keywords 12-hour shift, daily overtime, long working hours, sleepiness, weekly overtime, work intensity J. Sleep Res. (2008) 17, Shift work

Transcription

1 J. Sleep Res. (2008) 17, Shift work doi: /j x Effects of long working hours and the night shift on severe sleepiness among workers with 12-hour shift systems for 5 to 7 consecutive days in the automobile factories of Korea MIA SON 1, JEONG-OK KONG 1, SANG-BAEK KOH 2, JAEYOUNG KIM 3 and MIKKO HÄRMA 4 1 Department of Preventive Medicine, Medical College, Kangwon National University, Chuncheon, 2 Institute of Occupational Medicine, Wonju College of Medicine, Yonsei University, Wonju, South Korea, 3 Harvard School of Public Health, Boston, MA, USA and 4 Finnish Institute of Occupational Health, Helsinki, Finland Accepted in revised form 2 May 2008; received 10 June 2007 SUMMARY We investigated the effects of 12-hour shift work for five to seven consecutive days and overtime on the prevalence of severe sleepiness in the automobile industry in Korea. [Correction added after online publication 28 Nov: Opening sentence of the summary has been rephrased for better clarity.] A total of 288 randomly selected male workers from two automobile factories were selected and investigated using questionnaires and sleep-wake diaries in South Korea. The prevalence of severe sleepiness at work [i.e. Karolinska Sleepiness Scale (KSS) score of 7 or higher] was modeled using marginal logistic regression and included theoretical risk factors related to working hours and potential confounding factors related to socio-economic status, work demands, and health behaviors. Factors related to working hours increased the risk for severe sleepiness at the end of the shift in the following order: the night shift [odds ratio (OR): 4.7; 95% confidence interval (CI): )], daily overtime (OR: 2.2; 95% CI: ), weekly overtime (OR: 1.6; 95% CI: ), and night overtime (OR: 1.6; 95% CI: ). Long working hours and shift work had a significant interactive effect for severe sleepiness at work. Night shift workers who worked for 12 h or more a day were exposed to a risk of severe sleepiness that was 7.5 times greater than day shift workers who worked less than 11 h. Night shifts and long working hours were the main risk factors for severe sleepiness among automobile factory workers in Korea. Night shifts and long working hours have a high degree of interactive effects resulting in severe sleepiness at work, which highlight the need for immediate measures to address these characteristics among South Korean labor force patterns. keywords 12-hour shift, daily overtime, long working hours, sleepiness, weekly overtime, work intensity INTRODUCTION Correspondence: Jaeyoung Kim, Harvard School of Public Health, Landmark Center, 401 Park Drive, Boston, MA 02138, USA. Tel.: ; fax: ; jaeykim@ hsph.harvard.edu The world has become a 24-hour society, eliciting the need for flexible working hours and night working hours [1]. The downside of this strategy is that working hours have been extended into evenings and nights as well as weekends [1,2]. This situation is primarily evident within the rapidly developing Asian countries like South Korea. Many developing Asian countries such as Korea, have added diverse and modified shift work systems to the already existing 12-hour day shift, extending the length of working hours among different types of shifts. Extended work periods, where 12-hour shifts are staffed by two teams for five to seven consecutive days per week, are still common in Korea. Approximately 20% of companies in Korea have adopted shift work with extended work periods (60 68 h a week) and among all firms that have adopted shift work about 78% have this shift system [3]. Ó 2008 European Sleep Research Society 385

2 386 M. Son et al. One area of concern with the 12-hour shift system in Korea is overtime [4]. As the 12-hour shift schedule for five to seven consecutive nights per week is still prevalent, the total number of working hours has been increased by the long duration of the 12-hour shift itself, but also by extending the number of hours worked in a day into the weekend. The most apparent problem with the 12-hour shift system requiring an average of working hours per week is that although the shift is long, there may be only a single or no weekly resting day. The 12-hour shift workers in Korea are therefore exposed to daily and weekly overtime [5]. The trend of long working hours has become increasingly more noticeable since the economic crisis of However, there are insufficient data from other developing countries that have adopted provisions for daily or weekly overtime using the 12-hour shift system for five to seven consecutive days per week. It is widely acknowledged that night shifts are related to sleep disturbance and sleepiness [6,7]. Previous studies have also noted that the 12-hour and ÔdoubleÕ shifts have a detrimental effect on sleepiness and fatigue [4]. Kecklund et al. (2001) showed that a shift