Obesity Research & Clinical Practice (2014) 8, e163—e171
ORIGINAL ARTICLE
Associations between adolescent and adult socioeconomic status and risk of obesity and overweight in Danish adults Sinead M. Boylan a,∗, Timothy P. Gill a, Helle Hare-Bruun b, Lars B. Andersen c,d, Berit L. Heitmann a,b,c a
Boden Institute for Obesity, Nutrition, Exercise and Eating Disorders, University of Sydney, NSW 2006, Australia b Research Centre for Prevention and Health, Glostrup University Hospital, DK-2600 Glostrup, Denmark c Institute of Sports Science and Clinical Biomechanics, and National Institute of Public Health, University of Southern Denmark, DK-5230 Odense M, Denmark d Department of Sports Medicine, Norwegian School of Sport Sciences, Norway Received 13 December 2012 ; received in revised form 21 March 2013; accepted 24 March 2013
KEYWORDS Obesity; Overweight; Socioeconomic
∗
Summary Background: It has been suggested that socioeconomic status (SES) may influence the risk of obesity; however it is important to consider individual changes in SES over the life-course in addition to SES at specific time-points to better understand the complex associations with obesity. We explored the relationship between lifetimespecific and life-course SES and risk of obesity and overweight in Danish adults. Methods: Data were used from the Danish Youth and Sports Study (DYSS) — a 20—22 year follow-up study of Danish teenagers born between 1964 and 1969. Baseline data gathered in 1983 and 1985 included self-reported BMI, SES and physical activity. The follow-up survey (2005) repeated these assessments in addition to an assessment of diet. Complete data on adolescent and adult SES and BMI were available for 623 participants. Results: Following adjustments, adolescent SES had no significant association with overweight/obesity in this sample, however females of low or medium adult SES were significantly more likely to be overweight/obese compared to those of high SES (low SES: OR: 2.7; 95% CI: (1.3—5.8); p = 0.008; medium SES: OR: 4.0, 95% CI (1.6—10.2); p = 0.003). Females who decreased in SES during adulthood were significantly more likely to be overweight/obese compared to those who remained of high SES (OR: 3.1; 95% CI (1.1—9.2); p = 0.04).
Corresponding author. Tel.: +61 2 9036 3006; fax: +61 2 9036 3184. E-mail address: [email protected] (S.M. Boylan).
1871-403X/$ — see front matter © 2013 Asian Oceanian Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.orcp.2013.03.006
e164
S.M. Boylan et al. Conclusion: Effects of early life-factors may be conditional upon the environment in adulthood, particularly for the women. Further research should consider the timing of SES exposure and the mechanisms which may be responsible for the socioeconomic gradients in prevalence of obesity and overweight. © 2013 Asian Oceanian Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved.
Introduction The socio-economic gradients in health are of great public health concern [1]. The last couple of decades have seen a plethora of literature on socioeconomic differences in health outcomes such as coronary heart disease, stroke and all-cause mortality [2,3]. Obesity is believed to be one of the mechanisms responsible for such associations; therefore the socioeconomic gradients among obesity rates have been extensively examined in recent years. One of the largest bodies of evidence, a review by Sobal and Stunkard [4] found consistently inverse associations between socioeconomic status (SES) and obesity for women in developed societies, and a higher likelihood of obesity among those of higher SES in developing societies. The association between obesity and SES has been supported by more recent reviews [5,6] and there is now reasonable evidence that associations between SES and obesity may vary according to gender. Stronger associations are found among females from high income countries compared to males [4,7]. The relationship between obesity and SES may be bi-directional as obesity may influence SES, however SES may also influence the risk of obesity [8]. It has become widely recognized that social and biological factors that have developed over the lifecourse may influence adult health status and this has led to an increased interest in the life-course determinants of obesity [9—11]. Examining SES at a single time-point in life does not take into account the temporal nature of this association and may only partly explain the contributions of socioeconomic factors to health status or how these change over time [12]. In an attempt to best capture the individual cumulative and dynamic nature of SES, several models have been developed to examine life-course SES. The sensitive-periods model recognize that individuals may be more vulnerable to low SES exposure during certain time-points, leading to permanent (critical period) or modifiable (sensitive period) changes in disease risk [13]. Indeed, a recent meta-analysis found inverse associations between childhood socioeconomic status (SES) and
