| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research Communication |
1 From the Departments of Epidemiology and Health Promotion (ML-K, PP, and EV) and Health and Functional Capacity (MH), National Public Health Institute, Helsinki.
2 Supported by a research grant from the Finnish Cultural Foundation.
3 Address reprint requests to M Lahti-Koski, Nutrition Unit, National Public Health Institute, Mannerheimintie 166, FIN-00300 Helsinki, Finland. E-mail: marjaana.lahti-koski{at}ktl.fi.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: We investigated associations of body mass index (BMI) and obesity with physical activity, food choices, alcohol consumption, and smoking history. In addition, we examined the consistency of these associations over time, with the aim of assessing whether the significance of lifestyle variables as correlates of obesity increased over a 15-y period.
Design: Independent cross-sectional surveys were carried out in 1982, 1987, 1992, and 1997. Altogether, 24604 randomly selected men and women (aged 25–64 y) participated in these surveys. The subjects' weights and heights were measured, and data on lifestyle were collected with self-administered questionnaires.
Results: In men and women, perceived general health, leisure-time physical activity, and daily vegetable consumption were inversely associated with obesity, as were bread consumption in women and activity at work in men. Consumption of sausages, milk, and sour milk and heavy work (in women only) were positively associated with obesity. Obesity was also associated with alcohol consumption and smoking history. Most associations were constant over the 15-y period. However, the inverse associations of BMI with physical activity in women and with perceived health in men seemed to strengthen over time.
Conclusions: A physically active lifestyle with abstention from smoking, moderate alcohol consumption, and consumption of healthy foods maximizes the chances of having a normal weight. The significance of avoiding sedentariness increases over time as a factor associated with normal weight.
Key Words: Obesity • population studies • FINRISK Studies • lifestyle • physical activity • food choices • body mass index
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our society is becoming increasingly "obesogenic" (15). Thus, although obesity has a strong genetic background (16), environmental factors are commonly considered to be the underlying cause of the increase in obesity by promoting or exacerbating the problem (16, 17). In Britain, for example, the increase in the prevalence of obesity was attributed to a reduced level of physical activity, which was determined to more plausibly represent the dominant factor than was the intake of energy-dense food (18). Analyses in these studies, however, are usually based on population-level estimates of environmental factors. Surveys in which an increase in body mass index (BMI) is examined in relation to changes in lifestyle variables measured within the same population are scarce.
In previous studies, we showed that the prevalence of obesity in the Finnish working-age population is high: almost 20% of both men and women are obese. Furthermore, an upward trend in BMI was observed, especially in men, between 1982 and 1997 (19). The purpose of the present study was, therefore, to investigate the associations of BMI and obesity with lifestyle variables (physical activity, food choices, alcohol consumption, and smoking) in this population. A further aim was to examine whether these associations changed over the 15-y period to assess whether the significance of any of these lifestyle variables as correlates of BMI and obesity had increased over this period.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
An independent random sample was drawn from the population register for each survey. The samples covered the age range of 25–64 y and were stratified according to the World Health Organization MONICA (Monitoring Trends and Determinants of Cardiovascular Disease) protocol (21), such that 
250 subjects of both sexes in each 10-y age group were chosen from each region. The total sample included 15761 men and 15518 women (Table 1
). The participation rate varied across the years between 70.0% and 79.1% in men and between 76.5% and 85.0% in women. The final sample comprised 11857 men and 12747 women. The protocol was approved by the Ethical Committee of the National Public Health Institute.
|
30), to investigate the associations of lifestyle with both relative weight and obesity. Data on mean BMI and the prevalence of overweight and obesity by survey year are presented in Table 2. The overall changes in BMI and obesity over time are described and discussed in more detail elsewhere (19).
|
The measurement of physical activity in the questionnaire had several components. Occupational activity was originally inquired about in a question with 4 response categories, from physically very light office work to strenuous work. In this study, we combined the first 2 and the last 2 groups, defining the work of participants as either physically light or physically heavy. To evaluate physical activity during travel to and from work, the subjects were asked whether they walked, cycled, or used motorized transportation and the daily duration of this activity. Retired or unemployed persons were kept separate in the analyses.
