Increased rates of overweight and obesity have been reported in recent years in developed countries. Current International Obesity Task Force estimates suggest that worldwide at least 1.1 billion adults are overweight, including 312 million who are obeseReference James, Rigby and Leach1. In the USA, 35% of adults are overweight and an additional 30% are obese2. Mean body mass index (BMI) as well as the prevalence of overweight increased in virtually all Western European countries from the early 1980s to the mid 1990sReference Silventoinen, Sans, Tolonen, Monterde, Kuulasmaa and Kesteloot3. In Switzerland, the prevalence of overweight increased in men aged 20–29 years by >10% between 1980 and 1990 to reach 25%; and in women in the same age group, the prevalence increased progressively since 1960 to reach 10.5% in 1990Reference Groscurth, Vetter and Suter4. The overweight/obesity prevalence continued to increase by 10% in men and 8% in women between 1993 and 2000Reference Galobardes, Costanza, Bernstein, Delhumeau and Morabia5. Increases in BMI are associated with increases in lean body and fat tissues.
In epidemiological studies, both low and high BMI have been associated with increased morbidityReference Allison, Gallagher, Heo, Pi-Sunyer and Heymsfield6–Reference Tremblay and Bandi9, including higher risk of coronary heart disease, hypertension, hyperlipidaemia, diabetes, cancer and osteoarthritisReference Ogden, Carroll and Flegal10, and excess mortalityReference Ogden, Carroll and Flegal10. BMI has been used to estimate an excess or deficit in body weight, but is an imprecise measurement of fatnessReference Stolarczyk, Heyward, Van Loan, Hicks, Wilson and Reano11, Reference Segal, Van Loan, Fitzgerald, Hodgdon and Van Itallie12. The direct measurement of fat-free mass (FFM) and body fat (BF) permits a more precise determination of whether morbidity and mortality are associated with the lean or fat body compartment. Thus, it would be of interest to determine the longitudinal population trends of FFM and BF. However, the differences in FFM or BF in older, compared with younger, subjects may be due to shorter height of older subjects or due to changes in body composition. The use of the fat-free mass index (FFMI) and fat mass index (FMI), similarly to BMI (weight/height2, kg m− 2), permits comparison of subjects of different heightsReference Van Itallie, Yang, Heymsfield, Funk and Boileau13.
In long-term studies, weight gain has been associated with low physical activity levelsReference Williamson, Madans, Anda, Kleinmann, Kahn and Byers14. However, the long-term effects of leisure-time activity on FFM and BF are not well known.
While increases in the prevalence of overweight/obesity have been reported previously in SwitzerlandReference Galobardes, Costanza, Bernstein, Delhumeau and Morabia5, throughout Western EuropeReference Silventoinen, Sans, Tolonen, Monterde, Kuulasmaa and Kesteloot3 and the USA2, at the present time little is known about the prevalence of low and high FFM and BF. The purpose of this population study of healthy subjects (n = 2433) was to determine the changes in prevalence of overweight and obesity in 2003, compared with 1993, and further to determine the association of age, sex and leisure-time activity with BMI, FFMI and FMI.
There are no previously published studies that report body composition trends over a 10-year period. Although other studiesReference Gallagher, Ruts, Visser, Heshka, Baumgartner and Wang15–Reference Heitmann and Garby17 have reported body composition changes over time, these studies are limited in length of follow-up or number of subjects.
Methods
Subjects
The study population comprised two samples of healthy volunteers (1993, n = 802; 2003, n = 1631) between the ages of 20 and 74 years in Geneva, Switzerland who were non-randomly recruited through advertisement in local newspapers, at trade fairs and fun runs, among public administration staff, and by invitations sent to leisure clubs for the elderly in 1993 and 2003. Identical procedures and measurements were used for both data collection points. Subjects with known acute pathologies or physical handicap were excluded. Volunteers were invited to participate in the study if they had not visited a physicians in the last 6 months for acute or chronic conditions. Subjects with conditions that might interfere with bioelectrical impedance analysis (BIA) measurements were excluded, including water or electrolyte imbalance (such as visible oedema and ascites), skin abnormalities (e.g. pachydermia secondary to hypothyroidism) and abnormal body geometry (such as amputation, limb atrophy). Study participants were exclusively Caucasians.
