Medicine & Science in Sports & Exercise:
Basic Sciences: Epidemiology
Adolescent Correlates of Adult Physical Activity: A 26-year Follow-up
BEUNEN, GASTON P.1; LEFEVRE, JOHAN1; PHILIPPAERTS, RENAAT M.4; DELVAUX, KATRIEN2; THOMIS, MARTINE1; CLAESSENS, ALBRECHT L.1; VANREUSEL, BART1; LYSENS, ROELAND2; EYNDE, BAVO VANDEN3; RENSON, ROLAND1
1Department of Sport and Movement Sciences, Faculty of Kinesiology and Rehabilitation Sciences, 2Physical Medicine and Rehabilitation, Faculty of Medicine, 3Department of Kinesiology, Faculty of Kinesiology and Rehabilitation Sciences, K. U. Leuven, Leuven, BELGIUM; 4Department of Movement and Sports Sciences, U. Gent, Gent, BELGIUM
Address for correspondence: Gaston P. Beunen, Ph.D., Department of Sport and Movement Sciences, Faculty of Kinesiology and Rehabilitation Sciences, K. U. Leuven, Belgium, Tervuursevest 101, B-3001 Leuven, Belgium; E-mail: firstname.lastname@example.org.
Submitted for publication December 2003.
Accepted for publication July 2004.
Purpose: It is hypothesized that adolescent physical activity, fitness, anthropometric dimensions, fatness, biological maturity, and family characteristics contribute to the variation in physical activity at 40 yr of age, and that these associations vary with age.
Methods: Subjects were 166 males followed from 1969 to 1996, between the ages of 14 and 40 yr from the Leuven Longitudinal Study on Lifestyle, Fitness and Health. Sports participation, fitness, anthropometric dimensions, fatness, and biological maturity were observed during the growth period. Also, sociocultural characteristics of the family were examined. The work, leisure time, and sport activity index of the Baecke Questionnaire and activity counts of a triaxial accelerometer were used as outcome variables at 40 yr.
Results: When upper and lower activity groups (quintiles) at 40 yr were contrasted, moderate associations were found (R2c varied between 0.1419 and 0.3736). No or low associations were found with the leisure time index. Body dimensions, fitness scores, sports practice, and family characteristics contributed to the explained variance in work, sport index, and activity counts. Multiple correlations were low (R2 = 0.037–0.085) for the work and leisure time activities, and were somewhat higher (R2 = 0.06–0.156) for the sport index and the activity counts in the total sample.
Conclusion: Adolescent somatic dimensions, fitness, sports participation, parental sociocultural characteristics, and sport participation contributed to a small-to-moderate extent to the contrast between high and low active adults.
Although there is accumulating evidence that physical inactivity is a serious health burden, little is known about the adolescent antecedents of adult physical activity behavior (11,13). Different pathways have been described that relate childhood and adolescent physical activity, fitness, and health to adult physical activity, fitness, and health (4,13). However, few prospective or longitudinal studies have been conducted during which indicators of health, fitness, and physical activity have been observed during childhood or adolescence, with extension of these observations into adulthood (13). These pathways are: 1) does adolescent physical activity contribute to better health (reduced risk factors) during adolescence, 2) does physical activity track over time, 3) does physical activity during adolescence result in better adult health, and 4) does adolescent health (risk factors) contribute to adult health?
Physically active American adult males, aged 23–25 yr, had better motor fitness scores as children (10–11 yr) and as adolescents (15–18 yr) than inactive young adults (8). In Swedish adults, maximum oxygen uptake (;V̇O2max) and percentage of Type I muscle fibers at 16 yr accounted for about 31 and 24% of the variance in leisure time physical activity at 27 yr of age in women and men, respectively (9). In the Leuven Longitudinal Study on Lifestyle, Fitness and Health (LLSLFH), a 27-yr follow-up study of Belgian males, adolescent fitness characteristics showed consistent associations with an adult sport index as measured with the Baecke Questionnaire (1). Explained variance is lowest (3%) for the longest interval, 13–40 yr, but is already considerable (17%) for the 15- to 40-yr interval, and is highest (23%) for the 18- to 40-yr interval. Other indicators of physical activity during work or leisure time, or counts of accelerometers, were not significantly associated with adolescent fitness (3). Surprisingly, several fitness items, for instance, strength and muscular endurance showed negative correlations with adult physical activity scores. Higher associations were found when extreme physical activity quintiles were contrasted (3). Similarly, in adult Dutchmen, physical activity at 33 yr was weakly associated with fitness levels observed during adolescence, and only maximal aerobic power showed significant associations (10). Tracking in physical activity, using a variety of indicators, is low from adolescence to adulthood (7,10,13) but tend to be higher in Belgian males for inactivity scores (23).
