This is a prospective study of 1605 adolescent students (the Young-HUNT Study) in Nord-Trøndelag County in Norway, who was examined twice with four years apart. The baseline study was conducted in middle school (age 12–16 years) between August 1995 and June 1997 (baseline). The Young HUNT 2 study (follow-up) took place between January 2000 and June 2001, when the students were in 12^th or 13^th grade in high school (age 16–20 years). The overall attendance was 95% at baseline and 86% at follow-up. The baseline and follow-up study were performed with identical and standardized procedures. Within a month after completing a self-administered questionnaire the students underwent clinical examinations, including measurements of height, weight, and waist and hip circumference. Among the 1605 participants with clinical assessments, 697 boys and 864 girls with registered information about maturational timing were eligible for analyses.

Self-reports of pubertal status in boys and age at menarche in girls were used to indicate timing of sexual maturation. We classified each individual as early or late relative to intermediate maturing boys and girls of the same age. At baseline, boys were asked to assess changes in secondary sexual characteristics by use of the Pubertal Developmental Scale (PDS). The PDS was originally developed by Petersen et al, and has later been validated 1819. In our study, reliability tests showed good internal consistency between the reported characteristics of sexual maturation (α = 0.85). The boys rated themselves according to growth spurt and pubic hair growth; and were asked to assess changes in voice and facial hair growth. On a scale ranging from 1 (have not begun) to 4 (development completed) they were asked to report the extent of pubertal changes. Total score for the four items was summarized and divided by 4, in order to maintain the original range from 1 to 4. Stratified by age, we classified boys who had PDS score in the lowest quintile as "late" matured, and those who were in the highest quintile as "early" matured. Those who scored between these extremes were classified as "intermediate" (Table 1).

At baseline, girls were asked to report their age at menarche (AAM) in years and months. Reliability of self-reported AAM is well established 20. One fifth of the responders had not yet reached menarche (n = 158), or had incomplete or missing data (n = 30). For these girls we relied on information on AAM collected at follow-up. Mean AAM was 13.2 years (SD = 1.2). We classified girls whose menarche occurred before 12.5 years (20%) as "early" matured, girls with AAM between 12.5–13.9 years (60%) were classified as "intermediate", and girls with AAM at 14.0 years or later (20%) as "late" (Table 1).

Anthropometry was measured with the participants wearing light clothes and no shoes. Height and waist circumference was read to the nearest centimetre (cm) and weight to the nearest half kilogram (kg). We used waist circumference, as assessed at baseline as indicator of level of central adiposity (Table 1). Participants with waist circumference above the median within each gender and age group (i.e. half-year groups) were classified as having "high waist" and those below or at the median, were classified as having "low waist". Validation studies have shown high correlation between waist circumference and trunk fat measured by DEXA- scan, that measures both fat and lean body mass, among 3–19 years old children and adolescents 2122.

As outcome, we used BMI and overweight as registered at follow-up in late adolescence. Body-mass index (BMI) was calculated as weight in kilograms divided by the squared value of height in meters (kg/m²). Overweight was classified according to The International Obesity Task Force (IOTF) standards, which link BMI percentiles during childhood and adolescence to the widely used adult cut-off-points of 25 kg/m² at age 18, as a definition of overweight 23.

Participation was voluntary and each student signed a written statement of consent. For students younger than 16 years of age, we also obtained written consent from their parents. The Young HUNT study was approved by the Regional committee for ethics in medical research, and by the Norwegian Data Inspectorate.

We examined the independent and modifying effect of early central adiposity (i.e. waist circumference > median) on the association between timing of sexual maturation and subsequent fatness (i.e. overweight and BMI) by multivariate logistic analyses and linear regression analyses, and by stratification. Logarithmic (ln) transformation of BMI at follow-up was used to fit model assumptions in linear regression analyses. Logistic regression analysis was used to study the association between timing of sexual maturation and prevalence of overweight at follow-up, in groups characterised by high (> median) or low (≤ median) waist circumference at baseline, adjusting for potential confounding of age, waist circumference, parental education level, frequency of leisure time physical activity, and frequency of intake of sweets and soft drinks as reported at baseline. We used a general linear model to compute multivariate adjusted mean BMI and mean differences in BMI at follow-up, conducted separately for early, intermediate and late matured boys and girls, adjusting for the potential confounding of age, parental education level, frequency of leisure time physical activity and of intake of sweets and soft drinks as reported at baseline. Precision of the estimates was assessed with 95% confidence intervals (CI) and all reported P-values are two-sided. All statistical analyses were performed using SPSS for Windows, version 13 (SPSS, Chicago, USA).