adult obesity, however effect sizes were typically reduced when adjustments were made for adult SES (education) [7]. Similarly, other studies which have adjusted for adulthood SES typically find reduced effect sizes. This suggests that adulthood SES is important in explaining the observed associations between childhood SES and obesity. The social-mobility model acknowledges that SES is dynamic and it incorporates the trajectory of socioeconomic mobility across one’s lifetime in determining disease risk [13]. Studies examining the effects of SES trajectory have produced inconsistent results with negative effects found among both the downwardly [9,14—16] and upwardly social mobile individuals [17]. Others did not find any association between social mobility and BMI or waist-hip ratio [18]. It may be the accumulation of multiple adverse circumstances experienced by lower SES groups throughout the life-course which lead to social inequalities in health [19]. This accumulation of risk model is the summed effect of SES that interacts to increase disease risk across a person’s life-course [13], and may therefore help understand the mechanisms by which social inequities in health develop and are maintained. However, there has been little examination of individual cumulative influences of adverse circumstances during a lifetime on social inequalities in obesity [20,21]. Irrespective of model choice, it is important to consider the range of confounders which exist in the obesity—SES relationship, particularly parental and childhood/adolescent BMI as it is well established that adolescent and parental obesity influences the development of obesity later in life [22]. Our hypothesis is that during a lifetime, SES influences the risk of adult obesity and overweight. Therefore, the aims of this current study were to explore the relationship between lifetimespecific and life-course SES and risk of adult obesity and overweight by (a) examining the relationship between adolescent and adult SES and the risk of adult obesity and overweight; (b) determining the impact of individual SES mobility on the risk of adult obesity and overweight; (c) evaluating the extent to which individual cumulative life-course SES is
Socioeconomic status and risk of obesity and overweight in Danish adults associated with the risk of adult obesity and overweight; and (d) exploring whether associations are gender-specific.
Subjects and methods Ethics statement The study is in accordance with the Helsinki Declaration and was approved by the Data Protection Agency (2005-2311-0111). All participants gave written informed consent.
Data collection This study used data from the Danish Youth and Sports Study (DYSS) which is a 20—22 year follow-up study of Danish teenagers born between 1964 and 1969. Details of this study is discussed elsewhere [23], however in brief, baseline data was gathered 1983 from students in gymnasium, vocational or trade school and in 1985 from teenagers drawn from the Danish Civil Registration system (n = 3008). The data gathered consisted of, among other measures, a general questionnaire including questions on self-reported height and weight from which body mass index (BMI) was calculated, SES and physical activity. Physical activity was based on information on leisure time physical activity, sports, and moderate and high intensity gardening. Each sport and activity was assigned a metabolic equivalent (MET) — details of this method have been published elsewhere [24]. The follow-up survey in 2005 consisted of the same questionnaire, a validated 195-item food frequency questionnaire [25] and a cheek scraping for DNA analysis. Parental BMI was calculated from heights and weights recorded by subjects at follow-up. At the follow-up survey, it was not possible to trace all participants from the baseline survey, however of the 1904 eligible persons invited to participate in the follow-up survey, 786 (41%) chose to participate. The data are available from the authors upon request.
SES assessment SES was assessed at baseline hence reducing the risk of introducing recall bias and was based on an open ended question on paternal occupation and education. If no SES could be assigned to the father, then maternal SES was used. Follow-up SES was based on the participant’s own occupation and education. SES was categorized into five categories based on both occupation and education as follows: SES
e165
group 1: >4 years further education (master level), white-collar workers with >50 subordinates, and self-employed with >20 employees); SES group 2: short or medium further education (up to 4 years), white collar workers with 11—50 subordinates (if not master level education), and self-employed with 6—20 employees (if not master level education); SES group 3: short education (up to 1 year), vocational education, white-collar workers with 1—10 subordinates (if not education corresponding to SES group 1 or 2), and self-employed with 0—5 employees (if not education corresponding to SES group 2); SES group 4: skilled workers, and white-collar workers with no subordinates (if not education corresponding to SES group 1, 2 or 3); SES group 5: semi-skilled and unskilled workers. SES groups were then further categorized as high (groups 1 and 2), medium (group 3) and low SES (groups 4 and 5). Life-course SES frameworks: Life-course frameworks based on those developed for the Framingham Offspring Study were developed and examined [26]. The sensitive-periods framework calculated the effect of SES measured during each period individually. The social-mobility framework examined life course trajectories. Four potential trajectories were investigated: stable low SES (low adolescent and low adulthood), decreasing SES (high adolescent and low adulthood), increasing SES (low adolescent and high adulthood) and stable high SES (high adolescent and high adulthood). An accumulation-of-risk model was created by summing scores for adolescent SES and adult SES. The cumulative SES scores ranged from 2 to 6 were categorized into high (score = 2), medium (score = 3—4) and low (scores = 5—6) cumulative SES similar to other studies [26—28].
Statistical methods Statistics were conducted using SPSS v19.0. The effect of adolescent and adult SES, social mobility and accumulation on adult overweight (BMI: 25—30 kg/m2 ) and obesity (BMI: >30 kg/m2 ) were assessed by logistic regression. Models examining the relationship between adolescent SES, adulthood SES and adulthood BMI mutually adjusted for adolescent or adult SES. In addition, all models were adjusted for adolescent BMI (kg/m2 ), parental BMI (categorical predictor: no parents overweight/obese, 1 parent overweight/obese, 2 parents overweight/obese), adult smoking (current smoker/non-smoker), energy intake (KJ) and physical activity. Although the log-likelihood ratio of the unadjusted model (170.0) did not differ much from the model adjusting for smoking energy intake,
e166
S.M. Boylan et al.
and physical activity alone (169.7), these covariates were included in the models as they have been suggested to be partly responsible for the association between SES and BMI [29—32]. Mis-reporters of energy intake (n = 76) were excluded based on a method described elsewhere [33,34]. Birth weight (grams) did not significantly contribute to the models nor was it significantly associated with adulthood BMI (Spearman’s correlation coefficient: r = 0.03, p = 0.61) therefore it was not included in the logistic regression models.
Results Sample characteristics Following exclusion of mis-reporters and those with incomplete data on adult BMI, adolescent and adulthood SES, the final sample consisted of 623 participants (269 males and 354 females), the characteristics of which are presented in Table 1. The mean (SD) age of the sample at baseline was 16.9 (1.0) years and at follow-up was 37.7 (1.4) years. The mean (SD) daily energy intake was 10.5 (2.6) MJ. The subjects were inactive with low mean (SD) MET totals 0.27 (0.23) evident and the majority of them were current non-smokers (79%). More adult males (39%) than females (20%) were classified as overweight, however the prevalence of obesity was similar for females (7%) and males (7%). The prevalence of combined obesity and overweight across SES categories are presented in Table 2. Approximately one-third of the sample was categorized with a low adolescent SES and almost 40% of the sample increased in SES since that time period.
Table 1
Socio-economic status, adulthood obesity and overweight Among females, adolescent, adult, cumulative SES and social mobility were linearly associated with overweight/obesity (Table 2). There were a significantly higher percentage of overweight/obese females from low adult SES compared to those of high adult SES. The small sample size made it difficult to reliably examine the association between SES and obesity, however, the data indicated there were more obese participants with a low adolescent SES (8% males; 9% females) than with a high adolescent SES (4% males; 3% females) and a significantly (p = 0.002) higher percentage of obese females with a low adult SES (12.5%) than a high adult SES (3%). In addition, there were a significantly (p = 0.01) higher proportion of obese females from a low stable SES (10.9%) compared to those from a stable high SES (2%) and a significantly (p = 0.001) higher proportion of obese females with a low cumulative SES (14%) than a high cumulative SES (2%). Table 2 also shows the unadjusted and adjusted relationships between adolescent and adult SES, and combined adult overweight and obesity. After adjusting for adult energy intake, physical activity, smoking, adolescent and parental BMI and adult or adolescent SES, adolescent SES had no significant association with overweight/obesity in this sample, however females of low or medium adult SES were significantly more likely to be overweight/obese compared to those of high adult SES. Following adjustment for the aforementioned covariates, females who decreased in SES were significantly more likely to be overweight/obese compared to those who remained of high SES.
Characteristics of study sample (n = 623).
n Mean age (years) (SD) Baseline Follow-up Follow-up BMI (%) 35 Smoker (%) Yes No Mean daily energy intake (MJ) (SD) Mean MET total (SD)
Males
Females
Total
269
354
623
16.9 (1.0) 37.6 (1.4)
17.0 (1.0) 37.7 (1.4)
16.9 (1.0) 37.7 (1.4)
0 54 39 6 1
3 70 20 4.5 2.5
2 63 28 5 2
20 80 11.7 (2.6) 0.29 (0.23)
21 79 9.5 (2.2) 0.25 (0.23)
21 79 10.5 (2.6) 0.27 (0.23)
Associations between SES and adult obesity/overweight. Males (n = 269) na
Females (n = 354)
Obese or overweight %
Unadjusted OR (95% CI)
Adjusted OR (95% CI)
n
Obese or overweight %
Unadjusted OR (95% CI)
Adjusted OR (95% CI)
Adolescent SES Low Medium High Trend p-value
86 83 100
49 48 42 0.3
1.3 (0.7—2.3) 1.3 (0.7—2.3) 1.0 ref
0.9 (0.4—1.9) 0.8 (0.4—1.7) 1.0 ref
137 101 116
33 26 21 0.03
1.9 (1.1—3.3) 1.3 (0.7—2.5) 1.0 ref
0.7 (0.3—1.5) 0.9 (0.4—2.0) 1.0 ref
Adult SES Low Medium High Trend p-value
63 49 157
52 55 41 0.07
1.6 (0.9—2.9) 1.8 (0.9—3.4) 1.0 ref
1.5 (0.7—3.3) 1.3 (0.5—3.3) 1.0 ref
104 56 194
40 34 18
ORIGINAL ARTICLE
Associations between adolescent and adult socioeconomic status and risk of obesity and overweight in Danish adults Sinead M. Boylan a,∗, Timothy P. Gill a, Helle Hare-Bruun b, Lars B. Andersen c,d, Berit L. Heitmann a,b,c a
Boden Institute for Obesity, Nutrition, Exercise and Eating Disorders, University of Sydney, NSW 2006, Australia b Research Centre for Prevention and Health, Glostrup University Hospital, DK-2600 Glostrup, Denmark c Institute of Sports Science and Clinical Biomechanics, and National Institute of Public Health, University of Southern Denmark, DK-5230 Odense M, Denmark d Department of Sports Medicine, Norwegian School of Sport Sciences, Norway Received 13 December 2012 ; received in revised form 21 March 2013; accepted 24 March 2013
KEYWORDS Obesity; Overweight; Socioeconomic
∗
Summary Background: It has been suggested that socioeconomic status (SES) may influence the risk of obesity; however it is important to consider individual changes in SES over the life-course in addition to SES at specific time-points to better understand the complex associations with obesity. We explored the relationship between lifetimespecific and life-course SES and risk of obesity and overweight in Danish adults. Methods: Data were used from the Danish Youth and Sports Study (DYSS) — a 20—22 year follow-up study of Danish teenagers born between 1964 and 1969. Baseline data gathered in 1983 and 1985 included self-reported BMI, SES and physical activity. The follow-up survey (2005) repeated these assessments in addition to an assessment of diet. Complete data on adolescent and adult SES and BMI were available for 623 participants. Results: Following adjustments, adolescent SES had no significant association with overweight/obesity in this sample, however females of low or medium adult SES were significantly more likely to be overweight/obese compared to those of high SES (low SES: OR: 2.7; 95% CI: (1.3—5.8); p = 0.008; medium SES: OR: 4.0, 95% CI (1.6—10.2); p = 0.003). Females who decreased in SES during adulthood were significantly more likely to be overweight/obese compared to those who remained of high SES (OR: 3.1; 95% CI (1.1—9.2); p = 0.04).
