Originals
Long Work Hours and Obesity in Korean Adult Workers
2013 Volume 55 Issue 5 Pages 359-366
Details
2013 Volume 55 Issue 5 Pages 359-366
Objectives: The present study was designed to identify the association between work hours and obesity in Korean adult manual and nonmanual workers, and to determine whether there is a gender difference in this association. Methods: The study was conducted using Korean National Health and Nutrition Examination Survey data collected between 2007 and 2010. Individuals aged below 25 or over 64 years, pregnant women, part-time workers, soldiers, housewives and students were excluded. The total number of individuals included in the analysis was 8,889 (5,241 male and 3,648 female subjects). The outcome variable was obesity, defined as body mass index ≥25 kg/m2. Variables considered in the model were age, education, income, marital status, alcohol drinking, smoking, daily energy intake, physical activity, sleep hours per day, the type of job, work hours, and work schedule. Work hours were categorized as <40, 40-48 (reference), 49-60, and >60 hours per week. Results: In the multiple SURVEYLOGISTIC regression analyses, the adjusted odds ratio of obesity for long work hours (>60 hours per week) in male manual workers was 1.647 (95% confidence interval 1.262-2.151). Long work hours did not significantly increase the odds ratio for obesity in male nonmanual workers and female manual and nonmanual workers. Conclusions: More than 60 work hours per week increased the risk of obesity in Korean male manual workers. This result might be helpful in preventing obesity in Korean adult workers, especially male manual workers.
(J Occup Health 2013; 55: 359-366)
Obesity is a well-known risk factor for medical diseases including cardiovascular disease, type 2 diabetes mellitus, and several cancers1). Obese workers are often less productive at work2), and have increased absenteeism3), occupational injuries4), morbidity and use of health services5). In recent decades, the prevalence of obesity has been increasing worldwide6). In Korea, the prevalence of obesity (body mass index ≥25 kg/m2) increased from 28.7% in 1998 to 39.9% in 20077).
A variety of factors including genetic8), behavioral9), psychosocial10), and environmental factors11) are known to affect body weight and obesity. Work-related factors such as work hours12) and shift work13) are also related to obesity. Long work hours might lead to weight gain through lack of exercise, undesirable eating habits, and reduced sleep hours12, 14). The presence of gender differences in the association between work hours and obesity is controversial. Ko et al.15) and Ostroy et al.16) reported that long work hours were associated with obesity in male but not female subjects, whereas Lallukka et al.17) reported that long work hours were associated with weight gain and obesity in both male and female subjects.
Most studies, except few literatures18, 19), on the relation between work hours and obesity have been conducted in Western countries where working hours are short compared with Asian countries such as Korea. Until 2007, Korean workers had the longest working hours among countries in the Organization for Economic Cooperation and Development (OECD), and from 2008 to 2011 had the second longest working hours20). In addition, previous studies about work hours and obesity have not considered the type of workload. Occupation-related physical activity and the related energy expenditure are associated with obesity21). The physical activity and workload of manual work differ from those of nonmanual work, so the effect of work hours on obesity might differ in manual and nonmanual workers. Thus, it is important to identify the association between work hours and obesity in a Korean population in order to prevent obesity-related diseases in Korean workers. Furthermore, it is necessary to determine whether the effect of work hours on obesity differs in manual and nonmanual workers. The present study was designed to determine the association between work hours and obesity in Korean adult manual and nonmanual workers, and to determine whether there is a gender difference in the association between work hours and obesity in Korean adult workers.
Data were derived from the Korean National Health and Nutrition Examination Survey (KNHANES) performed by the Korea Center for Disease Control and Prevention. The KNHANES has been performed periodically since 1998 to evaluate the health and socioeconomic status of Koreans. Participants to the KNHANES were selected from noninstitutionalized civilians using a stratified multistage clustered probability sampling design. This sampling method is certified as appropriate for representative statistics by the Korea Department of Statistics. The present study was conducted on KNHANES data collected between 2007 and 2010. This included 14,492 male and 17,612 female subjects. Individuals aged below 25 or over 64 years, pregnant women, part-time workers, soldiers, and the unemployed, including housewives and students, were excluded. Part-time workers were excluded so as not to include individuals with extremely short work hours (less than 10 hours per week) in the analysis. The total number of individuals included in the analysis was 8,889 (5,241 male and 3,648 female subjects). In the KNHANES, the subjects' height and weight were measured objectively in the health examinations, and the subjects were interviewed to collect information such as health behaviors and work characteristics by well-trained interviewers.