At baseline, pubertal status in boys and age at menarche in girls differed significantly between individuals classified as early, intermediate and late (Table 1). Age adjusted mean waist circumference (Table 1) was higher in early (73.6 cm) maturing boys, compared to boys who were intermediate (72.1 cm) (P-value = .03) or late matured (69.8 cm) (P-value < .001). Waist circumference did not differ significantly between early (68.4 cm) and intermediate (68.6 cm) maturing girls, whereas late maturing girls on average were leaner (65.8 cm) (P-value < .001) than intermediate matured girls. Age adjusted mean BMI at baseline differed significantly between all timing categories and was highest among early maturing boys (20.7 kg/m²) and girls (21.1 kg/m²) and lowest among late maturing boys (19.2 kg/m²) and girls (18.9 kg/m²).

During follow-up, mean BMI increased by more than 2 kg/m² and in late adolescence, overweight was more common in all timing categories than at baseline, except for boys with late sexual maturation. Among those classified as overweight at baseline, 80% of boys and 70% of girls were also classified as overweight at follow-up (data not shown).

At follow-up, 20% of boys and 19% of girls were classified as overweight. Overweight was more than twice as common among early matured boys and girls (26%), compared to boys (11%) and girls (10%) who had matured late. After adjustment for age, parental education level, frequency of leisure time physical activity and frequency of intake of sweets and soft drinks as reported at baseline, early maturing girls, but not boys, were slightly more likely (OR, 1.6, 95% CI 1.0 -2.5) to be overweight, compared to girls who were intermediate matured (P-value = .03). Late matured boys (OR, 0.5, 95% CI 0.3–0.9) and girls (OR, 0.5, 95% CI 0.3–0.8) were both less likely to be overweight compared to intermediate matured counterparts.

We studied whether the association between timing of sexual maturation and overweight at follow-up could be modified by level of central adiposity at baseline (indicated by waist circumference). The results showed that for girls, but not for boys, there was a statistically significant interaction between timing categories and baseline waist circumference (P-value = .006). Further stratified analysis showed that among girls with low waist circumference (≤ median), early AAM was not associated with subsequent overweight at follow-up (Table 2). Among girls with high waist (> median), early matured girls were 2.7 times as likely (95% CI 1.5–4.9) to be overweight, compared to intermediate matured girls within this group. Odds ratios were adjusted for age, waist circumference, parental education level, physical activity and intake of soft drinks and sweets as obtained at baseline (Table 2).

By use of linear regression analyses, we estimated the independent and combined influence of timing of sexual maturation and waist circumference on mean BMI at follow-up. In boys (Table 3), both pubertal status score (P-value = .005) and waist circumference (P-value < .000) at baseline (age-adjusted), contributed independently to BMI at follow-up (Model 1), but the interaction term was not statistically significant (Model 2). In girls, age at menarche (P-value = .002) and waist circumference (P-value < .000) were both independently related to BMI (Model 1), but in girls these effects were interactive (Model 2, P-value = .002). Inclusion of the interaction term, however, did not change the adjusted R² (Table 3).

Further stratified analyses showed that across all maturational timing groups, mean BMI at follow-up was significantly higher among boys and girls with high waist circumference (> median), compared to those with low circumference (≤ median) at baseline (Table 4). In girls, however, differences appeared to be more pronounced among those with early age at menarche (3.5 kg/m², 95% CI 1.7–5.3) than among intermediate (2.7 kg/m², 95% CI 1.8–3.6) and late matured girls (1.2 kg/m², 95% CI 0.1–2.2). Differences in age and height adjusted weight at follow-up, showed that early matured girls with high waist circumference at baseline (> median) weighed on average 11.2 kg more than those with low waist (≤ median) (P-value < .001) (data not shown).