In all previous studies, only physical activity and fitness scores have been correlated with adult physical activity, and it can be questioned whether indicators of growth, maturation, and family characteristics can explain a part of the variation in adult physical activity levels (3,22). In this article, it is hypothesized that in Belgian males, adolescent indicators of physical activity, physical fitness, anthropometric dimensions, fatness, biological maturation, as well as family characteristics contribute to the variance in indicators of physical activity examined at 40 yr of age. Furthermore, it is expected that the associations decrease with increasing time interval.
MATERIALS AND METHODS
Study design and selection of outcome and predictor variables.
To test the hypotheses formulated above, data from the LLSLFH were used (see Fig. 1). Different indicators of physical activity were used as outcome variables at 40 yr (16). Physical activity was assessed using different indices of the Baecke Questionnaire. The reliability and validity of this questionnaire have been documented in a subsample of adults of the same age level (16,18). Furthermore, total activity counts measured by accelerometry were used as a more objective indicator of physical activity (5,6). It was decided to select correlates of physical activity that have been previously investigated as predictor variables (22). These included adolescent sports participation, parental socioeconomic status (education, profession, degree of urbanization), and parental sports participation. Following Malina (12,14), adolescent physical characteristics were also included, namely body dimensions, maturity indicators, and fitness scores.
Because a large number of potential predictor variables were available, a selection was made based on the zero-order correlations (P ≤ 0.05) between the adolescent characteristics and the adult physical activity outcomes. For consistency across age-groups and physical activity indicators, some characteristics that showed borderline significant correlations (P ≤ 0.10) were also included (Table 1). None of the bivariate regressions showed a significant deviation from linearity (nonsignificant quadric or cubic trend).
The subjects were from the Leuven Growth Study of Belgian Boys (2) who were later on followed, at the ages of 30, 35, and 40 yr, in the LLSLFH (Fig. 1). In the Leuven Growth Study (phase 1) a total of 8963 Belgian boys from Flanders, Brussels, and Wallonia (ages between 13 and 18 yr) were observed between 1969 and 1974. Of this sample, 588 boys were followed at annual intervals from 13 to 18 yr of age. The Leuven Growth Study of Belgian Boys was a combination of a cross sectional survey and a longitudinal study. Each year between 1969 and 1974, a representative sample of one grade of the secondary school was studied, the first grade in 1969 and the sixth grade in 1974. Because the same schools were included each year in the sample, a considerable number of boys were observed at several occasions of which 588 boys were followed at annual intervals from 13 to 18 yr.
Data were collected on somatic dimensions, somatotype, body composition, biological maturation, health- and performance-related fitness components, sports activities, and sociocultural correlates. Of the 588 boys followed longitudinally, only the Flemish speaking (N = 441) were selected for further follow-up. In 1986 (phase 2), the LLSLFH started, and 174 of those 441 were measured in this first adult follow-up phase. In 1991 (phase 3), 176 subjects, aged 35 yr, took part, and in 1996 (phase 4), 166 subjects participated. For the present study, data were included from 166 subjects followed from 1969 to 1996. Longitudinal data between 14 and 40 yr were used (see Fig. 1). The major reported reason for drop out is lack of time to continue participation. For the follow-up phases, subjects were invited to the laboratory at the K. U. Leuven for a full-day test program. Written informed consent was given by all subjects at all phases; during adolescence, parental consent was given in addition.