system involving a double shift had a negative effect on fatigue, recovery, and health [8]. Smith et al. (1998) found that overtime work under the 12- hour shift system impairs alertness and performance in conjunction with fatigue [4]. In particular, a few studies have investigated the effects of long working hours and shifts on sleepiness using the sleep diary method [9,10]. Harma et al. (2002) showed that the major factors predicting the prevalence of severe sleepiness in the irregular shift system were the time of day (shift), the length of sleep, and the work shift period (day night) [9]. Nevertheless, there is insufficient knowledge on the health effects of long working hours. Moreover, it is rare to find any case studies on the interactive effects of shift work and extended working hours. Therefore, we conducted a study on the effects of 12-hour shifts coupled with extended weekly work periods. The purpose of the study was to investigate the effects of 12-hour shift work for five to seven consecutive days per week and overtime on the prevalence of severe sleepiness within the automobile industry of Korea. The hypothesis was that shift workers will experience more severe sleepiness during the night shifts than during the day shifts of the extended weekly work period. In addition, it was supposed that there could be an interaction between night work and overtime in relation to sleepiness at work. METHODS Participants To investigate the effects of a 12-hour shift, long working hours, and work intensity on severe sleepiness, we provided sleep diaries and distributed a questionnaire to a selected study population at two automobile factories in Korea. The study populations were randomly selected among two factory registries, utilizing a simple random sample. Random numbers were generated using the Statistical Analysis System 9.1 (SAS Institute Inc., Cary, NC, USA) software. We selected these two factories because both factories are affiliates and have the same shift system. Of the total 8700 workers from Factory A, 300 workers (approximately 3%) were selected. Of the total workers from Factory B, 200 workers (approximately 1%) were selected. This study conforms to the requirements of Kangwon National University and its Ethics Committee. This survey was conducted with the consensus of the labor union and automobile workers at the sites. More than 20 welltrained workersõ representatives received permission for this study from all the participants after they were briefed on the purpose of this study. The participants were provided instructions on how to fill in the sleep diary over 14 consecutive days and accompanying questionnaire. About 262 (87.3%) of the 300 workers completed the sleep diary randomly selected from Factory A; however, 104 of them did not complete the questionnaire and were subsequently excluded from this study. Therefore, 158 workers who successfully filled out both the sleep dairy and the questionnaire were included in the final study population. About 192 (96%) completed the sleep diary out of the 200 randomly selected from Factory B; however, 62 of them failed to complete the questionnaire and were excluded from this study. Therefore, 130 workers who completed both their sleep diary and their questionnaire were included in the final study population in this study. Three out of the 288 final participants completed only 7 days of the sleep diary. The resulting number of observation days with 288 participants was There were no significant differences in the mean age, body mass index (BMI), and shift work experience between the participants and the group that failed to complete the survey in this study. There were no significant differences in the general characteristics between the participants and the group that failed to complete the survey in this study, so that the failed group was assumed to have similar results to the participants in this research. Both factories have the same weekly rotation system of two teams operating in two 12-hour shifts in which workers work during the day and at night for five to seven consecutive days. Under Factory AÕs system, the day shift lasts from 07:30 hours to 19:30 hours and the night shift from 19:30 hours to 07:30 hours; in Factory B the day shift extends from 08:30 hour to 20:30 hours and the night shift from 20:30 hours to 08:30 hours. Although Korean labor law limits the working hours to 8 h a day, 5 days a week, and 40 h a week, most of them work 5 consecutive days often with weekend duties, usually laboring 7 days a week. A majority of the workers are manufacturers classified as blue collar workers. Both factories contain several conveyer lines composed of press, body welding, engine assembly, painting, and assembly duties as well as nonconveyer lines which include production management and maintenance. The majority of automobile workers were exposed to ergonomic hazards: repetitive movements, awkward postures, and physical load because of heavy materials.