Corresponding author. Tel.: +61 2 9036 3006; fax: +61 2 9036 3184. E-mail address: [email protected] (S.M. Boylan).
1871-403X/$ — see front matter © 2013 Asian Oceanian Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.orcp.2013.03.006
e164
S.M. Boylan et al. Conclusion: Effects of early life-factors may be conditional upon the environment in adulthood, particularly for the women. Further research should consider the timing of SES exposure and the mechanisms which may be responsible for the socioeconomic gradients in prevalence of obesity and overweight. © 2013 Asian Oceanian Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved.
Introduction The socio-economic gradients in health are of great public health concern [1]. The last couple of decades have seen a plethora of literature on socioeconomic differences in health outcomes such as coronary heart disease, stroke and all-cause mortality [2,3]. Obesity is believed to be one of the mechanisms responsible for such associations; therefore the socioeconomic gradients among obesity rates have been extensively examined in recent years. One of the largest bodies of evidence, a review by Sobal and Stunkard [4] found consistently inverse associations between socioeconomic status (SES) and obesity for women in developed societies, and a higher likelihood of obesity among those of higher SES in developing societies. The association between obesity and SES has been supported by more recent reviews [5,6] and there is now reasonable evidence that associations between SES and obesity may vary according to gender. Stronger associations are found among females from high income countries compared to males [4,7]. The relationship between obesity and SES may be bi-directional as obesity may influence SES, however SES may also influence the risk of obesity [8]. It has become widely recognized that social and biological factors that have developed over the lifecourse may influence adult health status and this has led to an increased interest in the life-course determinants of obesity [9—11]. Examining SES at a single time-point in life does not take into account the temporal nature of this association and may only partly explain the contributions of socioeconomic factors to health status or how these change over time [12]. In an attempt to best capture the individual cumulative and dynamic nature of SES, several models have been developed to examine life-course SES. The sensitive-periods model recognize that individuals may be more vulnerable to low SES exposure during certain time-points, leading to permanent (critical period) or modifiable (sensitive period) changes in disease risk [13]. Indeed, a recent meta-analysis found inverse associations between childhood socioeconomic status (SES) and
adult obesity, however effect sizes were typically reduced when adjustments were made for adult SES (education) [7]. Similarly, other studies which have adjusted for adulthood SES typically find reduced effect sizes. This suggests that adulthood SES is important in explaining the observed associations between childhood SES and obesity. The social-mobility model acknowledges that SES is dynamic and it incorporates the trajectory of socioeconomic mobility across one’s lifetime in determining disease risk [13]. Studies examining the effects of SES trajectory have produced inconsistent results with negative effects found among both the downwardly [9,14—16] and upwardly social mobile individuals [17]. Others did not find any association between social mobility and BMI or waist-hip ratio [18]. It may be the accumulation of multiple adverse circumstances experienced by lower SES groups throughout the life-course which lead to social inequalities in health [19]. This accumulation of risk model is the summed effect of SES that interacts to increase disease risk across a person’s life-course [13], and may therefore help understand the mechanisms by which social inequities in health develop and are maintained. However, there has been little examination of individual cumulative influences of adverse circumstances during a lifetime on social inequalities in obesity [20,21]. Irrespective of model choice, it is important to consider the range of confounders which exist in the obesity—SES relationship, particularly parental and childhood/adolescent BMI as it is well established that adolescent and parental obesity influences the development of obesity later in life [22]. Our hypothesis is that during a lifetime, SES influences the risk of adult obesity and overweight. Therefore, the aims of this current study were to explore the relationship between lifetimespecific and life-course SES and risk of adult obesity and overweight by (a) examining the relationship between adolescent and adult SES and the risk of adult obesity and overweight; (b) determining the impact of individual SES mobility on the risk of adult obesity and overweight; (c) evaluating the extent to which individual cumulative life-course SES is
Socioeconomic status and risk of obesity and overweight in Danish adults associated with the risk of adult obesity and overweight; and (d) exploring whether associations are gender-specific.