At age 40 yr, two indicators of physical activity were used as the outcome variables: self-reported activity indices from the Baecke Questionnaire (1), and a more objective activity counter using triaxial accelerometry. Both indicators showed sufficient validity compared with the doubly labeled water method (5,16,18). The Baecke Questionnaire was administered in an interviewer-assisted format. Responses on 16 questions were scored on a five-point scale with descriptors ranging from never (1 point) to always (5 points), with the exception of the questions of the main occupation and the types of the two most important sports played during the previous year. Three indices reflecting activities during work (work index), during sport activities (sport index), and during leisure time excluding sport activities (leisure time index) were derived from the different questions. The work index was the average score on questions 1 to 8, scored on a five-point scale. The sport score was calculated from a combination of intensity, amount of time per week, and the proportion of the year in which the sports were played. The average score on questions 9–12 and questions 13–16 resulted, respectively, in the sport and leisure time index. Triaxial accelerometry using the Tracmor (5,6) was measured during four consecutive days. The total activity counts were used in the analyses.
During adolescence a questionnaire was used to assess the sports participation of the boys (15). Sports activities (type of sport and time spent in sports activities) in structured (private or school sport club) and nonstructured (with friends, family, or alone) contexts were recorded. For this article, time spent in sports activities (min·wk−1) was used as an indicator of PA during adolescence.
During the Leuven Growth Study of Belgian Boys, a fitness test battery was administered at each occasion. This test battery was constructed based on reliability and validity studies conducted in the same age categories (15). The battery included components of performance- and health-related physical fitness. For indicators of performance-related fitness, static strength was measured with the arm-pull test, explosive strength or power with the vertical jump, running speed with 50-m shuttle run, and speed of limb movement with plate tapping. For health-related fitness, the following dimensions were included: flexibility (sit and reach), upper-body muscular endurance (bent arm hang), lower-body muscular endurance (leg lifts), and cardiorespiratory fitness (pulse recovery after a step test). A detailed description of these tests and procedures is given by Beunen et al. (2) and Ostyn et al. (15).
Anthropometric dimensions, adiposity, and skeletal maturity.
Lengths, breadths, circumferences and skinfolds were taken by trained anthropometrists (21). Stature, body mass, shoulder width, arm circumference, calf circumference, and sum of skinfolds (triceps, subscapular, supra-iliac, and calf) were selected for the present analyses. Also the trunk/extremity ratio ((subscapular + supra-iliac)/(triceps + calf)) was used as an indicator of subcutaneous fat distribution. Furthermore, the characteristics of the adolescent growth spurt (age at peak velocity and peak velocity) in stature and body mass were calculated (2), and skeletal maturity was assessed with the Tanner-Whitehouse method [TWII-radius-ulna and short bones (21)].
Sociocultural and demographic factors.
A standardized questionnaire was designed to investigate the sociocultural and demographic factors of the family in which the boys were raised. Also, the sports practice (in structured and nonstructured contexts) of the father and mother were investigated (15). For this study, parental professional status and education, degree of urbanization of the dwelling place, and sports practice of the parents were selected. For each of these sociocultural characteristics, a hierarchical scale was developed for the Belgian population (15,20). It has been shown that these factors are associated with physical activity and fitness during the adolescent years (20) and are also considered as correlates of adult physical activity (22). For these analyses, the sociocultural and demographic factors reported at the last observation, about 18 yr, were included as predictor variables at each time point during adolescence.
To study the associations between adolescent characteristics and adult physical activity, two approaches were used: 1) stepwise multiple regression analysis (F to entry, alpha = 0.15) was used to examine the associations between the different indicators of physical activity, that is, Baecke work index, Baecke sport index, Baecke leisure time index, and activity counts (Tracmor) at 40 yr, and the adolescent characteristics. A selection of predictors was made as explained under study design and selection of predictors and outcome variables. And 2) upper (upper quintile) and lower (lower quintile) adult (at 40 yr) physical activity groups were contrasted for the adolescent characteristics. Because tracking at the extremes is considerably higher than in the overall sample (23), it seemed necessary to examine contrasts in upper and lower groups. Groups were formed based on the different activity indices in adulthood by taking the upper and lower quintiles. Stepwise discriminant analysis (F to entry, alpha = 0.15) was applied to identify the predictor variables. The same characteristics selected for the multiple regression analysis were also introduced in the discriminant analyses. Both the active and inactive group comprised between 22 and 36 subjects, depending on the age level and activity index.
