International Journal of Obesity
SEARCH     advanced search my account e-alerts subscribe register
Journal home
Advance online publication
Current issue
Archive
Press releases
For authors
For referees
Contact editorial office
About the journal
For librarians
Subscribe
Advertising
naturereprints
Contact Springer Nature
Customer services
Site features
NPG Subject areas
Access material from all our publications in your subject area:
Biotechnology Biotechnology
Cancer Cancer
Chemistry Chemistry
Dentistry Dentistry
Development Development
Drug Discovery Drug Discovery
Earth Sciences Earth Sciences
Evolution & Ecology Evolution & Ecology
Genetics Genetics
Immunology Immunology
Materials Materials Science
Medical Research Medical Research
Microbiology Microbiology
Molecular Cell Biology Molecular Cell Biology
Neuroscience Neuroscience
Pharmacology Pharmacology
Physics Physics
Browse all publications
 
June 2000, Volume 24, Number 6, Pages 685-694
Table of contents    Previous  Article  Next   [PDF]
Paper
Sociodemographic factors associated with long-term weight gain, current body fatness and central adiposity in Swedish women
P H Lahmann1, L Lissner2, B Gullberg3 and G Berglund1

1Lund University, Department of Medicine, Malmö University Hospital, Malmö, Sweden

2Göteborg University, Department of Internal Medicine, Vasa Hospital, Göteborg, Sweden

3Lund University, Department of Community Medicine, Malmö University Hospital, Malmö, Sweden

Correspondence to: P H Lahmann, The Malmö Diet and Cancer Study, Lund University, Malmö University Hospital, Medical Research Center, S-205 02 Malmö, Sweden. Petra.Lahmann@smi.mas.lu.se

Abstract

OBJECTIVES: To examine sociodemographic factors associated with long-term adult weight gain and current general and central adiposity in women.

DESIGN: Cross-sectional analysis based on data from the Malmö Diet and Cancer prospective cohort Study (MDCS), Sweden.

SUBJECTS: 5464 women aged 45-73 y, who participated in the MDCS between 1994 and 1996.

MEASUREMENTS: Weight change was defined as the difference between measured weight and recalled weight at age 20. Body composition was estimated from bioelectrical impedance analysis. Waist circumference and waist-hip ratio (WHR) were indicators of central obesity. Sociodemographic factors studied included reproductive, lifestyle and socioeconomic characteristics obtained from a questionnaire.

RESULTS: At current age 13% of all women were obese (BMI30), and nearly one-third had a body fat content of >33%. Since age 20, the majority of women (77%) had gained more than 10% of their initial weight. On average, the mean weight change was 12.6±10.0 kg during adult life. Age-adjusted group mean comparisons identified a number of significant sociodemographic variables predicting past weight gain and current obesity. In multivariate analysis, significant independent correlates of weight gain were: age, initial small body size, high parity, early menarche, non-use of hormone therapy, low self-rated health, non-smoking, no consumption of alcohol, sedentary leisure activity, past change in diet, retirement, low education, low socioeconomic status, and low socioeconomic status of origin. Many of the same variables were independently associated with current waist circumference, WHR, and percentage body fat.

CONCLUSIONS: Large weight gain during adult life and a high rate of overweight were observed in this group of Swedish women. Many sociodemographic characteristics were associated with long-term weight gain and recent obesity indices. Initial BMI, smoking, age, physical activity and early socioeconomic status accounted for most of the explained variance in weight change. The findings indicate the importance of identifying environmental determinants of both weight gain and attained fatness, as well as fat distribution.

International Journal of Obesity (2000) 24, 685-694

Keywords

obesity; weight gain; body fat; central adiposity; reproduction; life style; socioeconomic status; health; under-reporting

Introduction

Although the detrimental effects of obesity on health are generally well documented, research has more recently focused on adult weight gain, its risk factors and prevention. Both overweight and body weight fluctuations during adult life are related to chronic disease risk.1,2,3 In particular, high weight gain has been associated with an increased morbidity from breast cancer,4,5 cardiovascular disease in middle6 and later7 adulthood, and diabetes.1 Still, the evidence for the effects of low weight, as well as weight loss or gain is somewhat controversial, with respect to the optimal weight for longevity, and it remains unclear which periods of life might be most crucial concerning weight-related health outcomes in old age.6,7,8 While the evidence for the contribution of environmental or non-genetic factors in the development of obesity is generally strong, the evidence of inherited factors is also clear.9,10