Subjects and methods Ethics statement The study is in accordance with the Helsinki Declaration and was approved by the Data Protection Agency (2005-2311-0111). All participants gave written informed consent.
Data collection This study used data from the Danish Youth and Sports Study (DYSS) which is a 20—22 year follow-up study of Danish teenagers born between 1964 and 1969. Details of this study is discussed elsewhere [23], however in brief, baseline data was gathered 1983 from students in gymnasium, vocational or trade school and in 1985 from teenagers drawn from the Danish Civil Registration system (n = 3008). The data gathered consisted of, among other measures, a general questionnaire including questions on self-reported height and weight from which body mass index (BMI) was calculated, SES and physical activity. Physical activity was based on information on leisure time physical activity, sports, and moderate and high intensity gardening. Each sport and activity was assigned a metabolic equivalent (MET) — details of this method have been published elsewhere [24]. The follow-up survey in 2005 consisted of the same questionnaire, a validated 195-item food frequency questionnaire [25] and a cheek scraping for DNA analysis. Parental BMI was calculated from heights and weights recorded by subjects at follow-up. At the follow-up survey, it was not possible to trace all participants from the baseline survey, however of the 1904 eligible persons invited to participate in the follow-up survey, 786 (41%) chose to participate. The data are available from the authors upon request.
SES assessment SES was assessed at baseline hence reducing the risk of introducing recall bias and was based on an open ended question on paternal occupation and education. If no SES could be assigned to the father, then maternal SES was used. Follow-up SES was based on the participant’s own occupation and education. SES was categorized into five categories based on both occupation and education as follows: SES
e165
group 1: >4 years further education (master level), white-collar workers with >50 subordinates, and self-employed with >20 employees); SES group 2: short or medium further education (up to 4 years), white collar workers with 11—50 subordinates (if not master level education), and self-employed with 6—20 employees (if not master level education); SES group 3: short education (up to 1 year), vocational education, white-collar workers with 1—10 subordinates (if not education corresponding to SES group 1 or 2), and self-employed with 0—5 employees (if not education corresponding to SES group 2); SES group 4: skilled workers, and white-collar workers with no subordinates (if not education corresponding to SES group 1, 2 or 3); SES group 5: semi-skilled and unskilled workers. SES groups were then further categorized as high (groups 1 and 2), medium (group 3) and low SES (groups 4 and 5). Life-course SES frameworks: Life-course frameworks based on those developed for the Framingham Offspring Study were developed and examined [26]. The sensitive-periods framework calculated the effect of SES measured during each period individually. The social-mobility framework examined life course trajectories. Four potential trajectories were investigated: stable low SES (low adolescent and low adulthood), decreasing SES (high adolescent and low adulthood), increasing SES (low adolescent and high adulthood) and stable high SES (high adolescent and high adulthood). An accumulation-of-risk model was created by summing scores for adolescent SES and adult SES. The cumulative SES scores ranged from 2 to 6 were categorized into high (score = 2), medium (score = 3—4) and low (scores = 5—6) cumulative SES similar to other studies [26—28].
Statistical methods Statistics were conducted using SPSS v19.0. The effect of adolescent and adult SES, social mobility and accumulation on adult overweight (BMI: 25—30 kg/m2 ) and obesity (BMI: >30 kg/m2 ) were assessed by logistic regression. Models examining the relationship between adolescent SES, adulthood SES and adulthood BMI mutually adjusted for adolescent or adult SES. In addition, all models were adjusted for adolescent BMI (kg/m2 ), parental BMI (categorical predictor: no parents overweight/obese, 1 parent overweight/obese, 2 parents overweight/obese), adult smoking (current smoker/non-smoker), energy intake (KJ) and physical activity. Although the log-likelihood ratio of the unadjusted model (170.0) did not differ much from the model adjusting for smoking energy intake,
e166
S.M. Boylan et al.
and physical activity alone (169.7), these covariates were included in the models as they have been suggested to be partly responsible for the association between SES and BMI [29—32]. Mis-reporters of energy intake (n = 76) were excluded based on a method described elsewhere [33,34]. Birth weight (grams) did not significantly contribute to the models nor was it significantly associated with adulthood BMI (Spearman’s correlation coefficient: r = 0.03, p = 0.61) therefore it was not included in the logistic regression models.