To obtain information about possible age-related changes in the associations between adult (40 yr) physical activity and adolescent fitness, three different age levels were considered in both analyses: 14, 16, and 18 yr. All calculations were made using SAS procedures (SAS Institute, Cary, NC).
The average somatic and fitness characteristics of the adolescent boys correspond quite closely to those observed in the general Belgian population (2,15). Furthermore, the adult physical activity scores do not deviate significantly from those observed in an independent sample of male adults aged 30–40 yr (16).
Table 1 summarizes the zero-order correlations between the adolescent characteristics and the adult physical activity indices. Correlations in absolute value vary between r = 0.11 and 0.23, indicating low associations. At 14 yr, body dimensions and fitness characteristics were correlated with physical activity at 40 yr. The Baecke sport index showed the highest number of significant (P < 0.05) associations. At 16 and 18 yr sports participation of the boys was also a significant predictor (P < 0.05) for most activity indices. Also, at these age levels, body dimensions, fitness characteristics, and sociocultural and demographic factors showed small (P < 0.15) associations. Surprisingly, several body dimensions and fitness scores were negatively correlated with adult physical activity.
When the Baecke work index or Baecke leisure time index, observed at 40 yr, was used as outcome variables, multiple correlations with characteristics observed during adolescence were low (between 3.72% and 8.46% of explained variance) (Table 2). Body dimensions, skeletal maturity, pulse recovery, sports participation, and education or profession of the father contributed to the explained variance. The Baecke sport index and the Tracmor counts correlated somewhat higher (between 6.17% and 15.58% of explained variance). Again, body dimensions, fitness scores, sports participation, and sociocultural factors contributed significantly to the explained variance. The work and sport index were, at each age level, negatively associated (negative standardized regression coefficients, see Table 2) with body dimensions (stature or shoulder width) and with some fitness characteristics (arm pull and bent arm hang). This indicated that smaller body dimensions and lower fitness levels during adolescence were characteristics of more active adults. No such negative associations were found between anthropometric dimensions or fitness scores and the leisure time index or the Tracmor counts. The negative associations for pulse recovery were in the expected direction because better pulse recovery (lower heart rates) was associated with higher adult activity scores.
Because the explained variance of the physical activity outcome variables is rather low, and considering that tracking at the extremes is higher (23), it was decided to contrast upper and lower (quintiles) activity groups (Table 3). Associations were moderate (squared canonical correlations vary between R2c = 0.142 and 0.374) for the work index, sport index, and the Tracmor counts. Associations were lowest for the longest period, 14–40 yr. Body dimensions, fitness characteristics, sports practice, and sociocultural factors were significant predictors. The difference between the upper and lower activity groups in the adolescent predictors was expressed in SD units (Table 3). For the work and sport index, and the Tracmor counts, these differences varied between 0.34 and 0.90, and most (16 of the 22) were above 0.50 SD units. The leisure time index was only marginally associated (4–6%) with the adolescent characteristics considered herein. Noteworthy again was the negative association between adult physical activity indices (work, sport, and leisure time indices) derived from the Baecke Questionnaire, with body dimensions (stature and sum of skinfolds) and some fitness items (arm pull and bent arm hang).
This study indicates low (R2 varies between 0.04 and 0.16) associations between adult (40 yr) physical activity indicators and adolescent body dimensions, fitness items, sports practice, sociocultural, and demographic factors. The associations are higher (R2c varies between 0.04 and 0.37) when extreme physical activity groups are contrasted. Somatic dimensions, fitness characteristics, sports participation, and sociocultural-demographic factors all contribute to the differentiation between the adult upper and lower activity groups. The differentiation is lower when the period from 14 to 40 yr is considered (compared with 16–40 yr or 18–40 yr). Furthermore, the differentiation varies considerably when different activity indicators are considered. The Tracmor counts showed the highest differentiation followed by the sport and work index of the Baecke Questionnaire.