3 Effect of long working hours and night shift 387 Sleep diary In the sleep diary, participants were to note the shift starting and ending hours, bedtimes, wakeup times, nodding-off or naps (self-notice of nodding off or falling asleep), Karolinska Sleepiness Scale (KSS) [11], sleep latency (Estimating time for falling asleep in the beginning of the main sleep period), frequency of wakings, sleep quality, total amount of coffee, and alcohol, and medicine consumed over the past 24 h [9]. Instructions were given to write down all sleep periods during each 24-hour period, in addition to the main sleep period. The length of a sleep period was calculated as an interval between the bed time and the wake-up time. The main sleep period was defined as the longest normal night sleep for daytime workers or daytime sleep for night shift workers. The total sleep period was summarized for every 24-hour period. KSS was employed to rate severe sleepiness, which the participants were instructed to keep track of at the beginning and end of each shift, at the starting and ending hours of each work cycle, and for any sleepy times during the shift. Severe sleepiness was defined to be a level greater than 7 in the 1 (very alert) 9 (very sleepy) KSS [11]. KSS consists of the following scores: 1 = very alert, 2 = between 1 and 3, 3 = alert, 4 = between 3 and 5, 5 = not sleepy nor alert, 6 = between 5 and 7, 7 = sleepy but not falling asleep, 8 = between 7 and 9, 9 = very sleepy, about to fall asleep. Any incidents of nodding-off or naps (i.e., less than 30 min) during the day were also registered in the sleep diary. For episodes of nodding-off, we asked the subjects to Draw a black dot (d) in your diary, if you notice you have momentarily nodded off/fallen asleep at work or home, either in bed or in the chair. Before their main sleep period, they also recorded daily information about sleep latency, waking frequency, sleep quality, and the total amount of coffee, alcohol, and medicine consumed during the past 24 h [9]. Sleep latency was asked by the following question: ÔEstimate, how quickly you fell asleep in the beginning of the main sleep periodõ. Questionnaire The questionnaire was designed to acquire information about job tasks, the conveyer line job, the total years of working in shifts, working hours (daily and monthly), extra working hours per month, office hours, work type (fixed daytime or day night shifts), shift type (day or night), changes of working conditions related to work intensity (5-point scale, from Ôless intensifiedõ to Ômore intensifiedõ for work pace, amount of work and overtime work), job stress (modified Karasek Questionnaire) [12], the Borg Scale, and the 15-point rating perceived exertion scale [6 20] used to measure the workerõs level of intensity during their work. The content of the 15-point scale is as follows: 6 20% effort [13], physical workload (QEC) [14], 11-item sleep questionnaire [15], KSS at usual times (at the start and end of the day and night shifts), Epworth Sleepiness Scale [16], usual sleeping hours, usual sleep quality (5-point scale, from Ôvery badõ to Ôvery goodõ), frequency of waking, sleep induction time, medical history, and health behaviors (caffeinated tea or coffee, drug medication, smoking, alcohol, and exercise). Definition of long working hours: daily overtime and weekly overtime Although the working hours are specified to 8 h a day, 5 days a week, 40 h a week by the labor law and collective agreements, adding two more hours to the regular 8 h a day seems to be an unwritten law. The working hours for most shift workers in factories A and B are 12 h a day, including meal times. If some workers avoid doing 2 h daily overtime work, their working hours are less than 12 h. Some of the department workers often come to work much earlier than the start-up time and do not get off work until much later; consequently, their daily working hours are easily summed up to 12 h a day. The extra working hours on the weekends also cause longer working days. Most of the shift workers work on Saturdays and Sundays; therefore, they usually work two weekends a month. Therefore, we classified the working hours into three groups: <11 h, 11 h working hours < 12 h, 12 h. We also defined Ôdaily overtimeõ to refer to any working hours beyond 12 h per day, Ôweekly overtimeõ to refer to working more than two extra working days within a 2-week period, and Ônight overtimeõ to refer to working more than 2 extra work nights during a 2-week period and working over 7 nights in a 2-week period. Exposure and confounding variables The theoretical risk factors which were hypothesized and analyzed in this study were age, tenure, total years of shift work, the shift system, daily overtime (working hours), weekly overtime (extra workdays), night overtime (night workdays, extra night workdays), change in work intensity, physical workload, job stress, total sleeping hours, sleep quality, and waking frequency during sleep. As for the potential confounders, we employed job posts and income as socio-economic factors, the office arrival hour, marital status, and number of family dependents as family factors, age and BMI as biological factors, and stimulants, exercise, drinking, smoking, caffeinated tea, and coffee as health behavioral factors. The reason for selecting socio-economic factors, family factors, and health behavioral factors as potential confounding variables is that a respondentõs social and income status may affect the dependent variable, severe sleepiness. We also thought that the higher the number of family dependents, the less time there was for respondents to rest, so they may experience a higher degree of sleepiness at work. The already proven sleep-related health factors such as stimulants, exercise, drinking alcohol, cigarettes, and physical variables such as age and body index were taken as confounding variables for

