METHODS

In the original longitudinal study, 694 children (378 boys and 316 girls) had participated, with measurements taken in 1996, 1997, and 1998. Of these children, 608 were tracked from school leaving lists and the local education authority database to 32 secondary schools. We excluded 14 schools on the basis that they had fewer than five pupils from the original study and were distantly located. The remaining 18 schools had 500 pupils who were targeted for follow up. Written consent to participate was obtained from both the pupils and their parents.

Data were collected in March–June 2001. Height was measured to 0.1 cm with a freestanding magnimeter stadiometer (Raven Dunmow). Weight was recorded to 0.1 kg without shoes or sweaters. The mean of three triceps measurements was taken.⁵ Waist circumference was measured 4 cm above the umbilicus. Body mass index (BMI)was calculated from the formula (weight in kg/height in m²). Data were converted to SD scores using the UK 1990 growth references for height, weight, and BMI,⁶ the 1975 Tanner references for triceps measurements,⁵ and the 2001 McCarthy references for waist circumference.⁴ Puberty was staged by self report.⁷ The International Obesity Task Force criteria³ were used to calculate the numbers of obese and overweight children (table 1).

SD scores of participants and nonparticipants were compared by unpaired Student’s t test. The percentages of overweight and obese children in each year were compared using McNemar’s test. Random effects regression was used to test for trends over the 6 years of the study.

RESULTS

Of the 500 possible children, 348, now aged 12–14 years, agreed to participate in the study. Complete growth measures were obtained from 315 (63% of 500) of whom 174 were boys and 141 were girls. As there was concern that children who did not consent might be more obese, participants’ and nonparticipants’ growth data from 1998 were compared (see table 2). There was no significant difference for any measurement, suggesting that this was not the case.

The demographics of the children used to generate data for figs 1 and 2 are shown in table 3. Fig 1 shows the mean (SD) scores for growth. There was a substantial increase in weight throughout the period 1996–2001. Mean (SD) height scores also increased in the final year. As expected, all subjects were in puberty. Only 39 of the girls were premenarchal. The mean age for menarche was 12.6 (0.7) years.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Changes in growth measures over time. The trend in mean (SD) score was assessed by random effects regression.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Changes in obesity measures over time. The trend in mean (SD) score was assessed by random effects regression.

Obesity measures are shown in fig 2. Mean (SD) BMI and waist circumference scores increased substantially, indicating a rise in obesity. The triceps measurements did not change significantly. Results did not change for any measurement when boys and girls were analysed separately.

Overweight and obesity rates using the International Obesity Task Force criteria (table 1) are shown in table 4. Numbers of overweight and obese children were higher than expected from the 1990 growth references. There was no significant trend over time, but there appeared to be a tendency towards girls being more overweight and obese than boys, with an increase over the years.

DISCUSSION

Over six years the population showed a substantial increase in height and weight, although the former was restricted to the final year. The most likely explanation for this rise in height is an earlier onset of puberty. However, we were unable to demonstrate this as the subjects were already in puberty. The mean age for menarche was 3.5 months younger than that found in a survey of British girls from the early 1980s,⁸ which provides little support for this suggestion. Pubertal status was ascertained by self report rather than examination, with somewhat inconsistent findings, so this relationship could not be explored further.

Despite the increase in height, BMI scores continued to increase substantially, indicating a further rise in obesity for the population as a whole. This was probably caused by children becoming more overweight as they got older,⁶ although we could not disentangle the secular trend from the age trend because of their co-linearity. The triceps measurements were not consistent with other measurements, as previously noted.⁶

In addition to providing further epidemiological data on the rise in children’s weight, this report provides information on waist circumference measurements. Waist circumferences were significantly larger in 1996–8 than in the late 1970s and 1980s when the references were obtained,⁴ and were also larger than BMI. They continued to rise through the course of the study, so that by 2001 children’s waists were on average 4 cm (two clothing sizes) larger than they had been 20 years ago. This figure is all the more disturbing when one reflects on how many notches on a belt this represents, and is consistent with the cross sectional survey of waist measurements recently reported in the BMJ.⁹ Waist circumference as a measure of obesity may be of particular significance, given the association between abdominal girth in adults and cardiovascular morbidity.

It has been recommended that we should now be utilising new criteria for identification of obesity. These have been developed by the International Obesity Task Force in order to allow for international comparison and consistency. As our numbers show, these criteria are more stringent than the previous cutoff points and the numbers of obese children now appear to be relatively small. It is to be hoped that this will not lull us into a false sense of security about the problem.

Our data are of concern. The previous rise in BMI levels reported through primary school⁶ has continued into secondary school and emphasises the need for more rigorous efforts to stem the tide of child obesity. Waist circumference measures provide an extra level of concern. Larger cohort studies specifically designed for the purpose are required to monitor obesity trends, and effective interventions are urgently needed to address this major public health issue.