Leisure-time physical activity was assessed with 3 questions. In the first question, the level of leisure-time physical activity was measured with 4 alternatives ranging from no activity to heavy competitive training several times per week. Because of the small number of subjects reporting competitive training, the third and fourth groups were combined. The frequency of leisure-time physical activity was determined in a question with 6 response alternatives, and the duration of these exercise sessions was determined in a question with 5 categories, ranging from 0 to 60 min per session. Weekly time spent on leisure-time physical activity was calculated as the product of frequency times duration. On the basis of their total weekly time of leisure-time physical activity, the subjects were divided into 4 groups: <20, 20–44, 45–150, and >150 min/wk. Subjects reporting being unable to exercise as a result of illness or injury were kept separate in the analyses of weekly exercising.
The number of questions assessing the subjects' diet and food choices varied across the surveys. In this study, we used only those questions (n = 10) included in all surveys. These questions were of 3 types. First, the type of food usually consumed was evaluated (eg, "What kind of milk, cooking fat, and fat on bread do you usually choose?"). Second, the amount of food consumed daily was assessed (eg, "How many glasses of milk and sour milk, slices of bread, and cups of coffee and tea do you have daily?"). Finally, the frequency of consumption of vegetables and sausages over the past 12 mo was inquired about with 6 alternatives ranging from "more seldom than once a month" to "daily."
Alcohol consumption was assessed with questions on the types (beer, wine, or liquor), frequency, and amount of alcohol consumed during the previous week. On the basis of this information, an alcohol index was calculated indicating the intake of alcohol in grams per week.
Smoking history was assessed by using a standard set of questions. According to their responses, the participants were classified into 3 groups: those who had never smoked (never-smokers), those who had quit smoking 6 mo ago (ex-smokers), and those currently smoking (smokers). We also defined as smokers those who had quit smoking <6 mo ago.
Perceived general health was used as an indicator of health status. It was measured by asking, "How would you assess your current health?" There were 5 response alternatives: good, fairly good, average, rather poor, and poor. The first 2 and the last 2 groups were combined because the extreme classes were too infrequent for data analyses.
Statistical analyses
Data from the 3 regions in Finland were pooled. Furthermore, data from the 4 surveys were analyzed together, with survey year used as a factor in all analyses. All analyses were carried out separately for men and women and were controlled for age and education. In this study, education level was measured as the total number of school years, on the basis of which the subjects were divided into 3 groups (low, middle, and high) within each birth year.
The associations between lifestyle factors and obesity were estimated by logistic regression analysis with use of the PROC LOGISTIC procedure of SAS (version 6.12; SAS Institute Inc, Cary, NC). In the series of logistic regressions, being obese was used as a dependent variable and lifestyle factors (for which binary variables were used to represent a categorical variable) were used as independent variables. Both separate models for each single lifestyle variable and a multivariate model were used. The separate models included only a single lifestyle variable at a time, controlled for age and education, whereas the multivariate model included all the lifestyle variables together with age and education. An association was defined as significant if the 95% CI for an odds ratio did not include unity. In addition, to study any inconsistencies in the associations of the variables with obesity over time, the interactions between survey year and the variable of interest were calculated from the logistic models by using the PROC GENMOD procedure of SAS.
Linear regression analysis was used to estimate the independent effect of single lifestyle variables on the variation in mean BMI as well as the stability of the possible association between BMI and this lifestyle variable over time. These analyses were carried out by using the generalized linear model procedure of SAS, with mean BMI as a dependent variable. In the first model, after adjustment for age and education, year and the lifestyle variable were included in the model to examine the main effects of these variables. In the second model, an interaction term, year by the variable of interest, was added to test whether the possible association between BMI and lifestyle variable varied across the survey years.
Perceived health summarizes a broad range of health-related information on individuals (22) that is reliable enough to be used in population surveys (23). Because the associations between obesity and ill health are amply documented in the literature, and general health is likely to covary with both lifestyle and BMI, perceived health in the current study was considered mainly as a confounding factor. Thus, all the analyses were also done by adjusting for perceived health. The main results did not change materially, however; thus, the final analyses were controlled for age and education only. The adjusted mean BMI values were obtained by using least-squares means.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). The association of activity level at work with obesity was constant in both sexes over time, as was the association with mean BMI in the women (Figure 1). In the men, however, the most prominent increase in the time trend of mean BMI occurred in those outside the work force.
|
|
15 min/d were less likely to be obese than were the women who commuted by motor vehicle or walked for a shorter duration (Table 3
). The pattern was less clear in men.