4 388 M. Son et al. analysis. Furthermore, the potential confounding variables may act as intermediate variables affecting the work environment and conditions that cause sleepiness, so the mutual relation between the confounding variables and severe sleepiness was also taken into account. Statistical method We used the generalized estimating equations approach, which incorporates within-subject correlation, to fit the multivariable model to determine the significant risk factors of severe sleepiness. We fitted marginal logistic regression models based on the generalized estimating equations approach [17], using a procedure (GENMOD; sas Institute) to determine significant risk factors of severe sleepiness [18]. Through bivariate analyses, each of the possible risk factors, confounding variables, and intermediate variables were tested on the dependent variable, severe sleepiness to find the degree of relationship between them. Among the risk factors, the variables with the highest degree of relation were placed in the final model. The selected confounding variables in this study such as the socio-economic factor, the family factor, and the health behavioral factor had a lower degree of relation with sleepiness than such risk factors as working hours and shift work. These covariates for the multivariate model were selected based on standard model building strategies [19]. Three final multivariate marginal models were selected for the total study population of day and night shift workers. We attempted to include the free time in the model. However, it was dropped from the analysis because of its very strong co-variation (multicollinearity) with working hours. We calculated marginal multivariable models with the interactions of shift type and working hours, and similarly consecutive week days and weekends, added to the final models one at a time, with respect to their effect on the prevalence of severe sleepiness. To accurately consider the time-dependent co-variates such as consecutive weekdays, we investigated the interactions of shift type and workdays in the model. All statistical analyses were performed using sas version 8.0 [18]. All statistical tests were two-tailed with a significance level of P < RESULTS Working hours and sleep length Shift workers on a night shift (hereafter night shift workers) slept less than the shift workers on a day shift (hereafter day shift workers) or day workers did, whereas the working hours for night shift workers were longer. The number of extra working days and working nights was also greater among the night shift workers than among the day-shift workers or day workers (Table 1). The average number of sleeping hours during workdays for day workers was 6.8 h, 7.1 h for the day shift workers, and 5.7 for the night shift workers. The average daily number of hours worked by the three groups was greater than 11 h and there was no significant difference found among the three groups; day workers, 11.2; day shift workers, 11.0; and night shift workers, 11.4 h. However, regarding weekly working hours, shift workers had longer working hours than day workers. For example, among the three groups who had Table 1 General characteristics and working hours of 286 workers during the 14-day sleep diary period Day workers Shift workers Day work Day shift Night shift Total number of workers (n) Same as day shift Total observations of sleep diary (n, total) (3808) 2420 (3808) Age (years, mean, SD) 37.1 (8.3) 34.1 (6.9) Same as day shift Tenure (years, mean, SD) 13.1 (7.1) 10.6 (5.9) Same as day shift BMI (mean, SD) 22.9 (2.2) 22.9 (2.3) Same as day shift Weight (kg, mean, SD) 67.0 (6.9) 67.3 (7.9) Same as day shift Height (cm, mean, SD) (4.2) (5.5) Same as day shift Total working hours (hours, mean, SD) (1.5) (1.7) (1.4) Extra work per 2 weeks (days, mean, SD) 2.00 (0.6) 1.6 (0.7) 1.7 (0.7) Night work per 2 weeks (days, mean, SD) 0.0 (0.0) 5.3 (1.0) 5.5 (1.0) Extra night work per 2 weeks (days, mean, SD) 0.0 (0.0) 1.2 (0.5) 1.3 (0.6) Working hours per week (working days 5) (mean, SD) 62.1 (9.5) 61.2 (9.7) 61.7 (8.0) Working hours per week (working days = 5) (mean, SD) 55.3 (2.9) 54.6 (6.6) 55.9 (4.0) Working hours per week (working days = 6) (mean, SD) 66.2 (4.8) 65.7 (5.4) 69.4 (4.3) Working hours per week (working days = 7) (mean, SD) 85.5 (2.1) 87.9 (9.3) 78.0 (1.4) Percentage of the workers who worked 5 days per 1 week (%) Percentage of the workers who worked 6 days per 1 week (%) Percentage of the workers who worked 7 days per 1 week (%) Percentage of the workers who worked 5 7 days per 1 week (%) Sleeping hours per day 6.8 (1.7) 7.1 (1.5) 5.7 (2.4) Sleep quality (bad or very bad, %)

5 Effect of long working hours and night shift 389 worked for 6 days, the longest working hours were 69.4 among the night shift workers. For 7 days a week, the longest weekly working hours were 87.9 h among day shift workers. Overall, the shift workers both on day and night had longer working hours than day workers because their working hours were extended into the weekend (Table 1). workers. For the whole study population, the most important risk factors for severe sleepiness were night shift, daily overtime, and weekly overtime; whereas the most important risk factors for day shift workers were daily overtime and weekly overtime; finally, the risk factors for night shift workers were physical workload, daily overtime, and night overtime. Prevalence of severe sleepiness in relation to the shift system The prevalence of severe sleepiness was higher among the shift workers than among the day workers (Table 2). The prevalence of severe sleepiness at the end of the work period was 9.5% for the day workers, 23.2% for day shift workers, and 61.2% for shift workers on a night shift. Evidently, over half of the night shift workers (61.2%) were suffering from severe sleepiness at the end of their night shift. Waking episodes diminished more notably toward the shift ending time than at the beginning of the night shift during the weekdays and the night