A number of biological, demographic, sociocultural, and behavioral factors are known to be associated with overweight,11 yet less is known about the history of weight change and its determinants in the general population. Age and initial degree of overweight are correlated with weight change over time.12,13,14,15 In women, both age and the onset of menopause influence changes in body mass and body composition, resulting in overall increased body mass index (BMI) or increased total fatness and central adiposity. However, the magnitude of reported changes differs and findings are in part dependent on the methodology used for assessment.16 Furthermore, parity is positively associated with weight gain17 and increase in abdominal fat.18,19

Environmental predictors and determinants of weight gain in women observed in a few previous studies include low level of education, low physical activity, alcohol consumption, being married, quitting smoking, high energy intake, and high fat intake.17,20 Still, results on the association between dietary fat intake and weight gain in adults are not consistent.14,21,22 In women, weight gain has also been related to dieting12 and psychological traits.23

The limited information on patterns of long-term weight change for adult women between young adulthood and older age, and on predictive factors for weight change, led to the following research questions for our study. The primary aim was to describe sociodemographic factors associated with (1) long-term weight gain during adulthood, (2) current body fat, and (3) current central adiposity in women aged 45-73 y, and secondly to examine how the sociodemographic variables differ in relation to the different obesity indices. We were also interested in which sociodemographic factors were independent of others when studied simultaneously in multivariate models.

Methods

Study population

Data from a sub-sample of women aged 45-73 y who participated in the Malmö Diet and Cancer Study (MDCS) were used for this study. The MDCS is a population-based prospective cohort study (1991-1996), designed to examine dietary risk factors for cancer in a population of 30,465 women and men, born between 1923 and 1950. Subjects were randomly invited by letter using the Municipal Registry from the total source population of 74,138 subjects, living in the city of Malmö, in southern Sweden. In addition, subjects were recruited by advertisements by local media and recommendations from participants.

The present analysis is conducted as part of a larger project to investigate weight gain and central adiposity in the etiology of breast cancer. The following inclusion criteria were based on this background. Female participants between December 1994 and December 1996 (n = 7 462) with information on reproductive characteristics and available dietary data were considered eligible for the study sample (n = 7 209). Women reporting surgical menopause, extreme age at menopause (<35 and >65 y), and history of cancer, heart attack, and stroke were excluded. The final analytic cohort included 5464 study women.

Measures

The participants completed a health questionnaire covering an array of sociodemographic, behavioral, and health status characteristics, including two items on weight history. The diet assessment method was a modified diet history method combining a 7 day menu book and a quantitative 168 item food frequency questionnaire for the previous year.24 Measurement of anthropometry was performed during a baseline examination at first visit and each participant was also asked to recall her weight at age 20.

Anthropometric examination. Weight was measured to the nearest 0.1 kg using a balance-beam scale with subjects wearing light clothing and no shoes. Standing height was measured with a fixed stadiometer calibrated in centimeters. Current body mass index (BMI) was calculated as weight in kg divided by height in meters squared (kg/m2). BMI at age 20 was calculated using recalled weight and current height. Weight change was defined as the difference of weight in kg measured at current age and recalled weight at age 20. To further explore weight history, weight change from age 20 was categorized as gain of >10%, stable weight within 10%, or loss of >10%.

Bioelectrical impedance analysis (BIA) was used for estimating body composition according to procedures provided by the manufacturer (BIA 103, RJL-systems, Detroit, MI, USA; single-frequency analyzer). Body fat percentage was derived from lean body mass calculations. Waist and hip circumferences of each participant were measured by a trained nurse (waist, midway between the lowest rib margin and iliac crest; hip, horizontally at the level of the greatest lateral extension of the hips) and used to construct a waist-hip ratio (WHR) as a measure of central obesity.

Predictive variables. Menopausal status: study women had reported the calendar year of last menstruation in the health questionnaire. Based on this information menopause status (pre- or post-menopausal) and age at natural menopause (year of last menses minus year of birth) were determined. Women who reported their last menstrual cycle one year before or during the year of cohort enrollment (n=286) were categorized as pre-menopausal, to ensure that women classified as post-menopausal had had amenorrhea for at least 12 months before the year of interview.

Hormonal therapy: information on current use of exogenous hormones at study entry was obtained from women's self-reporting on medications in the questionnaire and menu book. The type of hormone administration was classified according to the Anatomic Therapeutic Chemical (ATC) classification system25 and included estrogen and progestin regimens, with known systemic effects. For this analysis exposure to hormone therapy was defined as a dichotomous variable, users vs non-users.

Self-rated health: perceived health was originally measured as a seven-category variable. The questionnaire item was worded as follows: 'How do you feel right now, physically and mentally, with respect to your health and well-being? (make a choice between 1 and 7)', with (1) indicating 'feel very bad, could not feel worse' and (7) 'feel very well, could not feel better'. The two lowest levels 1 (n = 34) and 2 (n = 90) were collapsed into one category due to small numbers. Accordingly, self-rated health was used aordinal variable ranging from very bad (1) to very well (6).