The subjects who were moderately or highly active at leisure time were less likely to be obese than were the subjects with a low level of activity, ie, those who spent their leisure time mostly reading or watching television (Table 3). This association was constant in both sexes over time. The phenomenon of being more active and having a diminished likelihood for obesity was also seen when we investigated the total weekly time of leisure-time physical activity in a single-variable model. These associations, however, disappeared after we controlled for all the other lifestyle variables. The association between obesity and weekly time of leisure-time physical activity did not vary across the survey years in the men, whereas in the women, this association became stronger over time (P value for interaction between year and weekly time of leisure-time physical activity: 0.0023 for women).
In both the men and the women, the strongest upward trend in mean BMI over time occurred in the group with the lowest level of leisure-time physical activity (Figure 2). The association between total weekly time of leisure-time physical activity and mean BMI remained unchanged in men over the 15-y period, whereas in women, an upward trend in mean BMI over time was less likely in those exercising 45 min/wk (Figure 3).
|
|
). No association was found between type of cooking fat and obesity, but the men and women who chose the butter-oil mixture or butter on bread and the men who used margarine on bread were less likely to be obese than were the subjects who ate bread without a fat spread. When the type of fat used on bread was ignored, those who ate bread without a fat were more likely to be obese than were the subjects who used any fat on bread. These associations were invariable over the 15-y period, as were those with mean BMI, the use of fat on bread being an exception in women. The mean BMI of the women who reported eating bread without fat was much higher than that of fat users in the 1980s, but in the 1990s this difference disappeared.
Consumption of sausages, milk, and sour milk were associated with obesity in both sexes. In the women, consumption of coffee also had a positive association and consumption of bread an inverse association with obesity. In the single-variable models, inverse associations of obesity were observed with consumption of vegetables in both sexes and consumption of tea in the women, whereas in the men, consumption of coffee had a positive association with obesity (Table 3). These associations were constant over the 15-y period, as were those with mean BMI, excluding consumption of milk and sour milk in women. The association of mean BMI with milk consumption in women strengthened over this period, whereas the association with consumption of sour milk was no longer significant in 1997.
The women who reported no alcohol use and the men who consumed 10 portions during the previous week were more likely to be obese than were those who consumed 1–3 portions/wk. In single-variable models, these associations were observed both in the men and the women (Table 3). When alcohol consumption was regarded as a continuous variable, mean BMI increased with increasing alcohol consumption in the men, but decreased in the women. Compared with the reference group, the men who consumed no alcohol at all or 10 portions/wk were more likely to be obese and to be heavier in 1997 than in 1982 (P value for interaction between year and alcohol: 0.012 for obesity and 0.021 for mean BMI). In the women, the associations of alcohol consumption with obesity and mean BMI remained stable over the 15-y period.
Compared with never-smokers, obesity was more prevalent among ex-smokers for both the men and the women, but less prevalent among the female smokers. In the single-variable model, male smokers were more likely to be obese than were never-smokers (Table 3). The association of smoking history with obesity in both sexes and with mean BMI in men remained unchanged over the 15-y period (Figure 4). In women, the trend in mean BMI seemed to increase in smokers and ex-smokers but decrease in never-smokers.
|
). These associations were constant over the 15-y period, but the most prominent increase in mean BMI over time occurred in those with poor perceived health (Figure 5).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Leisure-time physical activity was inversely associated with obesity and mean BMI in both the men and the women. Interestingly, on the basis of repeated cross-sectional surveys, the inverse associations of the level of leisure-time physical activity with BMI significantly strengthened over the 15-y period in both sexes, as did that of weekly time of leisure-time physical activity with BMI and obesity in the women.
Our findings are consistent with the common observation made in cross-sectional surveys that overweight subjects tend to be physically inactive at leisure time (24–26), although in some reports, only weak relations between physical activity and body weight were found (27). Furthermore, in agreement with our results, others observed a tendency for obesity to be less prevalent among men who performed physically heavy work but more prevalent among women who performed heavy work (25).
The more prominent increase in BMI over time observed in the men who performed physically light work could be partly due to changes in sedentariness varying across occupational groups over the 15-y period. In 1982, 26% of the men who performed physically light work and 35% of the men who performed physically heavy work reported being sedentary at leisure time. Fifteen years later, these proportions had decreased to 23% and 21%, respectively. Thus, although heavy work for men may have become less physically demanding, the men seem to have compensated for this decrease in energy expenditure by being more active at leisure time. This would not have been the case for men performing physically light work, not to mention those men not working at all.
The results of a recent cross-sectional study that examined the effect of beliefs and behaviors on physical activity in Australians attempting weight control suggest that the value of physical activity, particularly moderate-intensity activity, is insufficiently recognized, especially by men (28). As shown in the present study, in our earlier studies (29), and in other studies carried out in Finland (30), the proportion of persons reporting leisure-time physical activity has increased. However, the prevalence of obesity in Finnish men in particular has also increased, which may be because the increase in leisure-time physical activity has not been enough to compensate for the decrease in overall physical activity (31).
The methods used to assess habitual physical activity in most studies have been criticized as being crude and imprecise, especially concerning occupational and household activities (32). This was the case in our study as well. We did not include a measure for household activities and the definition used to classify participants according to occupational activity was rather broad. In contrast, our measure for leisure-time physical activity had several components, which were consistent across the surveys. Regardless of the measurement used, however, we observed a strong inverse association of physical activity with obesity and BMI.
Even though also crude in type, the dietary questions could be used as indicators to rank subjects according to food choices. Not unexpectedly, sausage eaters were observed to be heavier than others. Similarly, consumption of vegetables showed an inverse association with obesity. This association, however, was attenuated after other lifestyle variables were controlled for, which may have been because of overcontrol in the statistical analysis. In the men, an interaction between the frequency of sausage consumption and time (P = 0.069, data not shown) suggested that consumption of sausages had become an increasingly more important indicator of BMI. Our findings are in line with a recent study showing that obese subjects appear to consume an energy-dense diet that is particularly associated with salty rather than sweet food items (33). Furthermore, high-fat food items were shown to more likely belong to the top 10 favorite foods of obese subjects, with men preferring meat dishes and women sweet-fat combinations (34).
In contrast with our expectations, persons who avoided the use of fats on bread or preferred skim milk were more obese than were fat users or subjects who drank no milk. Probable explanations for this are that obese subjects try to choose low-fat products and avoid fat to control their weight or that they tend to give socially desirable responses (35–37). These results may also be an indicator of a phenomenon presented by Rolls and Miller (38) that low-fat food choices may give the consumer license to overeat. However, in a recent Italian study, no significant differences were observed between normal-weight and obese subjects in their beliefs and attitudes toward the consumption of fat-containing foods (39).
Carbohydrate as well as fiber intake has been shown to be inversely associated with body weight in cross-sectional surveys (40–42). Our finding of an inverse association between bread consumption and body weight supports this association. Bread counts for the majority of cereal products consumed in Finland, which in turn corresponds to 
45% of total carbohydrate and 62% of fiber intake in the Finnish diet (43).
In contrast with our findings, no association between coffee consumption and body weight was observed in other cross-sectional (44, 45) or follow-up studies (8), although coffee drinking was found to be associated with an unhealthy lifestyle (46). Overall, consumption of nonalcoholic beverages in our study was positively associated with obesity, tea being an exception (with a tendency to an inverse association, especially in women). Drinking tea could be related to general health consciousness and better weight control (46).
Epidemiologic findings on the association between alcohol consumption and body weight are controversial (47). This is hardly surprising when bearing in mind that measurement of alcohol consumption is prone to reporting error and may be influenced by cultural differences (48, 49). Our findings, consistent with some (50) but not all (45) cross-sectional studies, suggest that men and women with low alcohol consumption tend to weigh less than do nondrinkers or subjects with higher alcohol consumption. In agreement with other reports (51), we also found a positive association between alcohol consumption and BMI in the men and a negative association in the women, which may have been due to the much higher number of abstainers among the women than among the men. These findings should, however, be interpreted with caution, because the drinking history of the respondents was not available. Thus, the nondrinker group included both former drinkers and lifetime abstainers with different reasons (eg, health, conviction, etc) for not drinking (52).
In most populations, smokers weigh less than do nonsmokers (13, 53). As suggested by Molarius et al (13), however, this may no longer be true in populations such as Finland's because of extensive antismoking activities and a reduced prevalence of smoking. In Finnish men, the relation between smoking and BMI was reported to change from an inverse association to a positive one in the late 1980s by Marti et al (54), a finding confirmed by other Finnish studies (55). The same tendency was seen in our results. Consistent with the findings of some earlier cross-sectional studies (41, 56, 57), we observed ex-smokers to be heavier than nonsmokers among both the men and the women. In many studies, this association was found in men but not in women (13, 45, 51, 58), whereas in some studies, no association was found in either sex (59).
To conclude, a physically active lifestyle together with abstention from smoking, moderate alcohol consumption, and consumption of healthy foods seems to offer the best chance of avoiding obesity. Avoiding sedentariness has become an even stronger factor during the past decades for the maintenance of normal weight.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Sneve and R. Jorde Cross-sectional study on the relationship between body mass index and smoking, and longitudinal changes in body mass index in relation to change in smoking status: The Tromso Study Scand J Public Health, June 1, 2008; 36(4): 397 - 407. [Abstract] [PDF]
|
|
|
|
 |
|
 K. Borodulin, T. Laatikainen, A. Juolevi, and P. Jousilahti Thirty-year trends of physical activity in relation to age, calendar time and birth cohort in Finnish adults Eur J Public Health, June 1, 2008; 18(3): 339 - 344. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. F. Krebs, J. H. Himes, D. Jacobson, T. A. Nicklas, P. Guilday, and D. Styne Assessment of Child and Adolescent Overweight and Obesity Pediatrics, December 1, 2007; 120(Supplement_4): S193 - S228. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Hu, P. Jousilahti, A. Nissinen, S. Bidel, R. Antikainen, and J. Tuomilehto Coffee consumption and the incidence of antihypertensive drug treatment in Finnish men and women Am. J. Clinical Nutrition, August 1, 2007; 86(2): 457 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. W Gunther, R. M Lyle, P. A Legowski, J. M James, L. D McCabe, G. P McCabe, M. Peacock, and D. Teegarden Fat oxidation and its relation to serum parathyroid hormone in young women enrolled in a 1-y dairy calcium intervention Am. J. Clinical Nutrition, December 1, 2005; 82(6): 1228 - 1234. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. I. G. Holt Obesity - an epidemic of the twenty-first century: an update for psychiatrists J Psychopharmacol, November 1, 2005; 19(6_suppl): 6 - 15. [Abstract] [PDF]
|
|
|
|
 |
|
 J. W. Stewart, K. Koehler, W. Jackson, J. Hawley, W. Wang, A. Au, R. Myers, and D. F. Birt Prevention of mouse skin tumor promotion by dietary energy restriction requires an intact adrenal gland and glucocorticoid supplementation restores inhibition Carcinogenesis, June 1, 2005; 26(6): 1077 - 1084. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. W Gunther, P. A Legowski, R. M Lyle, G. P McCabe, M. S Eagan, M. Peacock, and D. Teegarden Dairy products do not lead to alterations in body weight or fat mass in young women in a 1-y intervention Am. J. Clinical Nutrition, April 1, 2005; 81(4): 751 - 756. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. A. Breslow and B. A. Smothers Drinking Patterns and Body Mass Index in Never Smokers: National Health Interview Survey, 1997-2001 Am. J. Epidemiol., February 15, 2005; 161(4): 368 - 376. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. A. C. van Dijk, L. J. Schouten, L. A. L. M. Kiemeney, R. A. Goldbohm, and P. A. van den Brandt Relation of Height, Body Mass, Energy Intake, and Physical Activity to Risk of Renal Cell Carcinoma: Results from the Netherlands Cohort Study Am. J. Epidemiol., December 15, 2004; 160(12): 1159 - 1167. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Beebe Body Weight Issues in Preventing and Treating Type 2 Diabetes Diabetes Spectr, October 1, 2003; 16(4): 261 - 266. [Full Text] [PDF]
|
|
|
|
 |
|
 M. C Gulliford, D. Mahabir, and B. Rocke Food insecurity, food choices, and body mass index in adults: nutrition transition in Trinidad and Tobago Int. J. Epidemiol., August 1, 2003; 32(4): 508 - 516. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE |
BMI < 30) and obesity (BMI 