According to World Health Organization (WHO), obesity is generally defined as a body mass index (BMI) of 30 kg/m2 or more22). But the WHO in the Asia- Pacific region has recommended different ranges based on risk factors and morbidities23). According to these criteria, the cut-off points for overweight and obesity in Asian adults are 23 and 25 kg/m2, respectively. We used these criteria for obesity, the dependent variable in the present study. Independent variables used were age (20-29, 30-39, 40-49, 50-59, 60-64), education level (less than high school, high school, university; graduate school), income (first quartile, second quartile, third quartile, forth quartile), marital status (single, married), alcohol drinking (none or social, moderate or heavy), smoking (never smoker or ex-smoker, current smoker), daily energy intake (<2,000, 2,000-2,499, ≥2,500 kcal), physical activity (light, moderate, vigorous), sleep hours per day (<5, 5-6, 7-8, ≥9), the type of job (manual, nonmanual), work hours per week (<40, 40-48, 49-60, >60), and work schedule (day work, shift work). Moderate or heavy drinking was defined as drinking more than once a week and more than one bottle of Soju. Moderate physical activity was defined as >30 minutes of moderate activities (e.g., table tennis, swimming, volleyball) at least five times a week. Vigorous physical activity was defined as >20 minutes of vigorous activities (e.g., running, cycling, soccer, basketball) at least three times a week. Manual workers included service and sales workers, agricultural and fishery workers, craft and related workers, plant and machine operators and assemblers, and elementary occupations (construction and mining), and nonmanual workers included general managers and government administrators, professionals and office workers. The Institutional Review Board of Seoul St. Mary's Hospital approved the study.
Statistical analyses were performed separately for male and female subjects. Weighted frequencies of independent variables were calculated using weight presented in the KNHANES data, and chi-square tests were performed to identify whether or not the distribution of each variable was different between obese and nonobese subjects. The odds ratio of obesity according to work hours was estimated by SURVEYLOGISTIC regression analyses, which fits linear logistic regression models for discrete response survey data by using the maximum likelihood methods (SAS windows version 9.2). Age was adjusted in model 1, and age, education, income, marital status, alcohol drinking, smoking, daily energy intake, physical activity, sleep hours per day, and work schedule were adjusted in model 2. The regression analysis was performed separately for manual and nonmanual workers. In the SURVEYLOGISTIC regression analyses, the missing data in the independent variable were not used. All statistical analyses were conducted with SAS windows version 9.2 (SAS institute Inc., Cary, North Carolina). The statistical significance level was set at p<0.05.
Table 1 shows general and work-related characteristics of subjects by type of job. There were 3,165 (60.4%) male and 2,261 (62.0%) female subjects who were manual workers. The distribution of age, education, income, smoking, physical activity, sleep hours per day, and work hours per week were significantly different between manual and nonmanual male and female subjects (p<0.05). Male and female manual workers were older and less educated, and they earned less. Manual workers slept and worked longer hours compared with nonmanual workers.