EI:BMR ratio: the ratio 'reported energy intake' to 'calculated basal metabolic rate' (EI:BMR) can be used to evaluate the validity of reported energy intake. Estimates of BMR were derived from equations utilizing weight, height and age.26 Subjects were allocated into quartiles of EI:BMR to compare low energy reporters with high energy reporters.

Alcohol consumption: recent alcohol intake was assessed with the 7 day menu book, and estimated as ethanol grams per day. Data on consumption of alcohol was dichotomized as non-users and users.

Physical activity: leisure time physical activity was a composite measure of 18 different leisure time activities as queried in the health questionnaire. A physical activity score was obtained by applying an activity-specific factor and computing the sum of all activity products. The procedure was adopted from a previous validated questionnaire.27 Subjects were ranked into quartiles of physical activity.

Employment status: the original questionnaire item included five response categories. For this analysis, due to small numbers, housewives, students and unemployed were grouped into one category named 'other'. Employment status was analyzed as three-categorical variable (employed/retired/other).

Occupation: study of women's own (current or latest) and parental (breadwinner's) profession was assessed with the questionnaire. Socioeconomic status (SES) was classified according to the Nordic Occupation Classification System.28 For the present analysis the two occupation variables consisted of six categories. The group employer/self-employed comprises a heterogeneous group of professionals with diverse educational background, including among others business owners, free lancers, and farmers.

Other predictor variables: age (in 5 y categories: 45-49, 50-54, 55-59, 60-64, 65-69, 70-73), age at menarche (five levels: £11 (9-11), 12, 13, 14, 15 (15-18)), parity (number of births, six levels), smoking status (current/former/never), past change in diet (dichotomized as non-changer or changer), ethnicity (country of birth, dichotomized as Swedish-born or foreign-born), education (educational achievement in years, five levels).

Statistical analysis

Means and standard deviations were used to describe crude and adjusted baseline values and changes in body measurements within age, menopausal status or initial relative weight groups. Relative weight categories (BMI <25, 25-29, 30) were used according to current WHO recommendations21 and BMI at age 20 was dichotomized at the median (median split) to distinguish women with small and large initial body size. For analyses describing associations of reproductive, lifestyle or sociodemographic characteristics with weight change and obesity indices, an analysis of variance (general linear model) was conducted. Means across predictor variables were adjusted for age (in 5 y categories). To identify independent predictive variables, multivariate regression analysis was performed. Separate models for weight gain, percentage body fat, waist circumference, and WHR were analyzed. All predictive variables were simultaneously included in each of the four regression models, except for marital status and the ratio EI:BMR. Since marital status and living arrangement reflect a similar demographic characteristic and were the least strong correlates of weight change and current obesity indices in age-adjusted univariate analyses, only living arrangement was included in the multivariate analyses. The ratio EI:BMR was used to examine the degree of underreporting of energy intake by weight gain and obesity indices, but was not entered in the final multiple regression models. The explanatory variables in the multiple regression models were used as categorical variables and tests for linear trends included for some of these variables. Pearson and Spearman correlations were also calculated to determine associations between body size indices and predictive variables, as well as associations among explanatory variables.

The Statistical Package for Social Sciences (SPSS for Windows, version 7.5, 1997, Chicago, IL, USA) was used for all statistical analyses. Values of P < 0.05 were considered statistically significant.

Results

The mean age of women at enrollment into the study was 56.6±9.7 y. Forty-eight percent were pre-menopausal and 52% post-menopausal. Median age at menopause was 50.0 y. Participants' average attained weight, height and BMI were 67.9 kg, 1.64 m, and 25.3 kg/m2, respectively. At current age 46% of all women were overweight (BMI25) with 13% of women having a BMI of 30. In contrast, at age 20 only 5% were overweight with less than 1% categorized as obese. Seventy-seven percent were classified as weight gainers, 21% had stable weight within 10%, and only 2% had lost weight since age 20.

The distribution of women and their anthropometry according to age, menopausal status, and initial body size category are presented in Table 1. The mean BMI, percentage body fat, waist circumference and WHR increased with age from the 45-49 group to the 65-69 group, followed by a decrease in the oldest group (70+). Similarly, the mean current weight and past weight change increased with age, but started to decline earlier, ie in the 65-69 group. Height was inversely correlated with age (r = -0.28, P = 0.01). Women's body size also differed by menopausal status. Post-menopausal women had significantly higher attained BMI, percentage body fat, and central adipose tissue compared to pre-menopausal women. Mean past weight change was higher in post-menopausal women as well.

