Substantial evidence supports the hypothesis that physical inactivity and lack of physical fitness increase the risk of morbidity and mortality from a number of chronic diseases. In adults, a physically active and fit way of life protects against or retards the development of coronary heart disease, stroke, hypertension, obesity, non-insulin-dependent diabetes mellitus, and some cancers. It also appears to reduce depression and anxiety and improve mood, and it is important for the health of muscles, bones, and joints (6).
It is often assumed that physical activity and physical fitness during childhood and adolescence is beneficial for health during adulthood. A conceptual model illustrating possible relationships between physical activity and physical fitness during youth and health during adulthood has been proposed by Blair et al. (4). In this model, physical activity and fitness in youth may affect adult health in three pathways: a) directly, b) through its effect on youth health (which in turn affects adult health), or c) through its effect on adult physical activity and fitness (which in turn affects adult health).
The number of studies to support these different pathways is relatively small. Only longitudinal research in which physical activity, physical fitness, and health are measured repeatedly in the same individual over an extended period of time can provide a definite answer. However, longitudinal studies are relatively scarce. They not only require a large financial effort and much patience, but are also faced with a number of specific problems such as long-term commitment of staff and subjects, obsolescence of techniques, and confounding effects that inevitably occur in repeated measurements (27).
Although there are indicators of beneficial associations in some areas, relatively little evidence unambiguously relates adolescent physical activity or physical fitness to adolescent or adult health or a more favorable adolescent or adult disease-risk profile. It seems more reasonable to assume that adolescent physical activity or physical fitness may have an indirect influence on adult health status by increasing the likelihood of becoming an active adult, which in turn is beneficial for adult health (14).
In epidemiology, the analysis dealing with this phenomenon is called tracking. Tracking refers to 1) the stability of relative rank or position within a group over time or 2) the predictability of a measurement early in life for the value of the same variables later in life (15-17).
The available data are generally consistent in showing moderate tracking of physical activity during childhood and adolescence with more variable but lower levels of tracking across longer intervals within adolescence, from adolescence into young adulthood, and across various ages in adulthood. Longitudinal data for measures of performance- and health-related physical fitness through childhood, adolescence, and adulthood are limited but generally show higher interage correlations and thus somewhat better tracking than for indicators of physical activity (15-17).
Most studies concerning tracking of physical activity and physical fitness from childhood and adolescence into adulthood largely focus on young adulthood, the 20s and early 30s. Tracking of physical activity and physical fitness from adolescence to older ages in adulthood is less studied and generally suggests weak associations. Moreover, data are more available for males than for females (15-17).
The purpose of the present study is to investigate the tracking of physical fitness and physical activity from adolescence (14-18 yr of age) into middle adulthood (37-43 yr of age) in Flemish females.
MATERIALS AND METHODS
This study is a follow-up of a subsample from the Leuven Growth Study of Flemish Girls, a cross-sectional multidisciplinary survey designed to provide information on the growth status and physical fitness of Flemish school girls. In 1979-1980, a large representative sample of elementary and high school girls from 88 schools was tested. Observations for each girl included a physical fitness test battery, anthropometric dimensions, somatotype, skeletal maturity, age at menarche, sports practice, and sociocultural characteristics of the family. A detailed description of the design of the study, the sampling and measurement procedures, and reference data is given elsewhere (22).
Within the scope of the Leuven Longitudinal Study on Lifestyle, Fitness and Health, conducted by the Policy Research Centre Sport, Physical Activity and Health, a subsample of the oldest girls (14-18 yr in 1979-1980) was contacted in 2003-2004, now 37-43 yr of age. A sample of 138 women agreed to participate in the follow-up study (response rate: 32.9%).
The study was approved by the medical and ethical committee of the Katholieke Universiteit Leuven. Prior to participation, the purpose and procedures of the study were explained, and subjects gave their written informed consent. During adolescence parental consent was given in addition.