| Male | Female | |||
|---|---|---|---|---|
| Nonobesea (n=3,186) | Obeseb (n=2,055) | Nonobesea (n=2,698) | Obeseb (n=950) | |
| Age (years)*† | ||||
| 25-29 | 370 (66.5) | 186 (33.5) | 552 (88.3) | 73 (11.7) |
| 30-39 | 964 (61.5) | 603 (38.5) | 668 (82.9) | 138 (17.1) |
| 40-49 | 918 (58.3) | 656 (41.7) | 792 (73.1) | 291 (26.9) |
| 50-59 | 674 (59.8) | 453 (40.2) | 523 (61.5) | 328 (38.5) |
| 60-64 | 260 (62.3) | 157 (37.7) | 163 (57.6) | 120 (42.4) |
| Education*† | ||||
| Less than high school | 311 (63.7) | 177 (36.3) | 429 (57.0) | 324 (43.0) |
| High school | 335 (61.8) | 207 (38.2) | 309 (63.3) | 179 (36.7) |
| University | 1,238 (62.2) | 753 (37.8) | 973 (76.5) | 298 (23.5) |
| Graduate school | 1,301 (58.6) | 918 (41.4) | 987 (86.9) | 149 (13.1) |
| Income*† | ||||
| First quartile | 693 (63.1) | 406 (36.9) | 617 (71.0) | 252 (29.0) |
| Second quartile | 773 (58.9) | 540 (41.1) | 610 (68.2) | 285 (31.8) |
| Third quartile | 864 (62.6) | 517 (37.4) | 688 (76.0) | 217 (24.0) |
| Forth quartile | 803 (58.3) | 574 (41.7) | 750 (80.2) | 185 (19.8) |
| Marital status*† | ||||
| Single | 2,678 (59.6) | 1,812 (40.4) | 2,101 (70.7) | 872 (29.3) |
| Married | 494 (67.4) | 239 (32.6) | 593 (88.8) | 75 (11.2) |
| Alcohol drinking*† | ||||
| None or social | 2,445 (63.0) | 1,437 (37.0) | 2,604 (74.4) | 897 (25.6) |
| Moderate or heavy | 741 (54.5) | 618 (45.5) | 93 (63.7) | 53 (36.3) |
| Smoking | ||||
| Never smoker or ex-smoker | 1,450 (61.0) | 927 (39.0) | 2,478 (73.6) | 887 (26.4) |
| Current smoker | 1,537 (60.7) | 997 (39.3) | 184 (77.3) | 54 (22.7) |
| Daily energy intake (kcal) | ||||
| <2,000 | 1,625 (60.9) | 1,043 (39.1) | 2,078 (73.2) | 759 (26.8) |
| 2,000-2,499 | 669 (62.6) | 400 (37.4) | 378 (75.3) | 124 (24.7) |
| ≥2,500 | 892 (59.3) | 612 (40.7) | 242 (78.3) | 67 (21.7) |
| Physical activity*† | ||||
| Light | 2,291 (61.3) | 1,444 (38.7) | 2,053 (76.5) | 630 (23.5) |
| Moderate | 287 (64.9) | 155 (35.1) | 266 (64.6) | 146 (35.4) |
| Vigorous | 608 (57.1) | 456 (42.9) | 379 (68.5) | 174 (31.5) |
| Sleep hours per day*† | ||||
| <5 | 84 (64.6) | 46 (35.4) | 81 (66.4) | 41 (33.6) |
| 5-6 | 1,212 (57.7) | 888 (42.3) | 1,007 (73.5) | 363 (26.5) |
| 7-8 | 1,713 (62.6) | 2,024 (37.4) | 1,486 (75.4) | 485 (24.6) |
| ≥9 | 177 (64.6) | 97 (35.4) | 124 (67.0) | 61 (33.0) |
| Type of job*† | ||||
| Nonmanual | 1,208 (58.2) | 868 (41.8) | 1,180 (85.1) | 207 (14.9) |
| Manual | 1,978 (62.5) | 1,187 (37.5) | 1,518 (67.1) | 743 (32.9) |
| Work hours per week*† | ||||
| <40 | 326 (59.7) | 220 (40.3) | 478 (69.7) | 208 (30.3) |
| 40-48 | 1,256 (61.5) | 785 (38.5) | 1,213 (79.0) | 322 (21.0) |
| 49-60 | 1,037 (62.7) | 618 (37.3) | 593 (73.2) | 217 (26.8) |
| >60 | 567 (56.8) | 432 (43.2) | 414 (67.1) | 203 (32.9) |
| Work schedule | ||||
| Day work | 1,118 (60.4) | 1,389 (39.6) | 1,781 (72.3) | 683 (27.7) |
| Shift work | 429 (60.3) | 282 (39.7) | 364 (74.1) | 127 (25.9) |
Values are expressed as frequency (%). a BMI<25 m/kg2. b BMI≥25 m/kg2. *Significantly different between obese and nonobese subjects in male subjects (p<0.05). †Significantly different between obese and nonobese subjects in female subjects (p<0.05).
Table 2 shows general and work-related characteristics of subjects by body mass index. There were 2,055 (39.2%) male and 950 (26.0%) female subjects who were obese. The distribution of age, education, income, marital status, alcohol drinking, physical activity, sleep hours per day, type of job, and work hours per week were significantly different between obese and nonobese male and female subjects (p<0.05). In male subjects, there were more obese subjects who were older, more educated, single, and moderate to heavy drinkers. Sleep hours per day had an inverted U-shaped relation with the prevalence of obesity. In female subjects, there were more obese subjects who were older, the same pattern as observed in male subjects. However, education and income had a negative relation with obesity, which was different from the pattern observed in male subjects. Work hours had a U-shaped relation with the prevalence of obesity.
| Male | Female | |||
|---|---|---|---|---|
| Nonmanual (n=2,076) | Manual (n=3,165) | Nonmanual (n=1,387) | Manual (n=2,261) | |
| Age (years)*† | ||||
| 25-29 | 223 (10.8) | 333 (10.5) | 490 (35.3) | 135 (5.9) |
| 30-39 | 798 (38.4) | 769 (24.3) | 476 (34.3) | 330 (14.6) |
| 40-49 | 658 (31.7) | 916 (29.0) | 484 (23.2) | 761 (33.7) |
| 50-59 | 341 (16.4) | 789 (24.8) | 146 (6.5) | 761 (33.7) |
| 60-64 | 56 (2.7) | 361 (11.4) | 9 (0.7) | 274 (12.1) |
| Education*† | ||||
| Less than high school | 22 (1.1) | 466 (14.7) | 8 (0.6) | 745 (33.0) |
| High school | 29 (1.4) | 513 (16.2) | 20 (1.4) | 468 (20.7) |
| University | 472 (22.8) | 1,519 (48.0) | 423 (30.5) | 848 (37.5) |
| Graduate school | 1,552 (74.8) | 667 (21.1) | 936 (67.5) | 200 (8.8) |
| Income*† | ||||
| First quartile | 255 (12.5) | 844 (27.0) | 219 (16.0) | 650 (29.1) |
| Second quartile | 416 (20.3) | 897 (28.7) | 252 (18.4) | 643 (28.8) |
| Third quartile | 615 (30.0) | 766 (24.6) | 367 (26.7) | 538 (24.1) |
| Forth quartile | 762 (37.2) | 615 (19.7) | 534 (38.9) | 401 (18.0) |
| Marital status† | ||||
| Single | 1,777 (86.0) | 2,713 (86.0) | 848 (61.3) | 2,125 (94.1) |
| Married | 290 (14.0) | 443 (14.0) | 536 (38.7) | 132 (5.9) |
| Alcohol drinking | ||||
| None or social | 1,566 (75.4) | 2,316 (73.2) | 1,341 (96.7) | 2,160 (95.6) |
| Moderate or heavy | 510 (24.6) | 849 (26.8) | 46 (3.3) | 100 (4.4) |
| Smoking*† | ||||
| Never smoker or ex-smoker | 1,010 (53.2) | 1,367 (45.4) | 1,300 (95.4) | 2,065 (92.3) |
| Current smoker | 888 (46.8) | 1,646 (54.6) | 63 (4.6) | 175 (7.8) |
| Daily energy intake (kcal) | ||||
| <2,000 | 1,088 (54.4) | 1,580 (49.9) | 1,069 (77.1) | 1,768 (78.2) |
| 2,000-2,499 | 431 (20.8) | 638 (20.2) | 184 (13.3) | 318 (14.1) |
| ≥2,500 | 557 (26.8) | 947 (29.9) | 134 (9.6) | 175 (7.7) |
| Physical activity*† | ||||
| Light | 1,621 (78.1) | 2,114 (66.8) | 1,140 (82.2) | 1,543 (68.2) |
| Moderate | 86 (4.1) | 356 (11.2) | 84 (6.1) | 328 (14.5) |
| Vigorous | 369 (17.8) | 695 (22.0) | 163 (11.7) | 390 (17.3) |
| Sleep hours per day*† | ||||
| <5 | 38 (1.8) | 92 (2.9) | 22 (1.6) | 100 (4.4) |
| 5-6 | 918 (44.2) | 1,182 (37.4) | 512 (36.9) | 858 (37.9) |
| 7-8 | 1,059 (51.0) | 1,678 (53.0) | 814 (58.7) | 1,157 (51.2) |
| ≥9 | 61 (3.0) | 213 (6.7) | 39 (2.8) | 146 (6.5) |
| Work hours per week*† | ||||
| <40 | 139 (6.7) | 407 (12.8) | 213 (15.4) | 473 (20.9) |
| 40-48 | 1,048 (50.5) | 993 (31.4) | 825 (59.5) | 710 (31.4) |
| 49-60 | 659 (31.7) | 996 (31.5) | 281 (20.2) | 529 (23.4) |
| >60 | 230 (11.1) | 769 (24.3) | 68 (4.9) | 549 (24.3) |
| Work schedule | ||||
| Day work | 1,390 (89.6) | 2,117 (79.4) | 905 (87.9) | 1,559 (80.9) |
| Shift work | 162 (10.4) | 549 (20.6) | 124 (12.1) | 367 (19.1) |
Values are expressed as frequency (%). *Significantly different between manual and nonmanual subjects in male subjects (p<0.05). †Significantly different between manual and nonmanual subjects in female subjects (p<0.05).
Table 3 shows the results of the multiple SURVEYLOGISTIC regression analyses in male subjects. There was a U-shaped relation between the odds ratio for obesity and the categories of work hours. In manual workers, the odds ratio for obesity was significantly increased (p<0.05) with long work hours (>60 hours per week) in both model 1 and model 2 (Table 3). In nonmanual workers, the odds ratio for obesity was similar in all categories of work hours in both model 1 and model 2 (Table 3). Table 4 shows the results of the multiple SURVEYLOGISTIC regression analyses in female subjects. In contrast to the results of male subjects, there was no relation between the odds ratio for obesity and work hours in either manual or nonmanual workers (Table 4).
| Work hours per week | Manual workers (n=3,165) | Nonmanual workers (n=2,076) | ||
|---|---|---|---|---|
| Model 1 | Model 2 | Model 1 | Model 2 | |
| <40 | 1.314 (1.001-1.726) | 1.191 (0.861-1.647) | 1.334 (0.893-1.993) | 1.224 (0.741-2.023) |
| 40-48 | 1.000 | 1.000 | 1.000 | 1.000 |
| 49-60 | 1.145 (0.926-1.417) | 1.115 (0.856-1.452) | 0.864 (0.694-1.074) | 0.878 (0.650-1.187) |
| >60 | 1.549 (1.241-1.933) | 1.647 (1.262-2.151) | 1.176 (0.857-1.614) | 1.059 (0.699-1.603) |
Data are shown as odds ratio (95% confidence interval). Model 1 was adjusted for age, and model 2 was adjusted for age, education, income, marital status, alcohol drinking, smoking, daily energy intake, physical activity, sleep hours per day, and work schedule.
| Work hours per week | Manual workers (n=2,261) | Non-manual workers (n=1,387) | ||
|---|---|---|---|---|
| Model 1 | Model 2 | Model 1 | Model 2 | |
| <40 | 1.082 (0.806-1.453) | 0.882 (0.636-1.124) | 1.896 (1.197-2.971) | 1.622 (0.926-2.843) |
| 40-48 | 1.000 | 1.000 | 1.000 | 1.000 |
| 49-60 | 1.029 (0.762-1.389) | 0.961 (0.687-1.346) | 1.420 (0.924-2.182) | 1.340 (0.781-2.299) |
| >60 | 1.051 (0.788-1.404) | 0.863 (0.624-1.194) | 1.552 (0.745-3.234) | 0.916 (0.363-2.308) |
Data are shown as odds ratio (95% confidence interval). Model 1 was adjusted for age, and model 2 was adjusted for age, education, income, marital status, alcohol drinking, smoking, daily energy intake, physical activity, sleep hours per day, and work schedule.
There are three main results of the present study. First, long work hours were independently associated with obesity in Korean male adult workers. Second, the relation between work hours and obesity was different in manual and nonmanual male workers. Work hours over 60 hours per week were associated with obesity in male manual workers, but not in male nonmanual workers. Third, there was a gender difference in the relation between work hours and obesity, and this relation was not observed in females.
There are several mechanisms that could underlie the association between work hours and obesity. Long work hours might interfere with various activities including sleep12), and disruption of the circadian rhythm could play a role in the development of metabolic problems such as obesity24). Magee et al.14) reported that long work hours were indirectly associated with high body mass index, and short sleep might mediate this association. Short sleep could also be associated with the changes in hormone levels such as leptin and ghrelin, which play an important role in the regulation of energy storage and the sense of hunger and satiety in the human body25).
Another possible mechanism for the association between work hours and obesity is undesirable behaviors. It was reported that individuals working longer hours with a deprived sleep duration might exercise less, eat more fast food, and spend more time watching television26-28), and these harmful behaviors could lead to weight gain. Psychological stress can also increase undesirable behaviors and play a role in the association between work hours and obesity. Several studies have reported that individuals working longer hours are more stressed29, 30), and high stress may increase undesirable behaviors including smoking, alcohol use, and food consumption as a way of managing stress31, 32). In the study of Wardle et al.33), individuals working longer hours were more stressed, and had higher intake of energy, saturated fat, and sugar.
Another potential mechanism for the association between work hours and obesity is disturbance of the hypothalamic-pituitary-adrenal (HPA) axis34). The HPA axis plays an important role in the regulation of stress hormones35). Corticotrophin-releasing hormone (CRH) induces the release of adrenocorticotropic hormone (ACTH), and ACTH induces the release cortisol, which is related to weight gain36). Physical stress can induce the release of CRH. Yanagita et al.37) reported that, in an animal model, forced exercise strongly activated CRH neurons compared with spontaneous exercise, and other studies have reported that high intensity or prolonged physical activity is related to CRH and ACTH release38, 39). In the study of Robson et al.40), plasma cortisol concentrations were elevated after physical activities that resulted in fatigue. In the present study, long work hours were associated with obesity in male manual workers, but not in male nonmanual workers. The intensity of physical activity involved in working over 60 hours a week may be viewed as a physical stress that was sufficient to induce CRH and ACTH release in male manual workers but not male nonmanual workers.
In the SERVEYLOGISTIC analyses, although they were not significant, the odds ratios for short work hours (less than 40 hours per week) were above one. The workers working less than 40 hours per week have different working conditions and socioeconomic positions compared with the workers working 40 hours or more per week. Most workers working less than 40 hours are probably non-regular workers, and they could have more work stress due to their positions. These could account for the odds ratios for short work hours (less than 40 hours per week) being above one.
The association between work hours and obesity in female workers has not been consistent in previous studies. Lallukka et al.41) reported that working overtime was related to weight gain in both male and female subjects, but another study reported that the association between working overtime and weight gain was significant in male but not female subjects17). Other studies reported that the association between work hours and obesity was significant in male workers but not female workers5, 15, 42). In the present study, long work hours were not a significant risk factor for obesity in female workers. Women differ from men in terms of such things as menopause, menstruation, childbearing, sex hormones and metabolism. These differences may mediate the gender difference in the effects of long work hours on obesity43). The gender differences observed may therefore be due to biological gender differences. In addition, the effects of socioeconomic factors on obesity might differ according to gender44). In the present study, obesity was more prevalent in more educated male subjects, but there was a negative relation between the prevalence of obesity and education in female subjects. Furthermore, income had a negative relation with obesity rate in female subjects, but not male subjects. These socioeconomic factors could account for the gender differences observed in the present study.
The causal association between obesity and long work hours cannot be determined from the present study due to the cross-sectional study design. In addition, some factors that could be related to obesity, such as female menopause, were not considered. Also, the results of the present study may not be applied to general populations such as those in Western countries because our study subjects were Korean adult workers. Despite these limitations, the results of the present study are meaningful because the Korean workers work longer hours than workers in most countries in the OECD, the data were derived from a nationwide survey, and this study identified differences in the association between work hours and obesity according to gender and job type (manual vs. nonmanual work). Due to the long hours worked by Korean workers, the health hazards and related costs of long work hours could be very large; therefore, the results of the present study might be useful in preventing the obesity associated with long work hours in Korean adult workers, especially male manual workers.
Conflicts of interest: Our study was undertaken without any financial support.
The authors of this manuscript have no conflicts of interest.
The Institutional Review Board of Seoul St. Mary's Hospital approved the study (approval ID: KC12RISI0558).
