may be that lower-SES children tend to spend more time in screen-based sedentary behaviours compared with those from higher SES. However, overall, boys (regardless of SES) also tend to spend more time in sports participation than girls, suggesting that lower-SES boys find time for sedentary behaviours and physical activity(Reference Fairclough, Boddy and Hackett31).
In Great Britain changing perceptions of body weight in adults have also been noted(Reference Johnson, Cooke and Croker32), with a considerable proportion of the population, especially men, failing to recognise themselves or their children as overweight(Reference Jeffery, Voss and Metcalf33–Reference Crawford and Campbell35). Population surveys conducted in 1999 and 2007 have highlighted a significant decline in the proportion of individuals correctly identifying themselves as overweight despite the fact that the obesity epidemic is rarely out of the media(Reference Johnson, Cooke and Croker32). Possible reasons for this may be that we are continually exposed to images of severe obesity from the media, portraying that extreme adiposity is required to meet medical criteria for overweight, or that excess weight is now so familiar in the general population that we have a reduced awareness of what excess weight actually is(Reference Johnson, Cooke and Croker32). In addition, recent evidence has shown that the proportion of primary care practitioners addressing the issue of overweight with their patients is inadequate(Reference Michie36). Although the present study did not directly measure adolescent awareness or concern of being overweight, it seems likely that perception of body size over time has reduced in this population, seeing as the prevalence of overweight/obese adolescents has increased while those actively following weight-reduction diets has decreased.
The high response rate from the 1996 and 2007 surveys, together with the careful selection of a representative sample of adolescents from all school types, are strengths of the present research and provide confidence that the prevalence rates are reliable. On the other hand, the cross-sectional nature of these surveys cannot identify any causal factors for increasing overweight and obesity; but whatever the reasons, the condition is continuing to rise in adolescents from NI. It is well documented that obesity in childhood and adolescence tends to track into adulthood(Reference Whitaker, Wright and Pepe8, Reference Power, Lake and Cole9) along with adverse dietary and physical behaviours known to promote obesity(Reference Burke, Beilin and Dunbar37–Reference Janz, Dawson and Mahoney39). Obesity and accompanying dietary and physical behaviours are predictors of adult morbidity and mortality, in the form of CVD and diabetes which are major public health burdens(Reference Must and Strauss5, Reference Dietz6, Reference Reilly40). In the present research we chose to use the IOTF classification of overweight and obesity in children. The use of international cut-off points of BMI is controversial; some have expressed their concern that the IOTF classification underestimates the magnitude of the problem(Reference Reilly41), while others have stressed that the use of national standards inflates the problem in children, and a single definition needs to be consistently applied(Reference Jebb and Prentice42). Nevertheless, these international cut-off points are increasingly being used in national and international studies(Reference McMaster, Cullen and Raymond10, Reference Jebb, Rennie and Cole12, Reference Watkins, Murray and McCarron15, Reference Yarnell, McCrum and Patterson23, Reference Wang, Monteiro and Popkin43–Reference Hawkins, Griffiths and Cole45), and we used these definitions for the sake of consistency with our previous report and comparability with other studies.
Conclusions
The findings of the present study highlight that current government obesity strategies are not adequate and more focused interventions are warranted. The differing trends in overweight/obesity and dietary behaviour by gender and socio-economic background from the present research suggest that the development of separate strategies may be necessary for these factors. Finally, health professionals need to take greater action in the prevention and treatment of childhood obesity, as no significant increases in prescribed weight-reduction diets over time were observed, despite dramatic increases in childhood overweight and obesity.
Acknowledgements
This work was supported by funds from the Health Promotion Agency for Northern Ireland and the Child Health & Welfare Recognised Research Grouping, Research & Development Office, DHSS Northern Ireland. The authors declare that they have no conflicts of interest. C.R.W. analysed the data and drafted the paper; J.V.W., N.M., and M.D.S. helped draft the paper; B.P.G. helped plan the study with J.W.G.Y., who was Principal Investigator for the study. M.S. advised on the sample selection, assisted with the random selection of schools and advised and assisted with the statistical analysis. The authors thank the participating schools, head teachers, parents and children included in the study and acknowledge the meticulous assistance of fieldworkers Ruth Leathem and Pauline Nolan and the technical assistance of Laura Stevenson.