shift on Saturdays and Sundays equaling weekend overtime. Effect of 12-hour night shift and long working hours on the prevalence of severe sleepiness Table 2 shows the association between the risk factors of severe sleepiness using marginal logistic regression analysis while adjusting for other covariates. The risk factors that were significantly associated with severe sleepiness were night shift, daily overtime (working more than 12 h in one day), weekly overtime (three extra working days during a 2-week period), and night overtime (two extra work nights during a 2-week period and working over seven nights in a 2-week period) (Table 2). The night shift itself increased the risk of severe sleepiness by over 4.7 times as compared to the day shift. Especially noteworthy, the relationship between severe sleepiness at the end of the work period and long working hours remained significant after adjusting for theoretical confounding variables: health behaviors (caffeinated tea and coffee, drinking, smoking, exercise, stimulants), biological factors (age and BMI), socio-economic factors (job posts and income), working conditions (the office arrival hour), and family factors (marital status and number of family dependents) (Table 2). Health behaviors (coffee, drinking, smoking, stimulants) and biological factors (age and BMI) did not play a role as confounding variables in the relationship between working hours and severe sleepiness. Especially, the health behavior variables initially thought to be the most significant factor for severe sleepiness did not show any significance with the dependent variable. Also, other confounding variables showed little effect on the relationship between working hours and severe sleepiness. Therefore, the confounding variables were used only to control for the relationship between main risk factors and severe sleepiness in this research. Table 3 shows the three final multivariable marginal models for the total study population for day shift and for night shift Interaction between long working hours and 12-hour shift work and the prevalence of severe sleepiness There was an interaction effect between the daily working hours and shifts (night or day) with respect to their effect on the prevalence of severe sleepiness (Log-likelihood test, P < 0.05) (Table 4). Working over 12 h increased the severity of sleepiness by over 7.5 times for night shift workers as compared to day shift workers who worked less than 11 h a day (OR: 7.5; 95% CI: ) (Table 4). In addition, consecutive weekdays have a positive effect on the risk of severe sleepiness associated with each shift type. Working on Sunday increased the severity of sleepiness by 4.6 times relative to working on Monday for night shift workers as compared to day shift workers (OR: 4.6; 95% CI: ) (Table 5). However, there was no increase in severe sleepiness from Monday to Sunday for night shift workers, whereas day shift workers showed severe sleepiness particularly when working on Sundays. DISCUSSION Within this study, we have identified an important set of risk factors associated with severe sleepiness within the Korean automobile industry. These risk factors were the 12-hour night shifts of the extended working week, daily, and weekly overtime. There were interactions between the type of shift and total working hours and between the shift type and cumulative working days with respect to their effect on the prevalence of severe sleepiness. Health behaviors did not affect severe sleepiness nor did they play a role as confounding variables in the relationship between working hours and severe sleepiness. This study shows that long working hours connected with night shifts is a critical factor for severe sleepiness among shift workers working for extended work periods. Several studies reported the association with a higher risk for cardiovascular disease, self-rated health problems, fatigue, and shortened sleep [20]. Long working hours were related to an inverse association between long working hours and sleep [20]. Long working hours and overtime are associated with reduced time for recovery [21]. Baker et al. (1990) reports that overtime work with longer shifts may result in sleepiness while lacking recovery time [22]. For studies concerning the effects of 24-hour shifts on health, Motohashi and Takano (1993) found that 24-hour shifts alter the characteristics of circadian rhythms of ambulance personnel [23]. Smith et al. (1998) posit that the main negative effects of 12-hour shifts are fatigue and safety [4]. Dahlgren et al. (2006) also reports that sleep is negatively affected by overtime work shorter sleep episodes

6 390 M. Son et al. Table 2 Marginal logistic regression analysis of variables affecting the prevalence of severe sleepiness at work among the whole study population Variables Observations Severe sleepiness %of severe cases ORs adjusted for age (1) ORs adjusted for age and all other confounders (2) ORs 95% CIs ORs 95% CIs Age (years) Work type Fixed day time work Day and night shift Shift Day shift Night shift Working hours per day <11 h h working hours < 12 h h Extra working day per 2 weeks 1 day days days Change of extra working days Not changed Changed Night work per 2 weeks 1 days < days days days days Extra night work per 2 weeks 0 days days days Extra night work per 2 weeks <2 days days Total years for shift work in life 10 years years Week Monday Tuesday Wednesday Thursday Friday Saturday Sunday Physical workload Less intensive Intensive QEC, questionnaire for physical workload (QEC, Li and Buckle, 1998); CI, confidence intervals. ORs (1): odds ratios adjusted for age. ORs (2): odds ratios adjusted for all other confounders (job status, income, workplace arrival time, marital status, number of family dependants, body mass index, drugs, exercise, alcohol, smoking, tea, and coffee including caffeine). *Other departments: product control, quality control, maintenance, trials, casting, and alloying.