When such long-term prospective studies are conducted, it is important to verify the subjects’ representativeness of the sample. The subjects of the LLSLFH are those who completed their secondary school education in 6 yr and were willing to cooperate in the later follow-up, and therefore selective in educational success and motivation. But, at 18 yr, none of the characteristics under consideration (somatic dimensions, fitness, sports participation, sociocultural characteristics of the family) of those who volunteered to participate in further follow-up (phase 2, N = 174) deviated significantly from the total longitudinal sample (N = 588). Of interest is to determine whether the subjects of the longitudinal sample are selective with respect to the characteristics under consideration. At 18 yr, the means of the longitudinal sample of 166 subjects correspond quite closely (means are situated between percentile 43 and 55 of the reference group) to those of the Belgian reference group (15) for the anthropometric dimensions, but for the fitness characteristics, the longitudinal sample has better average results (means between percentile 52 and 68) than the reference group. This longitudinal sample is on average somewhat better performing in physical fitness and higher in socioeconomic class (17).
Importantly, the interindividual variability of the subjects of the longitudinal sample is very comparable to those observed in the total sample. Finally, the activity scores calculated from the Baecke Questionnaire are highly comparable to those observed in an independent sample of males aged 30–40 yr (16). The activity score in males of this sample is lower than that for workmen only for the Work index (16). Exploration of the original data showed that in the low-active subjects, based on their sport index, all subjects, except one, reported not to be involved in sports activity during the past year. About 75% of the high active group (upper quintile of the Baecke sport index) reported participation in two sports, with about 3 h·wk−1 for the main sport activity and 1.5 h·wk−1 for the second sport activity. Although this longitudinal sample is not a representative sample of the Flemish or Belgian population, the variability in the characteristics observed herein corresponds quite closely to those of a representative sample. The activity scores are furthermore similar to those observed in an independent sample of male adults, which leads to the conclusion that the findings can be generalized.
One of the striking findings is that static strength (arm pull) and upper-body muscular endurance (bent arm hang) are negatively associated with the adult sport index. This is in agreement with what was previously reported for a sample of 109 subjects from the LLSLHF (3). It was then speculated that confounding factors such as size, body composition, and biological maturation modify these associations. The present analysis confirms this hypothesis because stature correlates negatively with activity indicators; furthermore, strength and muscular endurance are associated with stature and body mass. When the activity groups are contrasted in the discriminant analysis this becomes clear; mean statures of the inactive groups are systematically taller than these of the active groups: for the sport index at 14 yr of age, 160.7 (9.0) cm versus 155.2 (7.5) cm for the inactive and active group, and at 16 yr of age, 173.4 (7.8) cm versus 169.0 (7.2) cm for the inactive and active group, respectively. At 40 yr, however, there are no differences in body dimensions between the active and inactive using the Baecke sport index, but the active outperform the inactive on upper body muscular endurance (19). These findings are also confirmed by the positive association between age at peak height velocity and the sport index. A higher sport index at 40 yr is associated with a later age at peak height velocity. Active adult men (sport index) are thus characterized by later biological maturity and smaller body size, resulting in less static strength and upper-body muscular endurance. Later on, active adult men catch up and perform equally well or better than the nonactive (17).
Adult physical activity indicators are most frequently associated with isometric strength (arm pull), muscular endurance (bent arm hang), pulse recovery, stature, or another body dimension. At 16 and 18 yr, sports participation is also significantly associated with adult physical activity indicators. In their review article, Trost et al. (22) also noted that past physical activity is a correlate of adult physical activity levels. However, between 14 and 40 yr, the past sports practice does not contribute to the physical activity behavior. This reflects the lack of tracking in physical activity indicators over such long periods (7,10,13,23).
It is of interest to note that the magnitude of the associations varies with the activity indicators considered. This reflects what has been demonstrated before, namely that habitual physical activity is composed of three dimensions: physical activity during work, sports activities, and leisure time (16). These indices have been used in the present analyses.
The higher degree of urbanization of the dwelling place, the higher SES of the father and the activity levels of one or both of the parents are associated with higher adult physical activity levels of their male offspring (difference scores vary between 0.37 and 0.90 SD units). This reflects the impact of previous rearing styles and combined genetic and environmental conditions. Trost et al. (22) found only weak or mixed evidence for past family influences on adult physical activity. Note also that the associations between sociocultural background variables and adult physical activity indicators are as high as the associations between the sociocultural characteristics and sports participation at 18 yr (19).