Results Sample characteristics Following exclusion of mis-reporters and those with incomplete data on adult BMI, adolescent and adulthood SES, the final sample consisted of 623 participants (269 males and 354 females), the characteristics of which are presented in Table 1. The mean (SD) age of the sample at baseline was 16.9 (1.0) years and at follow-up was 37.7 (1.4) years. The mean (SD) daily energy intake was 10.5 (2.6) MJ. The subjects were inactive with low mean (SD) MET totals 0.27 (0.23) evident and the majority of them were current non-smokers (79%). More adult males (39%) than females (20%) were classified as overweight, however the prevalence of obesity was similar for females (7%) and males (7%). The prevalence of combined obesity and overweight across SES categories are presented in Table 2. Approximately one-third of the sample was categorized with a low adolescent SES and almost 40% of the sample increased in SES since that time period.
Table 1
Socio-economic status, adulthood obesity and overweight Among females, adolescent, adult, cumulative SES and social mobility were linearly associated with overweight/obesity (Table 2). There were a significantly higher percentage of overweight/obese females from low adult SES compared to those of high adult SES. The small sample size made it difficult to reliably examine the association between SES and obesity, however, the data indicated there were more obese participants with a low adolescent SES (8% males; 9% females) than with a high adolescent SES (4% males; 3% females) and a significantly (p = 0.002) higher percentage of obese females with a low adult SES (12.5%) than a high adult SES (3%). In addition, there were a significantly (p = 0.01) higher proportion of obese females from a low stable SES (10.9%) compared to those from a stable high SES (2%) and a significantly (p = 0.001) higher proportion of obese females with a low cumulative SES (14%) than a high cumulative SES (2%). Table 2 also shows the unadjusted and adjusted relationships between adolescent and adult SES, and combined adult overweight and obesity. After adjusting for adult energy intake, physical activity, smoking, adolescent and parental BMI and adult or adolescent SES, adolescent SES had no significant association with overweight/obesity in this sample, however females of low or medium adult SES were significantly more likely to be overweight/obese compared to those of high adult SES. Following adjustment for the aforementioned covariates, females who decreased in SES were significantly more likely to be overweight/obese compared to those who remained of high SES.
Characteristics of study sample (n = 623).
n Mean age (years) (SD) Baseline Follow-up Follow-up BMI (%) 35 Smoker (%) Yes No Mean daily energy intake (MJ) (SD) Mean MET total (SD)
Males
Females
Total
269
354
623
16.9 (1.0) 37.6 (1.4)
17.0 (1.0) 37.7 (1.4)
16.9 (1.0) 37.7 (1.4)
0 54 39 6 1
3 70 20 4.5 2.5
2 63 28 5 2
20 80 11.7 (2.6) 0.29 (0.23)
21 79 9.5 (2.2) 0.25 (0.23)
21 79 10.5 (2.6) 0.27 (0.23)
Associations between SES and adult obesity/overweight. Males (n = 269) na
Females (n = 354)
Obese or overweight %
Unadjusted OR (95% CI)
Adjusted OR (95% CI)
n
Obese or overweight %
Unadjusted OR (95% CI)
Adjusted OR (95% CI)
Adolescent SES Low Medium High Trend p-value
86 83 100
49 48 42 0.3
1.3 (0.7—2.3) 1.3 (0.7—2.3) 1.0 ref
0.9 (0.4—1.9) 0.8 (0.4—1.7) 1.0 ref
137 101 116
33 26 21 0.03
1.9 (1.1—3.3) 1.3 (0.7—2.5) 1.0 ref
0.7 (0.3—1.5) 0.9 (0.4—2.0) 1.0 ref
Adult SES Low Medium High Trend p-value
63 49 157
52 55 41 0.07
1.6 (0.9—2.9) 1.8 (0.9—3.4) 1.0 ref
1.5 (0.7—3.3) 1.3 (0.5—3.3) 1.0 ref
104 56 194
40 34 18