7 Effect of long working hours and night shift 391 Table 3 Three marginal logistic regression models of variables affecting the prevalence of severe sleepiness at work for all workers, the day shift workers and the night shift workers Model I for all workers Model II for day shift workers Model III for night shift workers Three model variables OR CI OR CI OR CI Intercept Age Work type Fixed day time work 1.0 Day and night shift Shift Day shift 1.0 Night shift Working hours per day <11 h h working hours < 12 h h Extra work per 2 weeks 1 day days days Sleep quality Normal, fair, good Bad, very bad Sleeping hours 6 h 1.0 <6 h Week Monday 1.0 Tuesday Wednesday Thursday Friday Saturday Sunday Extra night work per 2 weeks 0 days days days Change of extra working days Not changed 1.0 Increased OR, odds ratio; CI, confidence intervals. Each model was adjusted for all other confounders (job status, income, workplace arrival time, marital status, number of family dependants, body mass index, drugs, exercise, alcohol, smoking, tea, and coffee including caffeine). during overtime work and greater problems with fatigue and sleepiness [10]. Rosa (1995) points out that unscheduled overtime is more disruptive as it exacerbates the difficulty in organizing sleep, recovery, and social regimens [24]. Several existing studies report the effects of long working hours such as 12-hour shift work and night shift on worker health in Europe and the United States [4]. Nevertheless, it may not be possible to generalize the case studies from Europe and the United States to the situation in Asia because of the large difference in the average weekly working hours. The 12- hour shift systems in Europe and the US are often called Ôcompressed working hours,õ indicating that there are normally only four consecutive 12-hour shifts followed by at least 3 days off. That is clearly different from the Korean work system of five to seven consecutive 12-hour shifts in addition to frequent overtime. In this study, we investigated the cumulative effect of severe sleepiness because of working consecutive shifts from Monday to Sunday. Surprisingly, no cumulative detrimental effects from consecutive workdays were apparent among shift workers on night shifts, whereas severe sleepiness increased during the weekend, especially on Sunday, during the 12-hour day shifts. The reason why consecutive night shifts did not increase sleepiness from Monday to Sunday might be that there was actually no sufficient recovery period after the preceding week, as the shift workers usually worked until Sunday night, even

8 392 M. Son et al. Table 4 The interactive effect of shift and working hours on severe sleepiness Variable Observations Cases* Prevalence** ORs 95%CI b (parameter estimates) SE P-value Day shift and working hours < 11 h Day shift and 11 working hours < 12 h Day shift and working hours 12 h Night shift and working hours < 11 h <.0001 Night shift and 11 working hours < 12 h <.0001 Night shift and working hours 12 h <.0001 OR, odds ratio; CI, confidence intervals. *Cases: cases of severe sleepiness. **Prevalence: prevalence of severe sleepiness. ORs adjusted for all other confounders (job status, income, workplace arrival time, marrital status, number of family dependants, body mass index, drugs, exercise, alcohol, smoking, tea, and coffee including caffeine). Table 5 The interactive effect of shift and order of shifts on severe sleepiness Variable Observations Cases* Prevalence** ORs 95%CI b (Parameter estimates) SE P-value Day shift and Monday Day shift and Tuesday ) Day shift and Wednesday ) Day shift and Thursday ) Day shift and Friday ) Day shift and Saturday ) Day shift and Sunday Night shift and Monday <.0001 Night shift and Tuesday <.0001 Night shift and Wednesday <.0001 Night shift and Thursday <.0001 Night shift and Friday <.0001 Night shift and Saturday <.0001 Night shift and Sunday <.0001 OR, odds ratio; CI, confidence intervals. *Cases: cases of severe sleepiness. **Prevalence: prevalence of severe sleepiness. OR adjusted for all other confounders (job status, income, workplace arrival time, marrital status, number of family dependants, body mass index, drugs, exercise, alcohol, smoking, tea, and coffee including caffeine). up to the next Monday morning when they started night shift week on the same day. Table 5 indicates