To assess the representativeness of the study sample, adolescent age, anthropometric, physical fitness, and physical activity values of women participating in the follow-up were compared with those of nonparticipants during adolescence. Because the majority of the variables were not normally distributed, the nonparametric Wilcoxon rank-sum test was used. Results are shown in Table 1. Participants were significantly older (P < 0.01) in adolescence than nonparticipants. This could be due to the fact that because the tracking study was part of a larger study also investigating familial resemblance in physical fitness and physical activity, only women with at least two children older than 10 yr of age were included in the study to have sufficient familial correlations and because children ≤ 10 yr of age are generally unable to correctly perform the different tests. Participants had a significantly lower skinfold suprailiac (P < 0.05) and trunk-extremity skinfold index (P < 0.01) and needed significantly less time for the shuttle run (P < 0.01) compared with nonparticipants. Although statistically significant, the mean differences in age (3 months), skinfold suprailiac (1.1 mm), trunk-extremity index (0.5), and shuttle run (0.36 s) are very small and can be considered irrelevant. For the other variables, no significant differences were found between the groups. Hence, for the variables assessed in this study, the adolescent values of the subsample of 138 women participating in the follow-up study were reasonably representative for the total sample of 14- to 18-yr-old girls tested in 1979-1980.
All anthropometric dimensions were taken by trained staff; subjects were barefoot and in underwear (8). Standing height was measured to the nearest millimeter using a Holtain stadiometer, and body weight was measured to the nearest 0.1 kg using an electronic scale(Seca 841). BMI was calculated as body mass (kg) divided by squared height(m2). Skinfold thicknesses of biceps, triceps, subscapular, suprailiac, and medial calf were measured to the nearest millimeter using a Harpendencalliper.
The sum of four skinfolds (biceps + triceps + suprailiac + supscapular)and the trunk-extremity skinfold index ((scapula + suprailiac)/(triceps +medial calf)) were calculated.
In-field reliability of the anthropometric dimensions in adolescence was evaluated by reexamining about 3% of the subjects. The coefficients of variation, as an expression of the magnitude of the measurement error in function of the mean, were small. For body weight and height the coefficients were < 1%. For skinfolds the coefficients varied between 5 and 10% (12). In adulthood the study design did not allow reexamining a subset of participants to check in-field reliability. However, high quality of measuring techniques during adulthood was aimed at by careful selection of test instructors (M.Sc. in physical education or physiotherapy), thorough training by experienced kinanthropometrists (staff members of the Leuven Growth Study of Flemish Girls in 1979-1980), and regular supervision throughout the entire testing period.
Physical fitness items included both health- and performance-related fitness components. Performance-related fitness items were balance (flamingo balance), static strength (arm pull), explosive strength or power (vertical jump), running speed (50-m shuttle run), and speed of limb movement (plate tapping). Health-related fitness items were flexibility (sit and reach), upper-body muscular endurance(bent-arm hang) and lower-body muscular endurance (leg lifts). All tests were taken as described by Claessens et al. (8). Methods and procedures were kept as similar as possible during both phases of the study. Test-retest reliability of the fitness tests in adolescence ranged from r = 0.61 to 0.94 (12). The same explanation given for the absence of a reliability check on the anthropometric dimensions also applies to the physical fitness tests.
Data on sports participation were collected by a standardized questionnaire during adolescence via a retrospective approach: sports activities during the period of 1 yr before the date of investigation were recalled. Participants were asked to report compulsory physical activity at school as well as sports participation during leisure time. The questionnaire was completed by the girls' parents and cross-checked during an interview with each girl(20). The detailed information on the frequency of sports participation was combined into a global average score of sport practice per week over 1 yr. Reliability and validity of the questionnaire was determined in a previous study (18).
Sports participation of the women was assessed in a different way. In adulthood, subjects were asked to select their three most important sports out of a list of 196 different sports in a computerized questionnaire. For each of these three sports, they were asked to report frequency (from 1 wk·yr−1 to more than 7× wk−1) and duration (from less than 1h·wk−1 to more than 20 h·wk−1).The reported frequency and time of sports participation were combined into a global average score per week.
Descriptive statistics (means and standard deviations) were calculated for all anthropometric characteristics, physical fitness, and physical activity characteristics during adolescence and adulthood. Because the majority of the variables were not normally distributed, nonparametric statistics were applied. Wilcoxon signed rank tests were used to evaluate differences between adolescence and adulthood.
Tracking of anthropometric characteristics, physical fitness, and physical activity was assessed with Spearman's rank order correlations between measurements in adolescence and adulthood.
Tracking of BMI and sports participation was further explored using relative distribution. Subjects were divided into different BMI and physical activity groups (see below) and by means of cross-tabulation; the percentage of subjects remaining in the same group or shifting from one group to another from adolescence to adulthood was calculated. Subjects were divided into a normal-weight and an overweight group on the basis of the BMI. When applying the age- and sex-specific cutoff points for overweight in children and adolescents as described by Cole at al. (9), only 7.4% of the entire sample of 14- to 18-yr-old girls, and only 5.8% of the subsample of 138 girls were categorized as overweight compared with about 12% (88th percentile)in the study of Cole et al. (9). For this reason it was decided that these cutoff points could not be used in the follow-up study. Instead, for the entire sample of 14- to 18-yr-old girls divided into half-year ages, the BMI at 80th percentile was calculated, and these cutoff values were used to place the 138 girls into the normal-weight or overweight group as adolescents. For adults, the widely accepted cutoff point of 25 kg·m−2 for overweight was used. Depending on their reported hours of sports participation, subjects were divided into less active and active groups. In 1979-1980, the compulsory school physical education program in Belgium varied between one and three periods of 50 min·wk−1 depending on the school type. Adolescents reporting less than 3 h of sports per week were considered less active, meaning that they did not do any or only sporadically participated in sports during leisure time in addition to the compulsory physical education program at school. Girls reporting three or more hours of sports participation per week were considered active during leisure time. In adulthood, women reporting less than 1.5 h of sports per week in leisure time were considered less active, and those participating in 1.5 h or more per week were labelled active.
Odds ratios for less activity and overweight in adulthood according to adolescent activity or weight status were calculated.
For all statistical analyses, the SAS 9.1 package was used (2).
Descriptive statistics, differences, and interage correlations between adolescence and adulthood for anthropometric, physical fitness, and physical activity variables are shown in Table 2. All anthropometric variables and indices increased on average significantly from adolescence to adulthood. Except for flamingo balance, mean performances on all motor tests decreased significantly. Sports participation also decreased. Sports participation (r = 0.13, NS) and arm-pull strength (r = 0.20, P <0.05) both showed a low correlation between adolescence and adulthood, whereas correlations for height (r = 0.96) and sit and reach (r = 0.76) were high and significant (P < 0.001). For all other variables, correlations were significant (P < 0.001) and moderate, ranging from 0.34 to 0.56. It is remarkable that correlations for weight, BMI, all skinfolds, and the trunk-extremity skinfold index were virtually identical.
Percentage of subjects remaining in the same BMI group or shifting from one BMI group to another from adolescence to adulthood is shown in Figure 1. In adolescence, 84.1% of the girls had a normal weight and 15.9% were overweight. Of the normal-weight girls in adolescence, b84.5% remained normal weight in adulthood and only 15.5% became overweight. In contrast, 63.6% of overweight adolescent girls remained overweight and 36.4% became normal weight. Active and less active adolescents are compared with regard to level of sports participation in adulthood in Figure 2. During adolescence, 58.5% of the girls were active and 41.5% were less active. Of the active adolescent girls, 54.4% remained active and 45.6% became less active in adulthood. Of the less active adolescent girls, 62.5% remained less active and only 37.5% became active in adulthood. Odds ratios and 95% CI for less activity or overweight in adulthood according to adolescent activity or weight status are presented in Table 3. The odds of being overweight in adulthood are 9.53 times greater in overweight compared with normal-weight adolescents. The odds ratios for overweight in adulthood according to adolescent activity status and for less activity in adulthood according to adolescent activity or weight status vary between 1.34 and 2.18, but are not significant. The CI for the odds ratio of the overweight-overweight pairing is much wider than that of the other pairings. This could be attributed to the higher odds ratio itself and to the fact that only 22 (15.9%) girls were overweight in adolescence compared with 116 (84.1%) girls in the normal-weight group (see Fig. 1).
The purpose of this study was to investigate tracking of physical fitness and physical activity from adolescence into middle adulthood in Flemish females. First, for all variables interage correlations were calculated. Following the criterion of Bloom (5), all anthropometric and physical fitness characteristics (except for flamingo balance, plate tapping, leg lifts, and arm pull) are stable from adolescence to adulthood in Flemish females. On the other hand, sports participation is not a stable characteristic. Over a period of 23 yr, anthropometric and physical fitness variables demonstrate higher levels of stability than physical activity.
It is difficult to compare tracking coefficients across studies because of differences in methodology, age at first observation, and time span between repeated measures. It must also be noted that tracking refers to the relative position within a group over time. When tracking of a variable is high, it does not necessarily mean that absolute values of that variable remain at the same level.
Tracking coefficients found in the present study are comparable with those found in other studies. In Flemish males tracking was high for flexibility (r = 0.68-0.82) and low to moderate (r = 0.22-0.69) for other fitness parameters between adolescence (13-15-18 yr) and adulthood (30-35 yr) (3). The correlation between 18 and 30 yr of age for the BMI was 0.69 (11) and about 0.50 for skinfold thicknesses (13). In contrast, correlations for sports participation from adolescence (12-18 yr) to 35 yr of age were low: 0.14-0.20 (29). In females of the Cardiovascular Risk in Young Finns Study, correlations for a physical activity index between ages 15 and 36 yr and between ages 18 and 39 yr were 0.14 and 0.26, respectively (23).
Three potential explanations for the higher stability of physical fitness compared with physical activity are reasonable. First, physical activity is a behavior associated with more modifiable and volatile factors (e.g., social support, occupation) (24), whereas physical fitness is likely to be associated with more stable factors (e.g., genotype, morphology) (7). Second, the praising of physical activity may change with age-as do opportunities-especially with children in the household (15,17). A third factor concerns the possibly greater measurement error associated with assessment of physical activity by questionnaire compared with the more objective physical fitness tests (28).
To further investigate tracking of BMI and physical activity, the percentage of subjects remaining in the same BMI or physical activity group, or shifting from one group to another from adolescence to adulthood, was identified. In Flemish females it is clear that weight status in adolescence is indicative of weight status in middle adulthood: 84.5 and 63.6%, respectively, remained in the normal-weight and overweight group.
The present study confirms what has been observed in previous studies. For 6- to 18-yr-old females of the Fels Longitudinal Study at the 75th percentile on the CDC BMI-for-age growth charts, a probability of 40-59% of being overweight at 35 yr of age was reported. At the 85th percentile, the probability increased to 60-79.9% for 10- to 18-yr-old girls, and at the 95th percentile girls older than 10 yr of age had a ≥ of approximately 80% of becoming overweight adults (10). In Danish females, 71% remained in the upper quintile for the BMI from adolescence to young adulthood (1).
In the present study, less activity appears to track better than activity: 62.5% of less active adolescents remained less active, whereas only a small majority of 54.4% remained active.
Results of the stability of physical activity are somewhat more variable among studies of European samples. Similar results were obtained in Flemish males (29). Of inactive 17-yr-old boys, 78% remained inactive at 30 yr, whereas boys who were very active in sports during adolescence only had a slightly higher chance of belonging to the moderately active (38%) or active group (34%) as adults. On the other hand, among Finnish men who had taken physical exercise at least once a week during leisure time at 14 yr of age, the majority (71%) continued to do so at 24 yr of age, whereas 55% participating in no exercise or less than once a week in their youth remained in the same group in adulthood (19). In the Amsterdam Growth and Health Longitudinal Study, 42% of adolescent girls (13-17 yr) in the highest-risk group (under 25th percentile) for physical activity remained in the same risk group at 27 yr of age (26).
These variable results can be attributed, in part, to the different cut points used to categorize subjects into risk groups. In the present study, except for adolescent BMI, the groups were objectively defined according to absolute values, whereas in most other studies, groups were arbitrarily defined based on the distribution of values. Because results of a tracking analysis greatly depend on the arbitrary choice for division of subgroups, defining risk groups according to the distribution of values is suitable only in situations where no objective risk values are available (25).
Finally, odds ratios were calculated for less activity or overweight in adulthood according to adolescent activity or weight status. Overweight adolescent girls showed a 9.5-fold higher risk of becoming overweight adults.
This is consistent with the literature. In Flemish males, overweight 13- to 17-yr-old boys had a five to seven times higher risk of becoming an overweight at 40 yr (11). In the Fels Longitudinal Study, the odds of overweight at 35 yr for girls 15-18 yr at the 75th percentile for BMI varied between 2.9 and 7.3 times those for girls with BMI values at the 50th percentile (10).
The current study has several important strengths compared with other studies. It is a longitudinal study of females in which both tracking of physical fitness and physical activity from adolescence to middle adulthood were investigated. The literature on females is relatively scarce, and most studies focus on young adulthood. The longitudinal approach also circumvents recall biases, a major issue in many retrospective studies.
This study also has several limitations, especially concerning the definition and assessment of physical activity. Whereas physical activity is a complex behavior consisting of several dimensions (e.g., occupation, transportation, leisure time), only sports participation was taken into account because of lack of available data for other components of physical activity during the first phase of the study. Furthermore, the physical activity questionnaire used during adolescence differed from that in adulthood. However, this was also the case in the Cardiovascular Risk in Young Finns Study, where it did not seem to influence the results (23). Finally, the use of questionnaires may represent a weakness compared with more objective measures of physical activity such as heart rate monitoring or motion sensors (21).
In summary, results of this study show lower stability of physical activity compared with physical fitness from adolescence to adulthood. Weight status during adolescence is reasonably indicative of adult weight status, and a pattern of less activity rather than activity tends to be continued from youth into adulthood. This leads to the recommendation that during adolescence, effective strategies should be implemented to encourage physical activity and to prevent overweight. In addition, during life constant efforts should be made to keep people active, and efforts need to be made to reactivate those who have shifted from an active to a more sedentary lifestyle.
The Leuven Growth Study of Flemish Girls was supported by grants from the Ministry of Labor and Empoyment, the Belgian Olympic and Interfederal Committee, and Siemens n.v.
The Policy Research Centre Sport, Physical Activity and Health is supported by the Flemish Government.
1. Andersen, L.B. Tracking of risk factors for coronary heart disease from adolescence to young adulthood with special emphasis on physical activity and fitness. Dan. Med. Bull.
2. Base SAS 9.1 Procedure Guide. Cary, NC: SAS Institute Inc., 2004.
3. Beunen, G., M. Ostyn, J. Simons, et al. Development and tracking in fitness components: Leuven Longitudinal Study on Lifestyle, Fitness And Health. Int. J. Sports Med.
4. Blair, S. N., D. G. Clark, K. J. Cureton, and K. E. Powell. Exercise and fitness in childhood: implications for a lifetime of health. In: Perspectives in Exercise Science and Sports Medicine. Volume 2. Youth, Exercise and Sport
, C. V. Gisolfi and D. R. Lamb (Eds). Carmel,IN: Benchmark Press, 1989, pp. 401-430.
5. Bloom, B. S. Stability and Change in Human Characteristics
, New York: Wiley, 1964.
6. Bouchard, C., R. J. Shepard, and T. Stephens. PhysicalActivity, Fitness and Health: International Proceedings and Consensus Statement.
Champaign,IL: Human Kinetics, 1994.
7. Bouchard, C., R. M. Malina, and L. Pérusse. Genetics of Fitness and Physical Performance.
Champaign, IL: Human Kinetics, 1997.
8. Claessens, A.L. M., B. VandenEynde, R. Renson,and D. Van Gerven. The description of tests and measurements. In: Growth and Fitness of Flemish Girls: The Leuven Growth Study. HKP Sport Science Monograph Series. Volume 3
, J. Simons, G. P. Beunen, R. Renson, A.L. M. Claessens, B. Vanreusel, and J. A. V. Lefevre (Eds.). Champaign, IL: Human Kinetics Books, 1990, pp. 21-39.
9. Cole, T. J., M. C. Bellizzi, M. Flegal, and W. H. Dietz. Establishing a standard definition for child overweight and obesity worldwide: International survey. BMJ
10. Guo, S. S., C. C. Chumlea, and A. F. Roche. Predicting overweight and obesity in adulthood from body mass index values in childhood and adolescence. Am. J. Clin. Nutr.
11. Hulens, M., G. Beunen, A. L. Claessens, et al. Trends in BMI among Belgian children, adolescents and adults. Int. J. Obes.
12. Lefevre, J. A. V., G. P. Beunen, and R. Wellens. Data input and quality control. In: Growth and Fitness of Flemish Girls: The Leuven Growth Study. HKP Sport Science Monograph Series
, Volume 3, J. Simons, G. P. Beunen, R. Renson, A. L. M. Claessens, B. Vanreusel, and J. A. V. Lefevre (Eds.). Champaign, IL: Human Kinetics Books, 1990, pp. 47-56.
13. Lefevre, J., G. Beunen, A. Claessens, et al. Tracking at the extremes in health- and performance related fitness from adolescence through adulthood. I. V. Kinanthropometry, W. Duquet, and J. A. P. Day (Eds.). London: E. & F.N. Spoon, 1993, pp. 249-255.
14. Lefevre, J., R. Philippaerts, K. Delvaux, et al. Relation between cardiovascular risk factors at adult age, and physical activity during youth and adulthood: The Leuven Longitudinal Study on Lifestyle, Fitness and Health. Int. J. Sports Med.
15. Malina, R. M. Adherence to physical activity from childhood to adulthood: A perspective from tracking studies. QUEST
16. Malina, R. M. Physical activity and fitness: Pathways from childhood to adulthood. Am. J. Hum. Biology
17. Malina, R. M. Tracking of physical activity across the lifespan. Pres. Counc. Fitness Sports Res. Dig.
18. Philippaerts, R., and J. Lefevre. Reliability and validity of three physical activity questionnaires in Flemish males. Am. J. Epidemiol.
19. Pietilä, A.- M., M. Hentinen, and A. Myhrman. The health of Northern Finnish men in adolescence and adulthood. Int. J. Nurs. Stud.
20. Renson, R., and B. Vanreusel. The sociocultural and physical activity inventories. In: Growth and Fitness of Flemish Girls: The Leuven Growth Study.
>HKP Sport Science Monograph Series. Volume 3, J. Simons, G. P. Beunen, R. Renson, A. L. M. Claessens, B. Vanreusel, and J. A. V. Lefevre (Eds.). Champaign, IL: Human Kinetics Books, 1990, pp. 41-45.
21. Sallis, J. F., and B. E. Saelens. Assessment of physical activity by self-report: Status, limitations and future directions. Res. Q. Exerc. Sport
22. Simons, J., G. P. Beunen, R. Renson, A. L. M. Claessens, B. Vanreusel, and J. A. V. Lefevre. Growth and Fitness of Flemish Girls. The Leuven Growth Study.
HKP Sport Science Monograph Series, Volume 3. Champaign, IL: Human Kinetics Books, 1990.
23. Telama, R., X. Yang, J. Viikari, I. Välimäki, O. Wanne, and O. Raitakari. Physical activity from childhood to adulthood: A 21-year tracking study. Am. J. Prev. Med.
24. Trost, S. G., N. Owen, A. E. Bauman, and J. F. Sallis. Correlates of adults' participation in physical activity: review and update. Med. Sci. Sports Exerc.
25. Twisk, J., H. C. G. Kemper, and G. J. Mellenbergh. Mathematical and analytical aspects of tracking. Epidemiol. Rev.
26. Twisk, J., H. C. G. Kemper, and J. Snel. Tracking of cardiovascular risk factors in relation to lifestyle. In: The Amsterdam Growth Study. A Longitudinal Analysis of Health, Fitness, and Lifestyle.
HK Sport Science Monograph Series, Volume 6, H. C. G. Kemper (Ed.). Champaign, IL: Human Kinetics, 1995, pp. 203-224.
27. van Mechelen, W., and G. J. Mellenbergh. Problems and solutions in longitudinal research: From theory to practice. Int. J. Sports Med.
28. Vanhees, L., J. Lefevre, R. Philippaerts, et al. How to assess physical activity? How to assess physical fitness? Eur. J. Cardiovasc. Prev. Rehab.
29. Vanreusel, B., R. Renson, G. Beunen, et al. A longitudinal study of youth sport participation and adherence to sport in adulthood. Int. Rev. Sociol. Sport