These studies add to our knowledge about past fitness, physical activity, biological indicators, and sociocultural factors as correlates of adult physical activity behaviors. As previously demonstrated (3,8,9), some associations exist, but the associations vary considerably between studies. Most likely, the correlates become smaller when increasing age intervals are considered, which explains why Dennison et al. (8) and Glenmark et al. (9) found higher associations. Also, associations vary with the physical activity indicators considered.
Although a variety of indicators of physical activity have been used, correlations between adolescent and adult physical activity are low and do not permit predictions of adult physical activity. The correlations between sports participation during adolescence and the Baecke sport index at 40 yr fall within the range of inter-age correlations reported previously (12). Malina (12,14) wonders why tracking in physical activity is only moderate and tend to decrease within adolescence and from adolescence to adulthood. He argues that factors that influence physical activity during adolescence and adulthood, and the transition from adolescence to adulthood are multivariate in nature and are not included in most studies as covariates or correlates. Furthermore, indicators of growth, maturation, and fitness are not included in most studies (12,14). This was the major reason to include such indicators in the present analysis. Our data show that size, maturity, and fitness levels during adolescence are, at least to a small extent, associated with activity levels later in life.
The discrepancies between this study and those previously reported (3,8,9,10) also indicate that more research is needed in this area. Because long-term prospective studies are scarce, retrospective designs probably could add to our knowledge. These studies should be based on existing data banks of previous studies about physical fitness and physical activity of children and adolescents who are then reexamined at adult ages.
Studying the correlates or determinants of physical activity is an important prerequisite for designing relevant policies, and effective prevention or intervention programs. Because adolescent sports participation, fitness, fatness, and sports participation of the parents are all amenable to change and can be altered with adequate programs, this should be the focus of experimental studies. The results of the discriminant analysis between the upper and lower activity groups reported herein can help to the design of programs that intend to influence these characteristics. Subsequently the effect of such programs can be tested (e.g., in the school setting), and based on these findings, relevant policies can be developed, tested, and implemented.
In conclusion, adolescent somatic dimensions, sports participation, fitness scores, and parental sociocultural characteristics and sports participation contribute to a small-to-moderate extent to the contrast between high and low active adults (40 yr).
Contract grant sponsors for the Leuven Growth Study of Belgian Boys: The Administration of Sport, Physical Education and Open Air Activities of the Ministry of Nederlandse Cultuur and the Ministry of Culture Française, the Administration of Social Medicine of the Ministry of Public Health, and the Foundation for Medical Scientific Research. Contract grant sponsor for the Leuven Longitudinal Study on Lifestyle, Fitness and Health: National Scientific Fund; contract grant number: 3.0188.96; contract grant sponsor: ABB Assurance Company.
1. Baecke, J. A. P., J. Burema, and J. E. R. Frijters. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am. J. Clin. Nutr
. 36:936–942, 1982.
Beunen, G., R. M. Malina, M. A. van‘t Hof, et al. Adolescent Growth and Motor Performance: A Longitudinal Study of Belgian Boys
. Champaign, IL: Human Kinetics, 1988, pp. 1–102.
3. BEUNEN, G. P., R. M. PHILIPPAERTS, K. DELVAUX, et al. Adolescent performance and adult physical activity in Flemish males. Am. J. Hum. Biol
. 13:173–179, 2001.
4. Blair, S. N., D. G. Cark, K. J. Cureton, and K. E. Powell. Exercise and fitness in childhood: implications for a life-time of health. In: Perspectives in Exercise Science and Sports Medicine. Vol. 2: Youth, Exercise, and Sport
, C. V. Gisolfi and D. R. Lamb (Eds.). Indianapolis: Benchmark, 1989, pp. 401–422.
5. Bouten, C. V. C., W. P. H. G. Verboeket-Van De Venne, K. R. Westerterp, M. Verduin, and J. D. Jansen. Daily physical activity assessment: comparison between movement registration and doubly labelled water. J. Appl. Physiol
. 81:1019 –1026, 1996.
6. Bouten, C. V. C., K. R. Westerterp, M. Verduin W. P. H. G. Verboeket-Van De Venne, and J. D. Jansen. Assessment of energy expenditure for physical activity during a triaxial accelerometer. Med. Sci. Sports Exerc
. 26:1516 –1523, 1994.
7. Campbell, P. T., P. T. Katzmarzyk, R. M. Malina, D. C. Rao, L. Pérusse, and C. Bouchard. Prediction of physical activity and physical work capacity (PWC 150) in young adulthood from childhood and adolescence with consideration of parental measures. Am. J. Hum. Biol
. 13:190 –196, 2001.
8. Dennison, B. A., A. H. Straus, E. D. Mellits, and E. Charney. Childhood physical fitness tests: predictor of adult physical activity levels? Pediatrics
82:324 –330, 1998.
9. Glenmark, B., G. Hedberg, and E. Jansson. Prediction of physical activity level in adulthood by physical performance and physical activity in adolescence: an 11-year follow-up study. Eur. J. Appl. Physiol
. 69:530 –538, 1994.
10. Kemper, H. C. G., W. De Vente, W. van Mechelen, and J. W. Twisk. Adolescent motor skill and performance. Is physical activity in adolescence related to adult physical fitness. Am. J. Hum. Biol
. 13:180 –189, 2001.
11. Leonard, W. R. Assessing the influence of physical activity on health and fitness. Am. J. Hum. Biol
. 13:159 –161, 2001.
12. Malina, R. M. Adherence to physical activity from childhood to adulthood: a perspective from tracking studies. Quest
53:346 –355, 2001.
13. Malina, R. M. Physical activity and fitness: pathways from childhood to adulthood. Am. J. Hum. Biol
. 13:162–172, 2001.
14. Malina, R. M. Tracking of physical activity across the lifespan. Pres. Counc. Phys. Fitn. Sport Res. Dig
. 3(14):1– 8, 2001.
15. Ostyn, M., J. Simons, G. Beunen, R. Renson, and D. Van Gerven. Somatic and Motor Development of Belgian Secondary Schoolboys. Norms and Standards
. Leuven: Leuven University Press, 1980, pp. 1–158.
16. Philippaerts, R. M., and J. Lefevre. Reliability and validity of three physical activity questionnaires in Flemish males. Am. J. Epidemiol
. 147:982–990, 1998.
17. Philippaerts, R. M., J. Lefevre, K. Delvaux, et al. Associations between daily physical activity and physical fitness in Flemish males: a cross-sectional analysis. Am. J. Hum. Biol
. 11:587–597, 1999.
18. Philippaerts, R. M., K. R. Westerterp, and J. Lefevre. Doubly labeled water validation of three physcal activity questionnaires. J. Sports Med
. 20:284 –289, 1999.
19. Renson, R., G. Beunen, A. L. Claessens, et al. Physical fitness variation among 13 to 18 year old boys and girls according to sport participation. In: Children and Exercise
, G. Beunen, J. Ghesquiere, T. Reybrouck, et al. (Eds.). Stuttgart: F. Enke Verlag, 1990, pp. 136–144.
20. Renson, R., G. Beunen, M. De Witte, M. Ostyn, J. Simons, and D. Van Gerven. The social spectrum of the physical fitness of 12-to 19-year-old boys. In: Kinanthropometry II, M. Ostyn, G. Beunen, and J. Simons (Eds.). Baltimore: University Park Press, 1980, pp. 104–118.
21. Tanner, J. M., R. H. Whitehouse, N. Cameron, et al. Assessment of Skeletal Maturity and Prediction of Adult Height (TW2 method)
. London: Academic Press, 1983, pp. 1–108
22. Trost, S. G., N. Owen, A. E. Bauman, et al. Correlates of adults’ participation in physical activity: review and update. Med. Sci. Sports Exerc
. 34:1996 –2001, 2002.
23. Vanreusel, B., R. Renson, G. Beunen, et al. A longitudinal study on youth sport participation and adherence to sport in adulthood. Int. Rev. Soc. Sport
TRACKING; INTERAGE CORRELATIONS; FITNESS; SOMATIC DIMENSIONS; SOCIOCULTURAL DETERMINANTS
©2004The American College of Sports Medicine
Highlight selected keywords in the article text.