Participants with a small body size at age 20 gained significantly more during adult life than those who had a large initial body size. Despite the higher weight gain, women with a lower initial BMI remained significantly leaner regarding all obesity indices compared to women with a higher BMI at age 20 (Table 1). Additional analyses (results not shown) with initial BMI categorized as tertiles, quartiles or quintiles indicated similar negative trends with past weight change and similar positive trends with current anthropometric measures. As expected, weight change was positively associated with current relative weight categories (P < 0.0001, results not shown). Past weight gain (kg, mean±s.d.) was twice as high in overweight (16.2±6.9) and four times higher in obese (27.6±9.7) women, when compared to their normal weight (BMI <25) counterparts (7.0±6.6).

Individual correlates of weight gain and obesity indices

The age-adjusted means of anthropometric measures in relation to the levels of the cross-sectional correlates are presented in Tables 2,3 and 4. Correlates of long-term weight gain are described first, followed by correlates of body fat and central adiposity.

Parity was the strongest reproductive correlate of weight gain (P < 0.0001; Table 2). Weight gain increased with the number of childbirths, with highest weight gains in those with four or more pregnancies. Age at menarche was inversely associated with weight gain. The mean (s.d.) age of menarche was 13.5 (1.4) y. Comparing participants who experienced menarche at age 11 y or younger vs 15 years or older, there was a difference of 2 kg in absolute weight gain after adjusting for current age. Across current age groups (results not shown) study women differed significantly (P < 0.0001) in their age at menarche, with a range of 13.3-13.9 y between the youngest (45-49 y) and oldest (70-73 y) age group, suggesting a cohort effect.

Notably, menopausal status was significantly associated with weight gain independent of age (Table 2), but in the opposite direction as the non-age adjusted analysis (Table 1). Holding age constant, pre-menopausal women experienced a higher weight gain than post-menopausal women. Also, time since menopause (in years) was negatively correlated with weight gain independent of age (partial r = -0.05, P = 0.017). Current use of exogenous hormones was also predictive of weight gain. Hormone users (17%) had a lower level of weight gain than their counterparts not using hormones. When the association between menopausal status and weight change was controlled for both age and hormone use, the difference in weight gain between pre-menopausal (mean 13.6 kg) and post-menopausal women (mean 12.6 kg) became slightly smaller and only approached significance (P = 0.056).

Among lifestyle factors (Table 3), ex- and non-smoking, low physical activity, no consumption of alcohol, and past change in diet corresponded with higher weight gain. Women who reported a previous change in diet (22%) were more likely to be overweight and obese when compared to women with no change in diet (results not shown). Low energy reporting (EI:BMR ratio) was associated with high weight gain, indicating that underreporting was more common among women with major weight gains. Controlling for age and physical activity simultaneously did not alter the mean values of weight gain across the quartiles of the EI:BMR ratio (results not shown). The majority (74%) of women rated their health as better than average with 19% stating that their health was 'very well, could not be better'. The level of self-rated health was inversely associated with weight gain. There was nearly a 3 kg difference in absolute weight gain between women of the lowest and the two highest levels of self-rated health.

Socioeconomic characteristics (Table 4) were strong correlates of weight gain. Women with low educational attainment or low SES, assessed as their own or parents' occupation, had higher weight gains compared with women from higher educational or higher SES background, respectively. Married and widowed women had larger weight gains when compared to divorced and single women. A similar pattern emerged for living arrangement.

The same analyses were done to assess correlates of current adiposity and fat distribution (Tables 2,3 and 4). Parity was strongly correlated with body fat and fat distribution corresponding to the results found with weight gain (Table 2). In contrast to past weight gain, menopausal status was not associated with current body fatness (percentage body fat) or central adiposity (measured as waist or WHR), when adjusted for age. The lack of associations suggest that the independent effect of menopause transition appears to be more reflected in weight gain than in the actual level of general or regional adipose tissue in pre-menopausal women. This group is approaching menopause, while post-menopausal women have passed the time period of major hormonal changes. Users of hormones and women with late menarche had significantly lower percentage body fat. However, there was no relation between hormone therapy and central adiposity reflected as WHR, and a weaker, though still statistically significant negative association between age at menarche and WHR. After simultaneously controlling for age and current BMI the association between menarcheal age and WHR disappeared. The findings suggest that use of hormones and age at menarche may affect general adiposity, but only to a minor extent central adipose tissue.

Interestingly, smoking status had a different impact on central adiposity as compared to body fat mass or weight gain (Table 3). Current smokers had significantly higher waist-hip ratios than ex- or never-smokers, but had lower percentage body fat or level of weight gain compared to the non-smoking groups. This association also remained highly significant after adjusting for both attained BMI and age.

As observed with weight gain, SES of origin and own SES were highly predictive of current percentage body fat and central adiposity (Table 4). A lower parental or recent occupational background was associated with higher body fatness and central adipose tissue, respectively. Neither marital status nor living arrangement was significantly related to central adiposity reflected as waist circumference or WHR. Ethnicity was predictive of total and central body fatness, but not of weight change. Foreign-born women had higher percentage body fat and increased central adiposity when compared to Swedish-born women. The associations between the remaining correlates and body fatness or central adiposity were similar in direction and significance to the relations with weight gain.

Multivariate analyses

Multivariate analyses were performed to identify independent correlates of weight gain and obesity indices. Regression coefficients, standard errors and P-values for reproductive, lifestyle, and sociodemographic correlates of weight change and obesity indices are shown in Table 5. Since weight change was significantly related to age (r = 0.05, P = 0.01) and initial BMI (r = -0.19, P = 0.01) adjustments for these factors were made by including them in the multivariate model. The same was applied to the models with percentage body fat, waist circumference or WHR as the dependent variable.

Among reproductive characteristics, the independent associations and their directions with weight gain as well as the obesity indices remained unchanged. The difference in weight gain between pre- and post-menopausal women, however, was not significant anymore.

Smoking status, level of leisure time physical activity, alcohol consumption and past change in diet were strong lifestyle correlates in the tested multivariate models. Non-smoking (ex- or never-smoking) was independently associated with an increase in weight gain of approximately 2.5 kg. A few attenuated associations were observed, such as the association between ex-smoking and central adiposity (WHR, non-significant), and alcohol intake and weight gain (P = 0.05).

Among sociodemographic variables, those reflecting SES status were most predictive of weight change and obesity indices when controlled for all the other study variables. An increase of one educational level was associated with a decrease in weight change of 0.5 kg. Similarly, an increase of one level of own or parental occupation was associated with a decrease in weight change of 0.2 or 0.4 kg, respectively, indicating that SES of origin had a stronger influence on weight change in comparison to own occupational background. Moreover, the magnitude of weight change between each level of own occupation was dependent on the level of parental occupation, ie early SES. The most pronounced weight gains were found in the group of women with the lowest early SES across current occupational classes. Within each stratum of parental occupation, the amount of weight gain decreased with each higher level of own occupation and was apparent even for women who had the highest SES of origin. This was tested formally as an interaction between current SES and that of origin, with results suggesting significant effect modification between the two measures (P = 0.02, results not shown). Regarding current body fat mass or distribution (WHR), the effect of original or current SES was similar in magnitude. Furthermore, there was a significant difference in weight gain, body fatness and central adiposity between employed and retired women, suggesting that the transition to retirement results in higher weight gains (1.4 kg) and increased general and central adiposity. Ethnicity remained a significant predictor only for percentage body fat, but not for waist circumference or WHR, and suggestive for weight gain.

Discussion

This sample of 5363 middle-aged and older Swedish women had gained, on average, 12.6 kg during adult life or 0.37 kg/y since age of 20, respectively. Major weight losses had occurred in only a minority of women. This study differs from many other cross-sectional studies in that body composition measurements were used to assess adiposity, rather than body mass index. Moreover, the outcome variables waist circumference and waist-hip ratio were examined separately, as alternative indicators of centralized adiposity.

In the analysis of these data, it was demonstrated that a number of reproductive and environmental factors are related to both weight gain and current weight status. When analyzed simultaneously in a multivariate model, significant independent correlates of weight gain were: advancing age, initial small body size, high parity, early menarche, non-use of hormone therapy, low self-rated health, non-smoking, no consumption of alcohol, sedentary leisure time, past change in diet, retirement, low education, low socioeconomic status, and low SES of origin. These findings confirm and extend previous work that identified determinants of weight gain in Finnish women.17 Many of the same variables in the present study were also associated with attained body fatness and central adiposity. Our findings in respect to fat distribution may also suggest that WHR is poorly explained by the environmental influences examined here (r 2=6%).

Some limitations of these data should be noted. The estimates of weight gain are based on self-reported weights at age 20. The recall period ranged from 25 to 53 y for the youngest and oldest subjects, respectively. While previous studies have shown that self-reported weight at a younger age has good validity, there may remain some unknown biases in the long-term recall of weight history. However, evidence suggests that past body weight is reported with reasonable accuracy in comparison with measured past weight among elderly29 and middle-aged adults,30 even when recalled up to 30 y later. The Pearson correlation coefficient between BMI at age 20 (calculated from recalled weight at age 20 and current height) and measured attained BMI in our study (r = 0.43, P = 0.01) is comparable to the correlation (r = 0.41, P£0.0001) reported for post-menopausal women of the Iowa Women's Health Study.31 Also, there was good concordance between our calculated weight change variable and a separate question in which subjects described how their body weight changed over the years. It should be also pointed out that the statistical significance of the associations reported here reflects in part the relatively large size of the sample. It is clear that the demographic and lifestyle-related variables described here provide an incomplete picture of the determinants of weight gain and adiposity, as evidenced by the fact that explained variances were 11% for weight change and among current obesity indices 6% for WHR, 18% for waist circumference and 20% for body fat. In comparison, among Finnish women17 18% of the overall variation of weight change was accounted for, with weight change assessed for a period of 5 y only and a wider age range (25-64 y) among subjects.

Another issue is the use of dietary data and the problem of obesity-related underreporting in dietary interviews which has been well documented.32 In our study heavier subjects and those who had larger past weight gains displayed a significantly lower ratio of energy intake to predicted basal metabolic rate than non-obese women, even after adjusting for physical activity level. Our findings are in line with recent evidence that underreporting increases as overweight increases. For this reason, it was not considered meaningful to include dietary intake data in the present work without performing a more extensive analysis which was beyond the scope of this paper. Also, given that the overweight women and weight gainers were significantly more likely to report past changes in diet than the other women, even an unbiased assessment of current diet would be unlikely to provide a good explanation for development of a weight problem.

With these caveats, the following aspects of the findings were of particular interest in the context of the natural history of overweight. Menopausal status after adjusting for age was not a strong predictor of the weight measures, and was not significant in the multivariate model. Thus, our study results are not in general agreement with a recent review of Tchernof et al,16 which concluded that central body fatness is increased significantly at menopause, independent of advancing age. When examining the role of hormone therapy clear differences were noted, specifically that users of hormone therapy were significantly leaner and had gained less weight than non-users. In fact, users already had lower weight-for-height than non-users at age 20, possibly reflecting a selection bias with respect to estrogen use. In the same multivariate analysis young age at menarche and parity were both independent predictors of weight gain; however, parity seemed to predict central obesity while low menarcheal age seemed more predictive of percentage body fat. Previous evidence on the relation between early menarche and general or central adiposity is equivocal,18,19,31,33 while previous reports, including our study, have consistently shown substantial weight gain17 and increased WHR18 to be associated with childbearing.

A number of other lifestyle characteristics were associated with the obesity indices even adjusted for reproductive factors described above. For instance, alcohol users appeared to have both less body fat and less central fat. Overall, these results are consistent with previous evidence from cross-sectional studies, showing that the relation between alcohol intake and adiposity is generally negative for women.34 Smoking status was highly predictive of past weight change and adiposity. Although current smokers had a significantly lower weight gain, and body fat mass than ex- or never-smokers, they had significantly higher WHR. Similar observations of greater abdominal circumferences relative to hip circumferences in female current smokers were described in previous cross-sectional analyses.31,33,35 Additionally, in the present study, ex-smokers appeared to represent an intermediate group with percentage body fat and WHR.

Finally, several social variables also predicted body fat and weight gain in the multivariate model that included lifestyle and reproductive variables. For instance, Swedes had lower body fat content than non-Swedes, but no differences in fat distribution were observed. Similarly, living alone appeared to be more associated with body fatness than with fat distribution. Occupational variables related both to weight gain, current adiposity and fat patterning, with lower socioeconomic groups having higher values. Few previous studies of SES and obesity36,37 or health38 have found that early and late SES both have an independent predictive value for obesity. The question of whether they represent separate influences has rarely been addressed and if so, only in regard to BMI.38,39 Interestingly, in this study, SES of origin, as ascertained by the occupation of the parent who had been breadwinner, was a strong predictor of general obesity, central adiposity and weight gain, even after controlling for the woman's own socioeconomic status and a number of other background characteristics. In addition, the interaction between parental and own occupation in relation to weight change demonstrates that social mobility significantly affects the magnitude of weight change in this group of Swedish women, particularly among low socioeconomic classes. This indicates the importance of early influences, as well as proximal ones, in the etiology of obesity and weight gain.

In conclusion, many sociodemographic characteristics were associated with long-term weight gain and recent obesity indices in this large group of Swedish women. These results may become increasingly relevant as this cohort of women is followed for disease. Previous investigations of weight in relation to chronic disease have frequently not been designed to distinguish effects of long-term vs short-term obesity. Those studies which have, however, collected remotely recalled weight have been able to demonstrate independent effects of attained weight and weight gain on, for instance, breast cancer,4,5 and cardiovascular disease.6,7,40 This indicates the importance of identifying environmental determinants of both weight gain and attained fatness, as well as fat distribution.

Acknowledgements

The authors would like to thank Elisabet Wirfält and Irene Mattisson for their helpful input and Sivert Carlsson and Ghassan Salameh for technical assistance. This work was supported by grants from the Swedish Cancer Society (2684-B93-05XAA) and the Swedish Medical Research Council (B93-39X-09534-03C).

References

1 World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation on obesity, Geneva, 3-5 June 1997. WHO/NUT/NCD/98.1. WHO: Geneva, 1998. ,

2 Rippe JM, Crossley S, Ringer R. Obesity as a chronic disease: modern medical and lifestyle management. J Am Diet Assoc 1998; 98: S9-S15, MEDLINE

3 French SA, Jeffery RW, Folsom AR, Williamson DF, Byers T. Relation of weight variability and intentionality of weight loss to disease history and health-related variables in a population-based sample of women aged 55-69 years. Am J Epidemiol 1995; 142: 1306-1314. MEDLINE

4 Magnusson C, Baron J, Persson I, Wolk A, Bergström R, Trichopoulous D, Adami HO. Body size in different periods of life and breast cancer risk in post-menopausal women. Int J Cancer 1998; 76: 29-34. MEDLINE

5 Barnes-Josiah D, Potter JD, Sellers TA, Himes JH. Early body size and subsequent weight gain as predictors of breast cancer incidence (Iowa, United States). Cancer Causes Control 1995; 6: 112-118. MEDLINE

6 Manson JE, Willett WC, Stampfer MJ, Colditz GA, Hunter DJ, Hankinson SE, Hennekens CH, Speizer FE. Body weight and mortality among women. N Engl J Med 1995; 333: 677-685. MEDLINE

7 Harris TB, Savage PJ, Tell GS, Haan M, Kumanyika S, Lynch JC. Carrying the burden of cardiovascular risk in old age: associations of weight and weight change with prevalent cardiovascular disease, risk factors, and health status inthe Cardiovascular Health Study. Am J Clin Nutr 1997; 66: 837-844. MEDLINE

8 Tayback M, Kumanyika S, Chee E. Body weight as a risk factor in the elderly. Arch Intern Med 1990; 150: 1065-1072. MEDLINE

9 Bouchard C. Current understanding of the etiology of obesity: genetic and nongenetic factors. Am J Clin Nutr 1991; 53: 1561S-1565S. MEDLINE

10 Stunkard AJ, Harris JR, Pedersen NL, McClearn GE. The body mass index of twins who have been reared apart. N Engl J Med 1990; 322: 1483-1487. MEDLINE

11 Seidell JC, Flegal KM. Assessing obesity: classification and epidemiology. Br Med Bull 1997; 53: 238-252. MEDLINE

12 French SA, Jeffery RW, Forster JL, McGovern PG, Kelder SH, Baxter JE. Predictors of weight change over two years among a population of working adults: the Healthy Worker Project. Int J Obes Relat Metab Disord 1994; 18: 145-154. MEDLINE

13 Stevens J, Knapp RG, Keil JE, Verdugo RR. Changes in body weight and girths in black and while adults studied over a 25 year interval. Int J Obes 1991; 15: 803-808. MEDLINE

14 Colditz GA, Willett WC, Stampfer MJ, London SJ, Segal MR, Speizer FE. Patterns of weight change and their relation to diet in a cohort of healthy women. Am J Clin Nutr 1990; 51: 1100-1105. MEDLINE

15 Rissanen A, Heliövaara M, Aromaa A. Overweight and anthropometric changes in adulthood: a prospective study of 17 000 Finns. Int J Obes 1988; 12: 391-401. MEDLINE

16 Tchernof A, Poehlman ET. Effects of the menopause transition on body fatness and body fat distribution. Obes Res 1998; 6: 246-254. MEDLINE

17 Rissanen AM, Heliövaara M, Knekt P, Reunanen A, Aromaa A. Determinants of weight gain and overweight in adult Finns. Eur J Clin Nutr 1991; 45: 419-430. MEDLINE

18 Björkelund C, Lissner L, Andersson S, Lapidus L, Bengtsson C. Reproductive history in relation to relative weight and fat distribution. Int J Obes Relat Metab Disord 1996; 20: 213-219. MEDLINE

19 Troisi RJ, Wolf AM, Manson JE, Klingler KM, Colditz GA. Relation of body fat distribution to reproductive factors in pre- and postmenopausal women. Obes Res 1995; 3: 143-151. MEDLINE

20 Klesges RC, Klesges LM, Haddocck CK, Eck LH. A longitudinal analysis of the impact of dietary intake and physical activity on weight change in adults. Am J Clin Nutr 1992; 55: 818-822. MEDLINE

