Numerous epidemiological studies in adults have identified environmental and physiological risk factors that are associated with increased risk for cardiovascular disease (CVD). Among the many that have been identified 1, the major modifiable risk factors include physical inactivity, poor cardiorespiratory fitness, excess adiposity or obesity, and inappropriate dietary habits.

Although the clinically relevant effects of CVD are often not manifest until middle age or later, it is now generally well accepted that the disease is likely to have its antecedents in childhood 234. In addition, children and adolescents have been shown to exhibit many of the potentially modifiable CVD risk factors that have been identified in adults. For example, of the 1015 adolescents aged 12y and 15y who participated in the first phase of the Northern Ireland Young Hearts Project, 18–34% were considered to have excess body fat, 24–29% had low physical activity levels, and 26–34% had poor cardiorespiratory fitness. Furthermore, mean total fat intakes were higher than desirable 5. Similarly, in West Virginia, of the 5,887 school children who participated in the school-based Coronary Artery Risk Detection in Appalachian Communities (CARDIAC) Project, almost 43 percent were considered to be overweight, over a quarter were obese, and the high rate of obesity was positively associated with the prevalence of other CVD risk factors 6. The conclusions drawn from these studies and others 789 have been largely unanimous: minimising the risk of morbidity or premature mortality associated with CVD in adulthood, should begin in childhood or adolescence.

However, if health promotion interventions in younger life are to have any hope of success, it must be assumed that physiological and behavioural risk factors exhibited in early life track into adulthood. Tracking has been defined as the maintenance of relative position in rank of behaviour over time, such that subjects who rank highly for unfavourable risk profiles at a young age are likely to maintain their ranks through into adulthood 1011.

The aim of the current study was to use data from the Northern Ireland Young Hearts Project to ascertain the extent of tracking, between adolescence and young adulthood, of selected modifiable risk factors for CVD. Specifically, we investigated the tracking of physical activity, aerobic fitness, selected anthropometric variables, and diet.

The physical characteristics, aerobic fitness and physical activity levels of the Northern Ireland Young Hearts cohort at age 15y and at follow-up (young adulthood) are summarised in Table 1. At young adulthood, weight, height, BMI and skinfold thicknesses were significantly greater than at age 15y in males and females. In both sexes, VO₂max assessed at young adulthood was significantly lower than at age 15y. At age 15y, males and females were 4.8% and 8.4% heavier than the British age-specific reference population, while at young adulthood, they were 7.9% and 9.9% heavier respectively. Details of the reference populations are described in Annex 1 of the 'Dietary Reference Values for Food Energy and Nutrients for the United Kingdom' 22.

The energy and macronutrient intakes reported by the Young Hearts cohort at age 15y, and at follow-up, are presented in Table 2. At young adulthood, the males reported significantly lower intakes of energy (MJ/d; P 0.04), total fat (g/d and % energy; both P < 0.001) and total carbohydrate (g/d and % energy; both P < 0.001) than at age 15y. In contrast, intakes of protein (g/d and % energy; both P < 0.001) reported by males at young adulthood were significantly greater than at age 15y. Similar patterns were observed for females, with the exception that % energy derived from total carbohydrate did not change significantly between age 15y and young adulthood.

Table 3 summarises the extent of tracking of physical characteristics, fitness and physical activity levels between age 15y and young adulthood, in males and females. In males, tracking of height was moderate, while in females, it was very good. Figure 1(a), which illustrates the 3 × 3 matrix constructed for the tracking of height in females between age 15y and young adulthood, demonstrates that there was very little drift of subjects between low, medium and high categories over time; hence the relatively high κ value. In both sexes, tracking of BMI was moderate. In males, the tracking of weight, four skinfold thicknesses (biceps, triceps, subscapular, suprailiac) and sum of skinfolds was fair (κ 0.21–0.40) between age 15y and young adulthood. The magnitude of the κ values obtained for biceps and subscapular skinfold thicknesses in females were also greater than in males, while triceps and suprailiac skinfold thicknesses tracked to a similar extent in both sexes. In males, the tracking of aerobic fitness (VO₂max) was poor, but was greater than the κ value obtained for the females (κ 0.150 vs κ 0.076). A similar pattern was observed for physical activity scores (κ 0.202 vs κ 0.021).

The extent of tracking of energy and macronutrient intakes reported at age 15y and at young adulthood is presented in Table 4. In males, the κ values for energy, protein, total fat and total carbohydrate were poor, ranging from 0.019 (% energy from protein) to 0.169 (energy). Similarly, κ values observed in the females ranged from 0.051 (% energy from fat) to 0.202 (protein).

The current paper describes the extent of tracking for a range of behavioural and biological risk factors for CVD, between adolescence (15y) and young adulthood (22y) in 245 males and 231 females from Northern Ireland.

In relation to nutrient intakes, the poor tracking between 15 and 22 years revealed in this study reflects previous findings in this cohort between 12 and 15 years 20. This suggests that individual dietary patterns exhibited at 15 years are unlikely to be predictive of dietary intakes at young adulthood. Intuitively, this lack of tracking is to be expected, as the transition from adolescence to adulthood is characterised by considerable physical, cognitive and psychosocial change. To date, however, there has been little evidence for tracking or otherwise of diet in this age group. Reasonably good dietary tracking has been reported for younger pre-school children 23, but this is not surprising given the high degree of control over diet exerted by parents in this age group. Reasonably high tracking coefficients for diet in the Amsterdam Growth and Health Longitudinal Survey were reported by Kemper et al. 24, but these were between the ages of 13 and 32 years. Although it is difficult to draw direct comparisons between studies due to differences in methodologies, one possibility for the poor tracking reported in the present study might be a particularly high degree of mis-reporting of intake in 15 year-old adolescents. This has been noted previously in relation to this cohort, particularly in relation to 'under-reporting' in 15 year-old females 20. It is also possible that the low κ values obtained for the dietary intakes in the present study indicate that adolescence is, indeed, associated with rapidly changing and erratic patterns of nutrient intake. Adolescents take increasing control of what, when and where they eat and typically consume a greater proportion of their total intake outside the home. Concerns about changing body shape and adiposity may also prompt sudden changes in eating behaviour.

While adolescence is widely regarded to be a time of transition, it could also be argued that young adulthood is an equally important time of change, especially with regard to dietary habits. This is a time when people in this age group are likely to move out of home, go to university, start a family or to encounter other environmental or psychosocial factors that influence food intakes. Thus it is possible that the poor maintenance of ranks that we have observed in this study has arisen simply because we attempted to assess tracking between two very unstable periods of time in the life-cycle.

It is also possible that at least part of the explanation for poor dietary tracking in a cohort of this nature lies in the unsuitability of the diet history method for this purpose. Although the diet history method has shown good validity at the group level in adolescents, it is prone to significant problems of precision at the individual level 15. Moreover, as it assesses perception and memory of usual diet and is susceptible to socially desirable responding 25, it is possible that changes in memory and motivation over time may contribute to poor tracking. Finally, it is entirely feasible that diet simply does not track well between two time points several years apart. Certainly, the data presented in this study suggest that individual dietary patterns reported at 15 years are unlikely to be predictive of energy and nutrient intakes reported at 22 years. It is clear, therefore, that individual subjects cannot be targeted for long-term dietary intervention based solely on data obtained at 15 years of age.

Both physical fitness and physical activity are now accepted as independent risk factors for several chronic diseases. The identification of low levels of fitness and/or activity at an early stage in the lifecourse might, therefore, enable early remedial strategies, provided that a degree of tracking for these risk factors is demonstrable. The results of the present study fail to provide such evidence, with poor tracking being demonstrated for fitness in both sexes, and poor tracking in females and only moderate levels of tracking in males for activity. By and large, these results are in keeping with the handful of other studies which have examined tracking of fitness and activity in this age group 2627. While the mechanisms responsible for this lack of stability between adolescence and early adulthood remain obscure, it is tempting to speculate that similar influences are responsible for the poor tracking of both diet and physical activity/fitness. For example, while much activity during adolescence is organised and school-based, by the time the individual reaches early adulthood, activity is likely to be more a matter of choice. In this respect, it is interesting to note that studies of the tracking of physical activity/fitness in both younger 28 and older 2930 age groups than that of the present study, show generally higher levels of tracking between time points.

In contrast to diet and physical activity/fitness, anthropometric variables relating to body weight and adiposity showed stronger and consistent tracking, particularly in females. Good tracking of BMI from adolescence to young adulthood has been noted in previous studies 31243233, with stronger tracking for females also highlighted. The results of the present study thus confirm the potential utility of identifying adolescents at the age of 15 years who are at risk of persistent obesity, and targeting such adolescents with appropriate long-term lifestyle advice. Our results for diet, physical activity and fitness, however imply greater instability from adolescence to young adulthood, and consequently the need for shorter-term, ameliorative strategies based on regular monitoring of these behaviours and attributes.

The authors declare that they have no competing interests.

CB conceived of the study, participated in its design and co-ordination and co-drafted the manuscript. PR supervised the collection of dietary data and co-drafted the manuscript. AG and GC carried out the statistical analysis. LM participated in the design and co-ordination of the project. JS conceived of the study and participated in its design and co-ordination. All authors read and approved the final manuscript.