The protocol to perform BIA measurements and obtain physical activity, health status and medication data was approved by the Geneva University Hospital Ethics Committee, and study subjects gave written informed consent.
Body composition measurements
Body height was measured to the nearest 0.5 cm and body weight to the nearest 0.1 kg on a balance beam scale. Subjects were dressed in indoor clothing without shoes and heavy sweaters or jackets. Whole-body resistance was measured with four surface electrodes placed on the right wrist and ankle, as previously describedReference Lukaski, Bolonchuk, Hall and Siders18. Briefly, an electrical current of 50 kHz and 0.8 mA was produced by a generator (Bio-Z2®; Spengler) and applied to the skin using adhesive electrodes (3M Red Dot T; 3M Health Care) with the subject lying supineReference Houtkooper, Lohman, Going and Howell19. The skin was cleaned with 70% alcohol.
The FFM was calculated by the following previously validated multiple regression equationReference Kyle, Genton, Karsegard, Slosman and Pichard20: FFM = − 4.104+(0.518 × height2/resistance)+(0.231 × weight)+(0.130 × reactance)+(4.229 × sex (men = 1, women = 0))
BF was calculated as weight − FFM. The BMI, FFMI and FMI were derived as FFM and BF (kg), respectively, divided by height (m) squared (kg m− 2).
FFMI and FMI (kg m− 2) were used to classify patients as normal or high FFMI, and normal, high or very high FMI. Ranges of FFMI and FMI were derived from polynomial regression equations for each of the BMI cut-offs (18.5, 25 and 30 kg m− 2) from our healthy subjects (n = 5635)Reference Kyle, Schutz, Dupertuis and Pichard21. These cut-offs correspond to World Health Organization categories for normal weight (18.5–25 kg m− 2), overweight (25–29.9 kg m− 2) and obese ( ≥ 30 kg m− 2). We did not consider the categories below 18.5 kg m− 2 because of the small number of subjects falling into this category.
The FFMI (kg m− 2) was consideredReference Kyle, Schutz, Dupertuis and Pichard21 ‘normal’ if < 19.7 (men) and < 16.7 (women); and ‘high’ if ≥ 19.8 (men) and ≥ 16.8 (women).
The FMI (kg m− 2) was considered ‘normal’ if < 5.1 (men) and < 8.1 (women); and ‘high/very high’ if >5.2 (men) and >8.2 (women);.
Leisure-time activity
Subjects completed a questionnaire to specify the hours and minutes of physical activity per week performed on a regular basis throughout the year, including seasonal variations. Leisure-time activity in this study was predominantly walking, with seasonal activities including skiing, swimming and bicycling. The subjects who performed >3 h of physical activity per week for longer than 2 months were classified as ‘physically active’. Only physical activity at moderate or high intensity (4.0 metabolic equivalents (METs) or more, as defined by the activity intensity codes by the Minnesota Leisure Time Activities QuestionnaireReference Taylor, Jacobs, Schucker, Knudsen, Leon and Debacker22), was counted. Subjects who reported less than 3 h of leisure-time activity per week were classified as ‘sedentary’, as previously reported by our groupReference Kyle, Gremion, Genton, Slosman, Golay and Pichard23, Reference Kyle, Morabia, Schutz and Pichard24.
Statistical methods
The results are expressed as the mean ± standard deviation. The differences between age, leisure-time activity groups and between 1993 and 2003 were analysed by unpaired t-tests using Statview 5.0. Chi-square was used to determine differences between body composition classifications. Statistical significance was set at P ≤ 0.05 for all tests.
Multivariable linear regression was used to model the evolution of body composition between the 1993 and the 2003 subjects. In addition to the time of measurement (1993 or 2003), age, sex and leisure-time activity were introduced in the model. This procedure was performed because we wanted to take into account any age, sex or physical activity differences between the two cohorts. It was reasonable to postulate that changes in demographic composition and leisure habits could have occurred in a 10-year time frame. Therefore, in order to capture the true ‘time’ effect, this modelling was performed in order to control for potential confounding variables. Coefficients associated with the time of measurement therefore captured the evolution of body composition adjusted for sex, age and leisure-time activity between two samples of convenience, recruited in similar conditions.
Results
Comparison of 1993 and 2003 cohort
There was no significant difference in the number of active compared with sedentary study participants between 1993 and 2003 (Table 1). On the other hand, the 2003 cohort included significantly more women than the 1993 cohort. Mean age was significantly higher in 2003 (41.4 ± 11.8 years) than in 1993 (38.1 ± 12.5 years).
Table 1 Characteristics of the two cohorts of subjects
FFMI – fat-free mass index; FMI – fat mass index; BMI – body mass index.Results are expressed as n (%) or mean ± standard deviation.
The mean BMI was significantly higher in 2003 (P < 0.001) (Fig. 1). A greater number of subjects were overweight/obese in 2003 (27.5%) compared with 1993 (17.2%, chi-square P < 0.001). The FFMI was not significantly higher in 2003 compared with 1993. However, significantly more subjects had a high FFMI in 2003 (30.2%) than in 1993 (21.8%, chi-square P < 0.001). The FMI was significantly higher in 2003 than in 1993 and a greater number of subjects had a high FMI in 2003 (28.0%) compared with 1993 (20.3%, chi-square P < 0.001) (Table 1).
Fig. 1 Frequency distribution of body mass index (BMI, kg m− 2) (top), fat-free mass index (FFMI, kg m− 2) (middle) and fat mass index (FMI, kg m− 2) (bottom) (unpaired t-test between 1993 and 2003)
After adjustment for the relevant covariates (age, sex and level of activity), the body composition indices were only slightly different from the unadjusted results (BMI 22.6 ± 0.1 kg m− 2 for the 1993 subjects and 23.5 ± 0.07 kg m− 2 for the 2003 subjects, P < 0.001; FMI 5.1 ± 0.07 kg m− 2 for the 1993 subjects and 5.6 ± 0.05 kg m− 2 for the 2003 subjects, P < 0.001; FFMI 17.5 ± 0.05 kg m− 2 for the 1993 subjects and 17.8 ± 0.03 kg m− 2 for the 2003 subjects, P < 0.001). However, the difference between the 1993 and the 2003 cohort concerning the FFMI became statistically significant when adjusted values were used.
Sedentary and active cohorts
The mean BMI, FMI and FFMI were significantly higher in 2003 than in 1993 in sedentary and active men and women (except for non-significant differences in sedentary women).
The mean BMI was significantly higher in sedentary subjects in both 2003 and 1993 (men, 25.0 ± 3.0 and 23.7 ± 2.6 kg m− 2; women, 23.0 ± 3.3 and 22.6 ± 3.3 kg m− 2) than in the active cohort (men, 24.1 ± 2.6 and 22.9 ± 2.6 kg m− 2; women, 22.0 ± 2.8 and 20.6 ± 1.9 kg m− 2, respectively, P < 0.05). Similarly, the mean FMI was significantly higher in the sedentary subjects (men, 5.5 ± 2.0 and 4.8 ± 1.8 kg m− 2; women, 6.9 ± 2.2 and 6.7 ± 2.3 kg m− 2) than in the active cohort (men, 4.6 ± 1.7 and 4.7 ± 1.4 kg m− 2; women, 5.7 ± 1.6 and 6.0 ± 2.0 kg m− 2, P < 0.001) in 2003 and 1993, respectively. FFMI was not significantly higher in 1993 in active than in sedentary subjects (men, 18.9 ± 1.2 and 18.9 ± 1.4 kg m− 2; women, 15.6 ± 1.0 and 16.0 ± 1.4 kg m− 2, P>0.05) and in 2003 in men (19.3 ± 1.4 and 19.5 ± 1.6 kg m− 2, P>0.05), but was significantly higher in sedentary than active women in 1993 (16.0 ± 1.4 and 15.6 ± 1.0 kg m− 2, P = 0.02).
In order to take into account the observed differences in sex and age distributions and to isolate the ‘pure’ time effect represented by the study period, a multivariable analysis that adjusts for these differences was used. The multivariate regression model (Table 2) shows that leisure-time activity was negatively associated with all body composition indices. Age and year of inclusion were associated with an increase in all three body composition indices. Men had a significantly higher BMI and FFMI, and a significantly lower FMI than women.
Table 2 Associations of physical activity, sex and age on body mass index and body composition indices
BMI – body mass index; FFMI – fat-free mass index; FMI – fat mass index; CI – confidence interval.
Discussion
Prevalence of overweight and obesity
Overall the prevalence of overweight/obesity was 10% higher and the prevalence of high FMI was 8.0% higher in 2003 than in 1993. This suggested that overall rates of overweight and obesity are on the rise in Switzerland. The study showed that 27.6 and 5.4% of sedentary subjects, respectively, were considered overweight and obese in 2003, which is similar to overweight and obesity rates of 29 and 8%, respectively, reported in a random survey in SwitzerlandReference Eichholzer, Bernasconi, Jordan and Gutzwiller25. This compared with overweight and obesity rates of 65% (35 and 30%, respectively) in the USA2, 60% in GermanyReference Mensink, Lampert and Bergmann26 and obesity rates of 19% in SpainReference Mataix, Lopez-Frias, Martinez-de-Victoria, Lopez-Jurado, Aranda and Llopis27, reported in the late 1990s/early 2000s. Thus the rate of overweight in Switzerland is gradually approaching that seen in the USA and Western Europe, but obesity rates remain slightly lower than in other developed countries. Only six ( < 0.01%) sedentary and active subjects had BMI >35 kg m− 2. Rates of overweight and obesity in this study were lower because of exclusion of subjects with health-related problems (recent hospitalisation, chronic disease). However, the BMI change from 1993 to 2003 could result in considerable increases in overweight and obesity in the future.
Our study showed that the higher prevalence of overweight/obesity in 2003 than 1993 resulted partly in a higher prevalence of high FFMI in active subjects. This effect can be considered beneficial, because higher lean tissue reserves are thought to protect from detrimental effects of malnutrition, such as sarcopenia and frailty in older subjectsReference Gallagher, Ruts, Visser, Heshka, Baumgartner and Wang15, Reference Baumgartner, Koehler, Gallagher, Romero, Heymsfield and Ross28.
The prevalence of low and high/very high FMI was similar to the prevalence of low and overweight/obese BMI. Overall the prevalence of high and very high FMI was underestimated by 1–2% by BMI. This suggests that in healthy, physically active subjects, BMI provides a good estimation of fatness. However, our previous studies have shown significant underestimations of fatness by BMI in patients at hospital admissionReference Kyle, Nicod, Raguso, Hans and Pichard29, Reference Kyle, Morabia, Unger, Slosman and Pichard30 and patients with respiratory insufficienciesReference Kyle, Raguso, Janssen and Pichard31. This finding suggests that excess body fat is poorly estimated by BMI in sedentary subjects or those with chronic diseases. Our study did not determine fat distribution, i.e. did not distinguish between central and subcutaneous adiposity, and no data are available on the possible shift of fat from limbs to trunk.
Association of leisure-time activity, sex, age and year of inclusion with BMI, FFMI and FMI
As expected, sex was positively associated with FFMI and negatively associated with FMI. Numerous studies have documented higher muscle mass and lower body fat in men than in womenReference Gallagher, Ruts, Visser, Heshka, Baumgartner and Wang15, Reference Kyle, Genton, Gremion, Slosman and Pichard32. BMI was also positively associated with sex, men having a higher BMI than women.
Our results show that leisure-time activity was negatively associated with BMI, FFMI and FMI. Leisure-time activity may, through the effects of increased energy expenditure, preserve both functional status and lean body mass, and contribute to reduce fat accumulationReference Tager, Haight, Sternfeld, Yu and van Der Laan33. In previous studiesReference Williamson, Madans, Anda, Kleinmann, Kahn and Byers14, low recreational activity reported at follow-up survey was strongly associated with major weight gain (>13 kg over the preceding 10 years).
Even though the age and gender composition of the two cohorts were different, the year of inclusion remained statistically significant in a multivariable analysis and was positively associated with BMI, FFMI and FMI. We are thus confident that we captured a real evolution of body composition in the Geneva population across a 10-year period. The association of age and year of inclusion (2003 versus 1993) with BMI increase was higher than the beneficial effects of leisure-time activity and thus resulted in higher BMI and FMI in both the sedentary and active cohort in 2003 than 1993 and also resulted in higher FFMI in 2003. This suggests that the higher weights and BMI, at least within the limits of change reported in this study, did not necessarily have a negative effect on body composition in active subjects in view of the higher prevalence of high FFMI and low FMI (data not shown). The lower rates of obesity in the active subjects also suggest that leisure-time activity had a positive effect on BMI and weight, and support a relationship between physical inactivity and the development of overweight and obesityReference Haapanen, Miilunpalo, Pasanen, Oja and Vuori34.
Haapanen et al. Reference Haapanen, Miilunpalo, Pasanen, Oja and Vuori34 found that increased physical activity was associated with small body mass gain and low physical activity, and, in particular, decreasing levels of activity during a 10-year follow-up period was strongly associated with large body mass gain.
Our active subjects expended at least 720 kcal week− 1 (180 min week− 1 at 4 METS) on leisure-time activity. Physical activity expending 1000 kcal week− 1 has been associated with a 30% reduction in all-cause mortality rates, and a slightly favourable effect on all-cause mortality has been noted with physical activity as low a 500 kcal week− 1 (Kesaniemi et al. Reference Kesaniemi, Danforth, Jensen, Kopelman, Lefebvre and Reeder35). Lee et al. Reference Lee, Jackson and Blair36 found that fit but overweight men (BMI ≥ 27.8 kg m− 2) had a similar rate of all-cause mortality to physically fit men of above normal weight and had a lower risk of all-cause mortality than unfit, normal weight men. Unfit men had substantially higher cardiovascular disease mortality than fit men in each BMI groupReference Lee, Jackson and Blair36.
Numerous studies have shown a J- or U-shaped relationship between BMI and mortalityReference Allison, Gallagher, Heo, Pi-Sunyer and Heymsfield6, Reference Rissanen, Heliövaara, Knekt, Reunanen and Aromaa37, Reference Cornoni-Huntley, Harris, Everett, Albanes, Micozzi and Miles38 and a U-shaped relationship between BMI and expenditure on health careReference Heithoff, Cuffel, Kennedy and Peters39. Heitman et al. Reference Heitmann, Erikson, Ellsinger, Mikkelsen and Larsson7 suggested that the apparent U-shaped association between BMI and total mortality may be the result of compound risk functions from BF and FFM, e.g. total mortality was a linear function of high BF and low FFM. Allison et al. Reference Allison, Gallagher, Heo, Pi-Sunyer and Heymsfield6 evaluated a hypothetical model in which death increased linearly with BF and decreased linearly with FFM. In spite of higher weights in our active subjects in 2003 than in 1993, the active subjects showed a desirable profile of body composition with preserved FFMI and smaller increases in FMI compared with sedentary subjects. On the other hand, sedentary subjects are at risk of becoming obese. Thus leisure-time activity appears to counteract some of the negative effects of high BMI. Public health initiatives should therefore stress the benefits of leisure-time activity on the positive effects on body composition in terms of morbidityReference Allison, Gallagher, Heo, Pi-Sunyer and Heymsfield6.
Study limitations
The main limitation of the study is that it consisted of two transversal measurements, made in two distinct samples of convenience. It is therefore possible that the body composition evolution observed between the two cohorts is accidental. However, the recruitment conditions and the measurement procedures were similar.
The socio-economic level was not assessed; however, we do not believe that it affected the prevalence of obesity. Our study did not adjust for smoking, menopausal status and other lifestyle determinants (diet, alcohol). Whole-body composition measurements in our study did not determine fat distribution, i.e. did not identify intra-abdominal versus subcutaneous obesity. Collection of body composition as well as of waist and hip circumference data in conjunction with body composition measurements is recommended for future studies.
Conclusion
Overall the prevalence of overweight/obesity was 10% higher and the prevalence of high FMI was 8.0% higher in 2003 than in 1993 in a sample of convenience recruited in a Swiss city, and was mainly associated with an increase in fat mass. This observation remained statistically significant after adjustment for age, sex and leisure-time activity.
Acknowledgements
Sources of funding: Foundation Nutrition 2000Plus.
Conflict of interest declaration: none.
Authorship responsibilities: U.G.K. was involved in data design, collection of data, analysis of data and writing of the manuscript. L.G. and M.P.K. carried out analysis of data and writing of the manuscript. C.P. was involved in data design, data analysis, writing of the manuscript, and obtaining Institutional Review Board approval and funding.