21 Seidell JC. Dietary fat and obesity: an epidemiologic perspective. Am J Clin Nutr 1998; 67: 546S-550S. MEDLINE

22 Lissner L, Heitmann B. Dietary fat and obesity: evidence from epidemiology. Eur J Clin Nutr 1995; 49: 79-90. MEDLINE

23 Korkeila M, Kaprio J, Rissanen A, Koskenvuo M, Sörensen TIA. Predictors of major weight gain in adult Finns: stress, life satisfaction and personality traits. Int J Obes Relat Metab Disord 1998; 22: 949-957. MEDLINE

24 Riboli E, Elmståhl S, Saracci R, Gullberg B, Lindgärde F. The Malmö Food Study: Validity of two dietary assessment methods for measuring nutrient intake. Int Journal of Epidemiology 1997; 26: S161-S173 ,

25 Capella D. Descriptive tools and analysis. In: Dukes MNG (ed). Drug utilisation studies. Methods and uses, 45thedn. WHO Regional Office for Europe: Copenhagen, 1993, pp 55-78.

26 FAO, WHO and UNU. Report of joint expert consultation: Energy and protein requirements. Technical Report Series 724. WHO Geneva, 1985. ,

27 Taylor HL, Jacobs Jr DR, Schucker B, Knudsen J, Leon AS, Debacker G. A questionnaire for the assessment of leisure time physical activities. J Chronic Dis 1978; 31: 741-755. MEDLINE

28 Statistics Sweden. Occupations in Population and Housing Census 1985. Statistics Sweden: Stockholm, 1985. ,

29 Stevens J, Keil JE, Waid R, Gazes PC. Accuracy of current, 4-year, and 28-year self-reported body weight in an elderly population. Am J Epidemiol 1990; 132: 1156-1163. MEDLINE

30 Casey VA, Dwyer JT, Berkey CS, Coleman KA, Gardner J, Valadian I. Long-term memory of body weight and past weight satisfaction: a longitudinal follow-up study. Am J Clin Nutr 1991; 53: 1493-1498. MEDLINE

31 Kaye S, Folsom AR, Prineas R, Potter JD, Gabstur S. The association of body fat distribution with lifestyle and reproductive factors in a population study of postmenopausal women. Int J Obes 1990; 14: 583-591. MEDLINE

32 Goldberg GR, Black AE. Assessment of the validity of reported energy intakes¾review and recent developments. Scand J Nutr 1998; 42: 6-9.

33 Tonkelaar den I, Seidell JC, Noord van PAH, Baanders-van Halewijn EA, Ouwehand IJ. Fat distribution in relation to age, degree of obesity, smoking habits, parity and estrogen use: a cross-sectional study in 11825 Dutch women participating in the DOM-Project. Int J Obes 1990; 14: 753-761. MEDLINE

34 World Health Organization. Physical status: the use and interpretation of anthropometry. WHO Technical Report Series 854. WHO: Geneva, 1995,, pp 312-344.

35 Lissner L, Bengtsson C, Lapidus L, Björkelund C. Smoking initiation and cessation in relation to body fat distribution based on data from a study of Swedish women. Am J Public Health 1992; 82: 273-275. MEDLINE

36 Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105: 260-275. MEDLINE

37 Goldblatt PB, Moore ME, Stunkard AJ. Social factors in obesity. JAMA 1965; 192: 1039.

38 van de Mheen H, Stronks K, Looman CWN, Mackenbach JP. Does childhood socioeconomic status influence adult health through behavioral factors? Int J Epidemiol 1998; 27: 431-437. MEDLINE

39 Braddon FEM, Rodgers B, Wadsworth MEJ, Davies JMC. Onset of obesity in a 36-year birth cohort study. Br Med J 1986; 293: 299-303.

40 Willett WC, Manson JE, Stampfer MJ, Colditz GA, Rosner B, Speizer FE, Hennekens CH. Weight, Weight change, and coronary heart disease in women. JAMA 1995; 273: 461-465. MEDLINE

Tables

Table 1 Relation of age, menopausal status, and initial body size to past weight change and current anthropometric measures in 5454 women

Table 2 Reproduction-related correlates of long-term weight change, current body fatness and central adiposity (age-adjusted)

Table 3 Lifestyle-related correlates of long-term weight change, current body fatness and central adiposity (age-adjusted)

Table 4 Socioeconomic-related correlates of long-term weight change, current body fatness and central adiposity (age-adjusted)

Table 5 Correlates (age-adjusted) of long-term weight change, current body fat and central adiposity derived from multiple linear regression analysis

Received 4 June 1999; revised 15 November 1999; accepted 5 January 2000
June 2000, Volume 24, Number 6, Pages 685-694
Table of contents    Previous  Article  Next    [PDF]