the degree of severe sleepiness felt by shift workers throughout the week. With Monday as the starting day of the week for the day shift workers, the degree of severe sleepiness is not exacerbated as they approach the end of the week except on Sundays. For night shift workers, Monday was the worst day in terms of experiencing severe sleepiness, but once again, they did not experience an increase in their severe sleepiness as they moved toward the end of the week. The reason might be that night shift workers did not have days off on the weekends for recuperation; thus, they might experience the highest degree of severe sleepiness even on the next Monday. What about the level of working hours in Korea? Korean labor law limits official working hours to 8 h a day and 40 h a week. The history of Korean labor law and working hours dates back to the year The initial level of work week hours, as decided in 1953, was 8 h a day and 48 h a week. After facing strong resistance from Korean workers, the government changed the work week to 46 h a week in 1989 and to 44 h a week in After the emergence of mass unemployment and the need for creating new employments throughout the sluggish economy period called ÔIMFÕ (International Monetary Fund) in 1998, the issue of reform to 40 weekly working hours created a lot of discussion, and in 2003, 8 h a day and 40 h a week were accepted as the official working hours in Korea. Does every Korean enterprise comply with the legal level of working hours, namely 8 h a day and 40 h a week? The answer is unequivocally ÔnoÕ. The average weekly working hours in 1970 and 1980 were h and h; however, this average has remained slightly under 50 h a week since 1989 and 1990 among all Korean workers [25]. The actual total work hours in the manufacturing sector were more than the

9 Effect of long working hours and night shift 393 legal 40 h: 49.0 h for men and 49.6 h for women in 1990; 49.2 and 49.8 respectively in 2000; and 47.0 and 49.6 respectively in 2005 [25]. As with the results of this study, what was the level of working hours for the automobile assembly line workers who participated in this study? In the collective agreements between workers and the automobile factories, the agreed level was 42 h in 1996 which was later reduced to 40 h in Despite the collective agreement reached between the workers and factory management, the workers were still asked to engage in overtime work and extra weekend duties to lower the cost of production by increasing factory operations and production. A deepening social inequality compounded by an increasing number of non-regular workers and population in poverty following the economic crisis in Korea has turned a number of laborers into the status of non-regular workers who receive only half the wages of regular workers. Thus, those workers living on low pay by simply working regular hours find it difficult to make a substantial living and so the workers appear to have fully accepted the overtime work and the extra weekend duties requested from their companies. Consequently, as shown in Table 1, the participants in this study have worked over 12 total hours a day and over 60 h a week (Table 1). A principal strength of our study is that it utilized a field survey that consisted of a 14-day sleep diary as well as a crosssectional questionnaire with randomized sampling. As for the limitation of this study, the number of study subjects might not have been sufficient to explore the interactions among long working hours, shift types, and work intensity on severe sleepiness. Another limitation of this study is that we could not control the diagnosed sleep disturbance, which may play the role of a potential confounding factor. In conclusion, the study shows that the 12-hour shift system, long working hours, and intensified work were the main risk factors for severe sleepiness among automobile factory workers who work 12 h per day, work the night shift, and work five to seven consecutive days per week in Korea. ACKNOWLEDGEMENTS We thank the workers who participated in this study in the automobile factories in Korea. This work was supported by Korea Research Foundation Grant funded by the Korean Government (MOEHRD) (KRF EP0079). AUTHOR CONTRIBUTION Son M initiated the conception, design, data collection, analysis, interpretation, and writing-up