A physically active lifestyle is known to have significant health benefits and is a common goal of health promotion initiatives (27). From an epidemiological view, familial resemblance reflecting genetic and environmental factors shared by family members could be a major determinant of the level of habitual physical activity or sedentarism. To date, however, the studies of the familial resemblance in physical activity level have provided controversial results.
There are inconsistencies in the magnitude of the familial resemblance in physical activity participation. Both low (0.08–0.09) (1,21) and moderate (0.45–0.49) (4,19,26) parent-offspring correlations for various physical activity phenotypes have been found. Correlations among other family members are also highly variable. For example, spousal correlations ranged from 0.16 (21) to 0.60 (19) in two studies on physical activity level. Some studies have found greater familial effect for inactivity (9) or habitual physical activity (21) than for strenuous modes of exercise, but the opposite trend has also been reported (24).
Similarly, heritability estimates for physical activity levels are widely divergent. They range from no genetic effect at all for exercise participation (21) to moderate effect for physical activity (16) or exercise (14,16), or to even high heritability for exercise (4), with twin studies providing generally greater heritability coefficients than family designs.
However, in the aggregate, these studies suggest that there is a familial component affecting the level of habitual physical activity and participation in moderate to vigorous physical activity, although its magnitude is controversial. Thus, the aim of this study was to further investigate the issue of familial resemblance for degree of inactivity, habitual physical activity level, and exercise participation using four phenotypes. For each of these phenotypes, a familial correlation model was applied to quantify the pattern of intrafamilial correlations and explore the relative roles of familial, environmental and genetic factors.
Study design and subjects.
In the second phase of the Québec Family Study (5), white French-Canadian nuclear families from the greater Québec City area were investigated between 1989 and 1997. A total of 200 families with 696 members (312 parents; 384 offspring; mean family size 3.6 ± 2.0 individuals) with complete physical activity phenotype data were included in the study. Informed consent was obtained from each subject, and the study protocol was approved by the Institutional Review Board of Laval University.
Physical activity assessment.
Using a 3-d activity diary, which included one weekend day, subjects were instructed to record the dominant activity for each 15-min period during 24 h by using a list of categorized activities. In those categories (from 1 to 9), activities were ordered in increasing magnitude based on the energy expenditure (METs) of each activity (6). Category 1 indicated very low energy expenditure such as sleeping or resting in bed, category 5 included light manual work such as carpentry or moderately fast walking, and category 9 consisted of intense manual work such as sawing with a hand saw or playing sports such as soccer.
Three different phenotypes were calculated from the diary. The number of 15-min periods for each category was first summed over the 3 d and weighted by its own category number, which ranged from 1 to 9, i.e., from the lightest activity category to the most strenuous activity category. Total physical activity was then calculated by summing over all nine categories. The inactivity phenotype was derived from categories 1 to 4 (including activities such as sleeping, eating, watching TV, washing oneself, cooking, driving a car, and strolling-type of walking), and the sum of categories 5–9 was used as a proxy for moderate to strenuous physical activity.
The reliability of the activity record was investigated among 61 adults and children from Québec who completed it twice 6–10 d apart. An intraclass correlation coefficient of 0.96 was obtained, with no major differences between children (ICC = 0.91) and adults (ICC = 0.97) (6). Furthermore, the activity level was positively related to an indicator of fitness and negatively correlated with fatness. A slightly modified version of the diary was found to provide a close estimate of total energy expenditure as measured by doubly labeled water in adolescents (7).
In addition, subjects were probed for their past year’s involvement in physical activity using a questionnaire. The questionnaire included 11 questions and several subquestions. For this study, the average number of times per week over the last year and the average duration at each session, for the physical activity most often engaged in, were extracted and used to compute the time spent on the most common physical activity (hours per week) during the previous year. These activities could include organized sports, walking, cycling, and any other activities with the exception of occupational activity and household chores.
Each physical activity phenotype was adjusted within sex: by-generation subgroups for age, age2, and age3 by a stepwise regression procedure. Because some of the Quebec Family study families were selected based on discordant sibpairs for obesity, regression parameters were calculated using subjects from randomly sampled families with outliers (±3 SD) temporarily set aside. These parameters were then applied to the whole cohort to compute the residual score. Only significant covariates (P < 0.05) were included in the model. Residuals were standardized to a mean of 0 and a standard deviation of 1. Analysis of variance was used to assess the within and between families variance components of each adjusted phenotype.
A familial correlation model was applied to each adjusted phenotype using a maximum likelihood program SEGPATH (22). A summary of hypotheses used for model testing is presented in Table 1. The likelihood ratio test consisting of the difference in minus twice the log-likelihoods (−2lnL) obtained in a general versus a reduced model was used for testing these hypotheses. The likelihood ratio, with the degrees of freedom being the difference in the number of parameters estimated in the nested competing hypotheses, is approximately distributed as a χ2. The Akaike’s information criterion (AIC) (2) was used to compare nonnested models and identify the most parsimonious hypothesis. The AIC is −2lnL plus twice the number of estimated parameters.
Testing was carried out in two blocks consisting of sex and generation hypotheses (models 2, 3, and 4) and hypotheses about the strength of the correlations (models 6, 7, and 8) among the nonrejected hypothesis. Finally, the most parsimonious model among all tested hypotheses was the model with the lowest AIC value.
The maximum likelihood estimates of the familial correlations obtained under the most parsimonious model were used in computing the maximal heritabilities (h2) as follows (23):EQUATION
The heritability arising from this equation includes both genetic and nongenetic sources of familial variance and adjusts for spouse resemblance. The confidence intervals are obtained by substituting the standard errors obtained from the familial correlations into the above equation. This heritability estimate will be referred to as a “maximal heritability” to emphasize that it includes both genetic and familial environmental factors transmitted from parents to offspring. The same formula was used to calculate maximal heritability confidence intervals from the standard errors of the familial correlations in the most parsimonious model. The model assumptions are no gene-environment interaction, no gene-environment correlation, no gene-gene interactions, and no assortative mating.
The basic descriptive data for age, body mass, and the activity phenotypes in fathers, mothers, sons, and daughters are presented in Table 2. Male subjects tended to be more physically active than female subjects. Furthermore, particularly among female subjects, parents were more active than offspring. The results of the ANOVA presented in Table 3 indicate that there were 1.40–1.52 times more variation in physical activity levels between families than within families.
The results of the model fitting are presented in Table 4. Based on the individual tests, the hypothesis of no familial correlation (model 9) was rejected for all phenotypes, whereas the sibling correlations (model 6) were not significant for any trait. For the inactivity phenotype, there were sex differences (model 2) but no spouse resemblance. For the moderate to strenuous physical activity phenotypes, there were no sex differences, and the parent offspring (model 7) and spouse correlations (model 8) were significant. For the total daily activity and the time sent in physical activity phenotypes, the only significant correlation was for spouse resemblance.
The most parsimonious model (Table 4) was derived by combining nonrejected hypotheses and selecting the model with the lowest AIC. Thus, the “best” hypotheses were the environmental model (all 8 correlations equal) in the case of inactivity, and a combination of no sex or generation effects and no sibling correlations for moderate to strenuous physical activity. For both total physical activity and time spent in physical activity, the most parsimonious model was for no sex or generation effects. Maximal heritabilities computed under the most parsimonious models ranged from 16 to 25% (Table 5). For all phenotypes, spouse resemblance was greater (0.14–0.43) than the parent-offspring and sibling correlations (0.13–0.16).
The present study revealed the presence of significant familial resemblance in physical activity and inactivity levels. Maximal heritability, a combination of genetic factors and shared environment, explained 16–25% of the phenotypic variation. The highest heritability level was for the inactivity phenotype. The pattern of familial correlations suggests that genetic factors alone do not explain the observed familial resemblance, and that the shared environment also contributes importantly to the observed familial resemblance in physical activity and inactivity phenotypes.
In a previous study, based on the phase 1 data of the Québec Family Study (1979–1981), shared family environment explained 0% and 12%, whereas genetic factors explained 29% and 0% of the variation in habitual physical activity level and exercise participation, respectively (21). Although physical activity level was defined differently in this preceding report, the level of familial resemblance for physical activity level was modest and in agreement with the estimates of the present study. In another study involving a total of 18,073 subjects from 4,678 families who were participants in the Canada Fitness Survey, low familial correlations (0.12–0.28) were also observed for physical activity levels based on data from a questionnaire (19).
In contrast, in the Leuven Longitudinal Twin Study (3), shared family environment explained 54% of the variation (confidence interval (CI): 0.06–0.77) of sports participation among girls, whereas for boys the family environment was not significant because variation in sports participation was explained by both genetic (83%, CI: 0.66–0.91) and unique environmental factors (17%, CI: 0.09–0.34).
In the present study, spousal correlations (0.14–0.43) were higher than correlations between same-sex siblings (−0.06 to 0.10) except for inactivity, for which the correlation among brothers reached 0.31. For the other phenotypes, except level of light to moderate physical activity, spouse correlations were relatively high (0.22–0.43). In other family studies, both high (4,19) and low (1,21) spouse correlations have been reported. Spousal resemblance in physical activity is likely to be indicative of shared environmental effects during the years of cohabitation, nonrandom mating, or both. One may argue that observable behavioral traits such as physical activity level may to some extent influence mate selection, hence favoring a spouse resemblance for such phenotypes. Interestingly, spouse resemblance has also been observed for physical fitness levels in family studies (13,19,20).
Time spent in sedentary activities showed greater family resemblance than indicators of other physical activity levels in this study. One possible explanation for this finding might be that familial resemblance occurs primarily at the upper or lower end of the distribution, i.e., for time spent engaging in vigorous exercise or inactivity, rather than for moderate physical activity levels. Again, a similar pattern has been reported for physical fitness (13). A greater familial aggregation effect has been noted particularly among highly active parents and their children (1,10) as well as sedentary parents and their children (1,9,10). The inability to find high familial resemblance for participation in moderate to strenuous physical activities in this study may have resulted from the combination of those two intensities under one phenotype. In addition, the physical activity categories of the diary included both demanding occupational and manual labor plus strenuous leisure-time physical activities. Occupational physical activities consist of less-voluntary activities than leisure-time physical activities, and thus they may have different determinants.
A unique feature of this study was that offspring were adults (mean age in the late 20s). Thus, these results imply that the familial resemblance for physical activity levels persist beyond childhood and adolescent years, well after the age when offspring and siblings move away from the family home and live apart. Among students living apart from their parents for 3 yr after entering university, parental influence explained almost as much of the variation (27%) in exercise habits as did peer influence (32%) (15). Whether the parent-offspring similarity increases or decreases with age needs to be more thoroughly investigated in a longitudinal family study design.
There are critical issues to consider in interpreting the results of family studies on physical activity level. Physical activity participation fluctuates across the life span (8,17). In particular, physical activity level tends to be sporadic during childhood (11,25). This may increase variability and thus decrease the amount of familial resemblance. For example, Kaprio et al. (12) reported heritability differences across the life span for physical activity level among young and older MZ and DZ twins. The heritability estimates ranged from 0.32 to 0.64, being highest for the 18- to 29-yr-old twins and lowest (and statistically nonsignificant) for 50- to 59-yr-old twins. Variations in heritability level across the life span may thus explain some of the apparent inconsistencies in the familial correlations even though the data are generally adjusted for age. There may also be some fluctuation in strenuous physical activity participation across the weekdays, and as long as 21 to 28 d physical activity assessments have been recommended for nonoccupational physical activities (18). Even though the 3-d diary used in the present study was characterized by a test-retest reliability coefficient (ICC = 0.96) for the mean energy expenditure over 3 d, lower reliabilities were observed for the more intense physical activity categories (r = 0.48 to 0.88) (6). Finally, it is important to keep in mind that the heterogeneity in results among the various studies of familial resemblance for physical activity or inactivity level may reflect cultural or cohort differences, unequal sex effects, small sample sizes, or dissimilar ascertainment of physical activity or sedentarism.
In summary, these data suggest that the contribution of genetic factors and family environment in determining physical activity level is relatively small. Inactivity has a slightly higher heritability level than engagement in moderate to strenuous physical activity or the overall level of physical activity. These observations have implications for those concerned by the high prevalence of a sedentary lifestyle and the generally low levels of habitual physical activity in North America.
The Quebec Family study has been supported by multiple grants from Medical Research Council of Canada (PG-11811, MT-13960, and GR-15187). This study was also supported by grants from Academy of Finland and Finnish Ministry of Education to R. Simonen. C. Bouchard is partially supported by the George A. Bray Chair in Nutrition.
Address for correspondence: Claude Bouchard, Ph.D., Pennington Biomedical Research Center, Human Genomics Laboratory, 6400 Perkins Road, Baton Rouge, LA 70808; E-mail: [email protected] pbrc.edu.
1. Aarnio, M., T. Winter, U. M. Kujala, and J. Kaprio. Familial aggregation of leisure-time physical activity: a three generation study. Int. J. Sports Med. 18: 549–556, 1997.
2. Akaike, H. A new look at the statistical model identification. IEEE Trans. Automat. Cont. 19: 16–23, 1974.
3. Beunen, G., and M. Thomis. Genetic determinants of sports participation and daily physical activity. Int. J. Obes. 23: S55–S63, 1999.
4. Boomsma, D. I., M. B. M. van den Bree, J. F. Orlebeke, and C. M. Molenaar. Resemblances of parents and twins in sports participation and heart rate. Behav. Genet. 19: 123–141, 1989.
5. Bouchard, C. Genetic epidemiology, association and sib-pair linkage: results from The Québec Family Study. In: Molecular and Genetic Aspects of Obesity, G. A. Bray and D. H. Ryan (Eds.). Baton Rouge, LA: Louisiana State University Press, 1996, pp. 470–481.
6. Bouchard, C., A. Tremblay, C. Leblanc, G. Lortie, R. Savard, and G. Theriault. A method to assess energy expenditure in children and adults. Am. J. Clin. Nutr. 37: 461–467, 1983.
7. Bratteby, L.-E., B. Sandhagen, H. Fan, and G. Samuelson. A 7-day activity diary for assessment of daily energy expenditure validated by the doubly labelled water method in adolescents. Eur. J. Clin. Nutr. 51: 585–591, 1997.
8. Caspersen, C. J., M. A. Pereirea, and K. M. Curran. Changes in physical activity patterns in the United States, by sex and cross-sectional age. Med. Sci. Sports. Exerc. 32: 1601–1609, 2000.
9. Fogelholm, M., O. Nuutinen, M. Pasanen, E. Myöhänen, and T. Säätelä. Parent-child relationship of physical activity patterns and obesity. Int. J. Obes. 23: 1262–1268, 1999.
10. Freedson, P. S., and S. Evenson. Familial aggregation in physical activity. Res. Q. Exerc. Sport 62: 384–389, 1991.
11. Janz, K. F., J. D. Dawson, and L. T. Mahoney. Tracking physical fitness and physical activity from childhood to adolescence: the Muscatine Study. Med. Sci. Sports. Exerc. 32: 1250–1257, 2000.
12. Kaprio, J., M. Koskenvuo, and S. Sarna. Cigarette smoking, use of alcohol and leisure-time physical activity among same-sexed adult male twins. In: Twin Research 3: Epidemiological and Clinical Studies. New York: Alan R. Liss, Inc., 1981, pp. 37–46.
13. Katzmarzyk, P. T., L. Pérusse, D. C. Rao, and C. Bouchard. Familial risk ratios for high and low physical fitness levels in the Canadian population. Med. Sci. Sports. Exerc. 32: 614–619, 2000.
14. Koopmans, J. R., L. J. P. Van Doornen, and D. I. Boomsma. Smoking and sports participation. In: Genetic Factors in Coronary Heart Disease, U. Goldbourt, U. de Faire, and K. Berg (Eds.). Dordrecht, The Netherlands: Kluver Academic, 1994, pp. 217–35.
15. Lau, R. R., M. Jacobs Quadrel, and K. A. Hartman. Development and change of young adults preventive health beliefs and behavior: influence from parents and peers. J. Health Soc. Behav. 31: 240–259, 1990.
16. Lauderdale, D. S., R. Fabsitz, J. M. Meyer, P. Sholinsky, V. Ramakrishnan, and J. Goldberg. Familial determinants of moderate and intense physical activity: a twin study. Med. Sci. Sports Exerc. 29: 1062–1068, 1997.
17. Lefevre, J., R. M. Philippaerts, K. Delvaux. Daily physical activity and physical fitness from adolescence to adulthood: a longitudinal study. Am. J. Hum. Biol. 12: 487–497, 2000.
18. Matthews, C. E., J. R. Hebert, P. S. Freedson, et al. Sources of variance in daily physical activity levels in the seasonal variation of blood cholesterol study. Am. J. Epidemiol. 153: 987–995, 2001.
19. Pérusse, L., C. Leblanc, and C. Bouchard. Inter-generation transmission of physical fitness in the Canadian population. Can. J. Sport. Sci. 13: 8–14, 1988.
20. Pérusse, L., C. Leblanc, A. Tremblay, et al. Familial aggregation in physical fitness, coronary heart disease risk factors, and pulmonary function measurements. Prev. Med. 16: 607–615, 1987.
21. Pérusse, L., A. Tremblay, C. Leblanc, and C. Bouchard. Genetic and environmental influences on level of habitual physical activity and exercise participation. Am. J. Epidemiol. 129: 1012–1022, 1989.
22. Province, M. A., and D. C. Rao. General purpose model and a computer program for combined segregation and path analysis (SEGPATH): automatically creating computer programs from symbolic language model specifications. Genet. Epidemiol. 12: 203–219, 1995.
23. Rice, T., J.-P. Després, E. W. Daw, et al. Familial resemblance for abdominal visceral fat: The HERITAGE Family Study. Int. J. Obes. 21: 1024–1031, 1997.
24. Sallis, J. F., T. L. Patterson, M. J. Buono, C. J. Atkins, and P. R. Nader. Aggregation of physical activity habits in Mexican-American and Anglo families. J. Behav. Med. 11: 31–41, 1988.
25. Simons-Morton, B. G., N. M. O’Hara, D. G. Simons-Morton, G. S. Parcel. Children and fitness. a public health perspective. Res. Q. 58: 295–302, 1987.
26. Willerman, L., and R. Plomin. Activity level in children and their parents. Child Dev. 44: 854–858, 1973.
27. Vuori, I. Does physical activity enhance health? Patient Educ. Couns. 33: S95–103, 1998.