of this study. Kim J contributed to conception, design, analysis, interpretation, and writing-up of this study. Kong JO contributed to collection of data and analysis of this study. Koh SB contributed to conception, design, and interpretation of this study. Ha rmä M was involved in conception, design, data collection, interpretation, and writing up of this study. REFERENCES Akerstedt, T. Shift work and disturbed sleep wakefulness. Occup. Med. (Lond), 2003, 53: Akerstedt, T. and Gillberg, M. Subjective and objective sleepiness in the active individual. Int. J. Neurosci., 1990, 52: Baker, T. L., Campbell, S. C. and Linder, K. D. et al. Control Room Operator Alertness and Performance in Nuclear Power Plants. Electric Power Research Institute Report NP Electric Power Research Institute Report, Research Reports Center, Palo Alto, CA, Borg, G. Perceived exertion as an indicator of somatic stress. Scand. J. Rehabil. Med., 1970, 2: Costa, G. Flexibility of working hours in the 24-hour society. Med. Lav., 2006, 97: Costa, G., Akerstedt, T., Nachreiner, F., Baltieri, F., Folkard, S., Frings Dresen, M., Gadbois, C., Gartner, J., Grzech Sukalo, H., Harma, M., Kandolin, I., Silverio, J. and Simoes, A. Flexible working hours, health and well-being in the European Union: preliminary data from a SALTSA project. J. Hum. Ergol. (Tokyo), 2001, 30: Dahlgren, A., Kecklund, G. and Akerstedt, T. Overtime work and its effects on sleep, sleepiness, cortisol and blood pressure in an experimental field study. Scand. J. Work Environ. Health, 2006, 32: Eum, K. D., Li, J., Jhun, H. J., Park, J. T., Tak, S. W., Karasek, R. and Cho, S. I. Psychometric properties of the Korean version of the job content questionnaire: data from health care workers. Int. Arch. Occup. Environ. Health, 2007, 80: Folkard, S. and Tucker, P. Shift work, safety and productivity. Occup. Med. (Lond), 2003, 53: Harma, M., Tenkanen, L., Sjoblom, T., Alikoski, T. and Heinsalmi, P. Combined effects of shift work and life-style on the prevalence of insomnia, sleep deprivation and daytime sleepiness. Scan. J. Work Environ. Health., 1998, 24: Harma, M., Sallinen, M., Ranta, R., Mutanen, P. and Muller, K. The effect of an irregular shift system on sleepiness at work in train drivers and railway traffic controllers. J. Sleep Res., 2002, 11: Hosmer, D. W., Taber, S. and Lemeshow, S. The importance of assessing the fit of logistic regression models: a case study. Am. J. Public Health, 1991, 81: van der Hulst, M. Long work hours and health. Scand. J. Work Environ. Health, 2003, 29: Johns, M. W. A new method for measuring daytime sleepiness: The Epworth Sleepiness Scale. Sleep, 1991, 14: Kecklund, G., Ekstedt, M., Akerstedt, T., Dahlgren, A. and Samuelson, B. The effects of double shifts (15.5 hours) on sleep, fatigue and health. J. Hum. Ergol. (Tokyo), 2001, 30: Li, G. and Buckle, P. A practical method for the assessment of workrelated musculoskeletal risks Quick Exposure Check (QEC). Proceedings of the human factors and ergonomics society 42nd Annual Meeting, 1998, 2: Motohashi, Y. and Takano, T. Effects of 24-hour shift work with nighttime napping on circadian rhythm characteristics in ambulance personnel. Chronobiol. Int., 1993, 10: National Statistics Office. Weekly hours worked by industry. (last accessed 30 January 2007). Park, J. S., Paek, D., Lee, K. B., Rhee, K. Y. and Yi, K. H. Shift work and sickness absence in Korean manufacturing industries. J. Prev. Med. Public Health, 1994, 27: Rissler, A. Stress reactions at work and after work during a period of quantitative overload. Ergonomics, 1977, 20: Rosa, R. R. Extended workshifts and excessive fatigue. J. Sleep Res., 1995, 4(S2):

10 394 M. Son et al. SAS Institute Inc. SAS STAT UserÕs Guide, Version 8. SAS Institute Inc., Cary, NC, Smith, L., Folkard, S., Tucker, P. and Macdonald, I. Work shift duration: a review comparing eight hour and 12 hour shift systems. Occup. Environ. Med., 1998, 55: Son, M., Koh, S. B., Kang, Y. J., Lee, U. S., Kong, J. O., Kim, J. S., Song, H. S., Moon, J. Y., Sun, J. Y., Cho, H. Y., Kim, H. M., Sun, J. W., Jung, J., Jun, U. and Jin, U. The Study on Irregular WorkersÕ Right for Health in Korea. National Human Rights Commission of the Republic of Korea, Seoul, Korea, 2003: Zeger, S. L. and Liang, K. Y. Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 1986, 42: