PSYCHOSOCIAL WORK CHARACTERISTICS PREDICTING

DAYTIME SLEEPINESS IN DAY AND SHIFT WORKERS

Masaya Takahashi ,1 Akinori Nakata,1 Takashi Haratani,1

Yasumasa Otsuka,1 Kosuke Kaida,1,2 and Kenji Fukasawa1

1National Institute of Occupational Safety and Health, Tama-ku, Kawasaki, Japan2Japan Society for the Promotion of Science, Chiyoda-ku, Tokyo, Japan

Characteristics of work organization other than working time arrangements may con-tribute importantly to daytime sleepiness. The present study was designed to identifythe psychosocial factors at work that predict daytime sleepiness in a sample of day andshift workers. Participants working at a pulp and chemical factory completed an annualquestionnaire regarding psychosocial factors at work using the U.S. National Institutefor Occupational Safety and Health Generic Job Stress Questionnaire (i.e., quantitativeworkload, variance in workload, job control, support from supervisor, coworkers, orfamily/friends, job satisfaction, and depressive symptoms), as well as daytime sleepi-ness (through the Epworth Sleepiness Scale [ESS]) and sleep disturbances for threeyears starting in 2002 (response rates, 94.6–99.0%). The present analysis included55 day workers (11 women) and 57 shift workers (all men) who participated in allthree years of the study, worked under the same work schedule throughout thestudy period, and had no missing data on any of the daytime sleep items. Arepeated-measures analysis of covariance (ANCOVA) was used to test the effects ofwork schedule (day vs. shift work) and psychosocial factors at work in 2002 on theESS scores in subsequent years, with sleep duration, insomnia symptoms, chronicdiseases, and sleepiness levels at baseline as covariates. Given significant and near-significant interactions of work schedules with psychosocial factor or study year, theANCOVA, with the factors of psychosocial work characteristics and study year, wasperformed by type of work schedule. The results indicated a significant main effectof psychosocial work characteristics ( p ¼ 0.010, partial ˛2 ¼ 0.14) and an almostsignificant main effect of study year ( p ¼ 0.067, partial ˛2 ¼ 0.06) and interactionbetween psychosocial work characteristics and study year ( p ¼ 0.085, partial˛2 ¼ 0.06) for variance in workload among the day work group. The day workersreporting high variance in workload in 2002 exhibited significantly higher ESSscores in 2003 and 2004 than did those reporting low variance in workload. TheANCOVA for the shift work group showed a main effect of psychosocial workcharacteristics for job satisfaction ( p ¼ 0.026, partial ˛2 ¼ 0.10) and depressivesymptoms ( p ¼ 0.094, partial ˛2 ¼ 0.06) with the interaction between psychosocial

Address correspondence to Masaya Takahashi Ph.D., National Institute of Occupational Safetyand Health, 6-21-1, Nagao, Tama-ku, Kawasaki 214-8585, Japan. Tel.: þ 81-44-865-6111;Fax: þ81-44-865-6124; E-mail: takaham@h.jniosh.go.jp

Chronobiology International, 23(6): 1409–1422, (2006)Copyright # Informa HealthcareISSN 0742-0528 print/1525-6073 onlineDOI: 10.1080/07420520601100963

1409

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

work characteristics and study year for job satisfaction ( p ¼ 0.172, partial ˛2 ¼ 0.04)and depressive symptoms ( p ¼ 0.035, partial ˛2 ¼ 0.07). The shift workers with lowjob satisfaction and high symptoms of depression in 2002 showed significantlygreater ESS scores in 2003 and/or 2004 than did those with opposite characteristics.These results may suggest a potential predictive value of variance in workloadfor day workers as well as job satisfaction and depressive symptoms for shiftworkers with respect to daytime sleepiness. The present findings may imply thatredesigning these aspects of work environment would be of help in managingdaytime sleepiness.

Keywords Daytime sleepiness, Psychosocial work characteristics, Work schedules, Shiftwork

INTRODUCTION

Work schedules have a considerable impact on the health, safety, andwellbeing of the working population. Extensive research has been madeinto shift work, long hours of work, and flexible working hours forprecise assessment and practical solutions (Akerstedt, 2003; Carusoet al., 2004; Costa et al., 2004; Folkard et al., 2005; Giebel et al., 2004;Rosa, 2005). One of the major concerns here is recognized not only assleep disturbances, but also as increased sleepiness (Akerstedt, 2003;Lockley et al., 2004; Rosekind, 2005). Elevated sleepiness (in otherwords, poor alertness) is shown to cause immediate unfavorableconsequences, such as inferior efficiency, more injuries and accidents atwork, and even lowered quality of life (Akerstedt et al., 2005; Brioneset al., 1996; Folkard et al., 2005; Rosekind, 2005).

A variety of factors are associated with a failure to maintain a requireddegree of alertness. The multidimensionality of sleepiness is a very import-ant concept in understanding, treating, and controlling sleepiness-relatedproblems (Kim and Young, 2005; Young, 2004). Among otherwise healthyworking people, when they should work is thought to primarily affect theirsleepiness, as noted above. Recent studies, however, have focused on thecharacteristics of work organization as a determinant of sleepiness, inaddition to working time arrangements (Akerstedt et al., 2002; Fukasawaet al., 2006; Kalimo et al., 2000; Takahashi et al., 2003). The Work,Lipids and Fibrinogen (WOLF) study in Sweden demonstrated thatwork demand on the demand-control-support model (Karasek and Theo-rell, 1990) was only associated with the likelihood of “unintentional fallingasleep” outside work hours in a dose-dependent manner (odds ratio, 95%confidence interval, CI): low demand ¼ 1 (reference), intermediatedemand ¼ 1.07 (0.86–1.33), and high demand ¼ 1.33 (1.01–1.76)(Akerstedt et al., 2002). Intriguingly, this study did not detect any signifi-cant associations of work demand, control, and social support with the

M. Takahashi et al.1410

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

likelihood of “unintentional falling asleep” during work hours, possibly dueto the masking effects of doing a job per se.

The studies noted above provided insight into a vital role of the psy-chosocial work environment on the sleepiness of workers. Nevertheless,the interpretation of the findings needs caution given that the studydesign was cross-sectional, which might introduce limitations regardingcausality. The current stage of research can thus be carried forward bylongitudinal assessment for assessing the relationship between psychoso-cial factors at work and sleepiness. Similar attempts, however, have beenmade in research to look at the prospective association betweenpsychosocial work characteristics and other relevant problems, such asinsomnia (Jansson and Linton, 2006), fatigue (Eriksen, 2006), or healthstatus (Cheng et al., 2000). Furthermore, a series of the MaastrichtCohort Study on Fatigue at Work have examined the associationbetween psychosocial work characteristics or work schedules and fatigue(i.e., an independent but overlap of symptoms) with sleepiness (Shenet al., 2006) in both cross-sectional and longitudinal manners (Bultmannet al., 2002; Jansen et al., 2003; van Amelsvoort et al., 2004).

The present study sought to examine whether psychosocial factors atwork predict daytime sleepiness during two years of follow-up in asample of day and shift workers. It is hypothesized that unfavorablepsychosocial conditions at work would lead to future increases indaytime sleepiness. Presumably, the patterns of the prospective associa-tions would differ by type of work schedule.

METHODS

Participants

Employees of a pulp and chemical factory were invited to participate inannual study for three years starting in 2002. A questionnaire was distrib-uted to 208, 208, and 204 workers and returned from 206 (response rate,99.0%), 204 (98.1%), and 193 (94.6%) workers in October of 2002, 2003,and 2004, respectively. At these worksites, almost half worked permanentday shifts (08:30–16:30 h) and the remaining worked rotating shifts. Theshift-working group worked under a backward rotating shift schedule witha cycle of three consecutive day shifts (07:00–15:00 h), a day off, threeconsecutive night shifts (23:00–07:00 h), a day off, three consecutiveevening shifts (15:00–23:00 h), and a day off.

This study analyzed data from a sample of 55 day workers (11 women,mean age in 2002 of 44.7 + 5.4 S.D. years and 44 men of 46.0 + 9.2 years)and 57 shift workers (all men, 43.2 + 10.2 years of age), as those workersparticipated in all three years of the study, worked under the same workschedule throughout 2002 to 2004, and had no missing data on any of

Psychosocial Predictors of Daytime Sleepiness 1411

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

the daytime sleepiness items. The study protocol was reviewed andapproved by the research ethics committee of the National Institute ofIndustrial Health, Japan, and was also conducted in accordance with theethical guidelines and standards of the journal (Touitou et al., 2004).

Questionnaire

The questionnaire asked the participants about psychosocial factors atwork, daytime sleepiness, sleep disturbances, and sociodemographic andother health factors. Psychosocial factors at work were measured withthe U.S. National Institute for Occupational Safety and Health GenericJob Stress Questionnaire: quantitative workload (4 items, Cronbach’salpha coefficient ¼ 0.88); variance in workload (3 items, 0.92); jobcontrol (16 items, 0.95); social support from the supervisor (4 items,0.74), coworkers (4 items, 0.82), or family/friends (4 items, 0.81); jobsatisfaction (4 items, 0.75); and depressive symptoms (the Center forEpidemiologic Studies Depression Scale, CES-D, 20 items, 0.83) (Harataniet al., 1996; Hurrell and McLaney, 1988; Nakata et al., 2004).

Daytime sleepiness was quantified with the Epworth Sleepiness Scale(ESS) (Johns, 1991). This eight-item questionnaire is widely used forrating the likelihood of dozing off in eight different situations of dailylife on a four-point scale (0 ¼ never, 1 ¼ slight chance, 2 ¼ moderatechance, 3 ¼ high chance). The possible range of the ESS score is 0–24.The Cronbach’s alpha coefficient was 0.68.

Sleep disturbances were assessed using three questions about per-ceived sleep during the last one-year period (Nakata et al., 2005):

1. “How long does it usually take you to fall asleep in bed? (0–10 min,11–30 min, 31–59 min, 1–2 h, �2 h)” to assess difficulty of initiatingsleep

2. “How often do you have difficulty staying asleep? (never or almostnever, few times/year, more than once/month, more than once/week, three times or more/week, almost every day)” to assess difficultymaintaining sleep, and

3. “How often do you wake up too early and can’t fall asleep again?(never or almost never, few times/year, more than once/month,more than once/week, three times or more/week, almost everyday)” to assess early morning awakening.

The presence of insomnia symptoms was defined if the participants had atleast one positive response to questions regarding difficulty initiating sleep(time to fall asleep of .30 min), difficulty maintaining sleep (�3 times/week), or early morning awakening (3 times/week). In addition, sleep dur-ation at night was measured by the question, “On the average, how much

M. Takahashi et al.1412

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

sleep at night did you usually get during the last year? (,5 h, 5–6 h, 6–7 h,7–8 h, 8–9 h, 9 h or more)”.

Information was also collected on sociodemographic and other healthvariables, including age, gender, marital status, educational level, occu-pation, body mass index, leisure time physical activity, smoking, alcoholconsumption, and the presence of chronic diseases.

Statistical Analysis

Scale scores of the psychosocial work characteristics in 2002 as theindependent variables were dichotomized at the mean. The ESS score in2002 through 2004 was treated as the dependent variable. A repeated-measures analysis of covariance (ANCOVA) was carried out to examinethe effects of work schedule (day work vs. shift work), psychosocial workcharacteristics (high vs. low), and study year (2002, 2003, and 2004) ondaytime sleepiness at follow-up while adjusting for the potential effectsof the presence of chronic diseases (yes, no), sleep duration (,6 h,�6 h), insomnia symptoms (yes, no), and ESS � 11 (yes, no) in 2002.The first three variables were included because of the well-known corre-lates of increased daytime sleepiness (Briones et al., 1996; Doi andMinowa, 2003; Kim and Young, 2005; Ng and Tan, 2005), and thefourth variable was entered to control for the possible influence of thedaytime sleepiness level at baseline (2002). According to those results, arepeated-measures ANCOVA was performed to evaluate the effects ofpsychosocial work characteristics and study year on daytime sleepiness atfollow-up while adjusting for the potential effects of the same fourcovariates. The Greenhouse–Geisser procedure was applied to correctfor violations of sphericity. The statistical analyses were conducted withSPSS for Windows release 10.0J (SPSS, Inc., Chicago, Illinois, USA).

RESULTS

The characteristics of the participants in 2002, the first year of thisstudy, are listed in Table 1. The mean ESS score (5.3 + 3.0 vs.5.3 + 2.6) and prevalence of increased daytime sleepiness (3.6% vs.5.3%) were comparable between day and shift workers. Approximately20% of the both work groups reported having chronic diseases and asleep duration of �6 h. Day workers were also characterized by highereducation ( p ¼ 0.006), less consumption of alcohol ( p ¼ 0.038), fewersymptoms of insomnia ( p ¼ 0.019), more managers ( p ¼ 0.060), andlower body mass index ( p ¼ 0.070) than were shift workers. Withrespect to psychosocial work characteristics, no significant differences

Psychosocial Predictors of Daytime Sleepiness 1413

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

were observed, except for significantly higher job control and lowerdepressive symptoms among day workers.

As summarized in Table 2, the results of repeated-measures ANCOVAon the ESS scores revealed significant and almost significant interactions ofwork schedules with psychosocial work characteristics (job control,p ¼ 0.066, partial ˛2 ¼ 0.04) or study year (quantitative workload,p ¼ 0.070, partial ˛2 ¼ 0.03; job satisfaction, p ¼ 0.020, partial ˛2 ¼ 0.04;depressive symptoms, p ¼ 0.025, partial ˛2 ¼ 0.04). Therefore, therepeated-measures ANCOVA with the factors of psychosocial work charac-teristics and study year was conducted by type of work schedule.

Table 3 presents the ANCOVA results on the ESS scores for day andshift-work groups. A significant main effect of the level of psychosocialwork characteristics was found in terms of quantitative workload

TABLE 1 Characteristics of the Study Participants in 2002

Day workers(n ¼ 55)

Shift workers(n ¼ 57)

Mean (SD) Mean (SD) F or x2 p

Age (yrs) 45.8 (8.5) 43.2 (10.2) 2.12 0.149Women (%) 20.0 0.0 12.64 ,0.001Married (%) 82.3 71.9 2.07 0.151Education�High school (%) 79.6 96.4 7.44 0.006

Managers (%) 13.5 2.0 4.75 0.060Body mass index (kg/m2) 22.2 (2.8) 23.3 (3.2) 3.35 0.070Leisure time physical activity�1 time/w (%) 43.6 52.6 0.91 0.352

Smokers (%) 41.8 52.6 1.31 0.252Alcohol consumption

No (%) 33.3 17.9 6.55 0.0381 glass/occasion (%) 31.5 23.2�2 glasses/occasion (%) 35.2 58.9

Chronic diseases (%) 19.6 23.6 0.25 0.615Sleep duration ,6 h (%) 21.8 22.8 0.02 0.900Insomnia symptoms (%) 13.0 31.6 5.51 0.019ESS 5.3 (3.0) 5.3 (2.6) ,0.01 0.987ESS � 11 (%) 3.6 5.3 0.17 1.000Quantitative workload 10.8 (3.8) 10.7 (3.6) 0.01 0.904Variance in workload 7.9 (3.2) 7.9 (2.8) 0.01 0.928Job control 45.3 (12.7) 38.7 (10.8) 8.42 0.005Social support from

Supervisor 15.4 (3.1) 14.8 (2.6) 0.90 0.346Coworkers 15.8 (3.2) 16.2 (2.2) 0.52 0.473Family/friends 15.3 (4.2) 15.5 (3.1) 0.07 0.791

Job satisfaction 9.4 (1.6) 8.9 (1.7) 2.61 0.109Depressive symptomsa 11.3 (7.3) 14.6 (6.7) 5.68 0.019

CES-D � 16 (%) 16.0 31.5 3.41 0.065

ESS: Epworth Sleepiness Scale.aMeasured by the center for epidemiologic studies depression scale (CES-D).

M. Takahashi et al.1414

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

TABLE 2 Results of the ANCOVAa to Test the Effects of Work Schedule, Psychosocial Work Characteristics, and Study Year on Daytime Sleepiness (ESS Scores)at Follow-Up

S L S � L Y S � Y L � Y S � L � Y

Quantitative workloadF 1.34 10.68 1.85 1.78 2.70 2.06 1.06p, partial eta squared 0.251, 0.01 0.002, 0.11 0.178, 0.02 0.173, 0.02 0.070, 0.03 0.131, 0.02 0.348, 0.01

Variance in workloadF 2.35 8.27 1.49 2.73 1.70 1.18 1.44p, partial eta squared 0.129, 0.03 0.005, 0.08 0.226, 0.02 0.069, 0.03 0.185, 0.02 0.310, 0.01 0.239, 0.02

Job controlF 0.49 0.00 3.47 1.67 1.87 0.18 1.26p, partial eta squared 0.487, 0.01 0.959, 0.00 0.066, 0.04 0.191, 0.02 0.158, 0.02 0.831, 0.00 0.285, 0.01

Social support from supervisorF 1.48 0.01 0.07 2.00 2.07 0.25 0.90p, partial eta squared 0.226, 0.02 0.926, 0.00 0.794, 0.00 0.139, 0.02 0.129, 0.02 0.778, 0.00 0.408, 0.01

Social support from coworkersF 1.08 1.41 0.03 1.81 1.73 0.13 0.64p, partial eta squared 0.301, 0.01 0.238, 0.02 0.853, 0.00 0.167, 0.02 0.181, 0.02 0.879, 0.00 0.525, 0.01

Social support from family/friendsF 2.32 0.02 1.25 2.07 2.27 0.54 1.21p, partial eta squared 0.131, 0.02 0.889, 0.00 0.266, 0.01 0.129, 0.02 0.106, 0.02 0.581, 0.01 0.301, 0.01

Job satisfactionF 3.02 4.15 0.97 1.43 3.98 1.20 1.29p, partial eta squared 0.085, 0.03 0.045, 0.04 0.327, 0.01 0.242, 0.01 0.020, 0.04 0.305, 0.01 0.278, 0.01

Depressive symptomsF 3.42 4.52 0.00 3.24 3.77 2.78 0.72p, partial eta squared 0.068, 0.04 0.036, 0.05 0.946, 0.00 0.042, 0.03 0.025, 0.04 0.065, 0.03 0.489, 0.01

Abbreviations: S ¼ type of work schedule (day work vs. shift work) in 2002; L ¼ level of each psychosocial work characteristics (high vs. low) in 2002; S � L ¼interaction of schedule with level; Y ¼ year of the study (2002, 2003, and 2004); S � Y ¼ interaction of schedule with year; L � Y ¼ interaction of level with year;S � L � Y ¼ interaction among schedule, level, and year.

aCovariates included chronic diseases (yes, no), sleep duration,6 h (yes, no), insomnia symptoms (yes, no), and ESS � 11 (yes, no) in 2002.

14

15

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

TABLE 3 Results of the ANCOVAa to Test the Effects of Psychosocial Characteristics Factor and Study Year on Daytime Sleepiness (ESS Scores) by Type of WorkSchedule at Follow-Up

Day work Shift work

L Y L � Y L Y L � Y

Quantitative workloadF 11.23 2.85 2.43 3.37 0.65 1.11p, partial eta squared 0.002, 0.21 0.064, 0.06 0.095, 0.05 0.073, 0.07 0.521, 0.01 0.334, 0.02

Variance in workloadF 7.22 2.82 2.56 1.94 0.78 0.33p, partial eta squared 0.010, 0.14 0.067, 0.06 0.085, 0.06 0.171, 0.04 0.461, 0.02 0.718, 0.01

Job controlF 1.22 2.21 0.86 2.00 0.27 0.31p, partial eta squared 0.276, 0.03 0.118, 0.05 0.425, 0.02 0.164, 0.04 0.759, 0.01 0.731, 0.01

Social support from supervisorF 0.22 2.63 0.07 0.10 0.18 1.15p, partial eta squared 0.642, 0.01 0.080, 0.06 0.928, 0.00 0.748, 0.00 0.834, 0.00 0.322, 0.03

Social support from coworkersF 0.27 2.18 0.53 1.49 0.28 0.45p, partial eta squared 0.606, 0.01 0.121, 0.05 0.587, 0.01 0.229, 0.03 0.754, 0.01 0.633, 0.01

Social support from family/friendsF 0.51 2.50 0.42 0.70 0.39 1.35p, partial eta squared 0.481, 0.01 0.089, 0.06 0.653, 0.01 0.409, 0.02 0.671, 0.01 0.264, 0.03

Job satisfactionF 0.66 3.13 0.29 5.24 0.15 1.79p, partial eta squared 0.421, 0.02 0.050, 0.07 0.745, 0.01 0.026, 0.10 0.857, 0.00 0.172, 0.04

Depressive symptomsF 1.48 3.26 0.81 2.93 0.94 3.49p, partial eta squared 0.231, 0.03 0.045, 0.07 0.445, 0.02 0.094, 0.06 0.395, 0.02 0.035, 0.07

Abbreviations: L ¼ level of each psychosocial work characteristics (high vs. low) in 2002; Y ¼ year of the study (2002, 2003, and 2004); L � Y: interaction of levelwith year.

aCovariates included chronic diseases (yes, no), sleep duration ,6 h (yes, no), insomnia symptoms (yes, no), and ESS �11 (yes, no) in 2002.

14

16

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

( p ¼ 0.002, partial ˛2 ¼ 0.21) and its variance ( p ¼ 0.010, partial˛2 ¼ 0.14) among the day workers. A significant or near significant effectof the study year was also observed in terms of quantitative workload( p ¼ 0.064, partial ˛2 ¼ 0.06), variance in workload ( p ¼ 0.067, partial˛2 ¼ 0.06), social support from supervisor ( p ¼ 0.080, partial ˛2 ¼ 0.06)and family/friends ( p ¼ 0.089, partial ˛2 ¼ 0.06), job satisfaction( p ¼ 0.050, partial ˛2 ¼ 0.07), and depressive symptoms ( p ¼ 0.045,partial ˛2 ¼ 0.07). The interaction between psychosocial work charac-teristics and study year tended to be significant in terms of quantitativeworkload ( p ¼ 0.095, partial ˛2 ¼ 0.05) and variance in workload( p ¼ 0.085, partial ˛2 ¼ 0.06).

Post hoc pairwise comparisons at each year indicated that the ESSscores of the day workers with high workload in 2002 were consistentlyhigher from 2002 to 2004 than the scores of those with low workload in2002 ( p , 0.05; see Figure 1). There were no significant differences inESS scores between the high and low groups of variance in workload in2002, and the ESS scores in 2003 and 2004 were significantly greater for

FIGURE 1 ESS scores (mean and standard error) in 2002–2004 by the high (filled circle) and low(empty circle) level of each psychosocial work characteristic in 2002 for day and shift workers.�p , 0.05 compared between the two groups in each year.

Psychosocial Predictors of Daytime Sleepiness 1417

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

the high group in 2002 than for the corresponding low group in 2002( p , 0.05).

Among the shift-work group, a main effect of the level of psychosocialwork characteristics was significant in terms of job satisfaction ( p ¼ 0.026,partial ˛2 ¼ 0.10) and tended to be significant in terms of quantitativeworkload ( p ¼ 0.073, partial ˛2 ¼ 0.07) and depressive symptoms( p ¼ 0.094, partial ˛2 ¼ 0.06). In addition, a significant interactionbetween psychosocial work characteristics and study year was obtainedin terms of depressive symptoms ( p ¼ 0.035, partial ˛2 ¼ 0.07). The two-way interaction was not statistically significant in terms of job satisfaction( p ¼ 0.172), but its partial ˛2 was 0.04. This value seemed to correspondto an approximately medium effect size (Huberty, 2002). According topost hoc pairwise comparisons for the shift workers (see Figure 1), signifi-cantly greater ESS scores in 2004 were found for the low job satisfactiongroup in 2002 than for the high job satisfaction group in 2002( p , 0.05). Likewise, significantly higher ESS scores in 2003 and 2004were observed for the high depressive shift workers in 2002 than for lowdepressive counterparts in 2002 ( p , 0.05).

The results for both the day-work and shift-work groups presentedabove remained consistent with those after additional adjustment formarital status (married, single), smokers (yes, no), alcohol consumption(0, 1, or �2 glasses/occasion), and leisure time physical activity (yes, no).

DISCUSSION

The present study examined whether psychosocial work characteristicsat baseline predicted subsequent levels of daytime sleepiness amongemployees working permanent day and rotating shifts. The resultsshowed that more variable workload was associated with an increase indaytime sleepiness at follow-up for day workers, and that reduced jobsatisfaction and more symptoms of depression were linked to elevateddaytime sleepiness over two years of follow-up for shift workers. A highquantitative workload yielded consistently greater levels of sleepinessduring the follow-up period as well as the initial year of the study, particu-larly for the day workers.

For reasons that are as yet unknown, day workers with a high variancein workload developed an elevation of daytime sleepiness during the twoyears of follow-up. However, extra efforts may need to be taken to meetthe production requirements under highly varying demands. Thesesituations are assumed to make the work environment more exhaustiveand unsafe. Indeed, the cross-sectional study in small and medium-sizedenterprises demonstrated a significantly higher risk of reported injuriesat work (adjusted odds ratio 1.71, 95% CI 1.32–2.17) in 1489 men whoperceived greater variability in workload (Nakata et al., 2006). Note that

M. Takahashi et al.1418

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

this odds ratio was the highest among that of the other psychosocial work-related risk factors addressed. It is likely that workers’ sleepiness may playa contributory role in those occupational injuries (Akerstedt et al., 2005;Folkard et al., 2005; Lockley et al., 2004; Nabi et al., 2006). The presentfindings, combined with the earlier data, may support the possibility thatexposure to highly fluctuated workload may be an occupational risk thatmight give rise to future increased sleepiness. The research questionsthat need to be pursued would be as to how and why such a relationshipis possible.

Highly depressive symptoms, as measured by the CES-D, were foundto be linked to subsequent increases in daytime sleepiness among thesample of shift-working men. The present data also showed significantlygreater depressive symptoms of the shift-work group at baseline thanthose of the day-work counterpart (see Table 1). A recent report demon-strated 2.6 times greater rates of depression, as determined by thephone interview using the Diagnostic Interview Schedule, amongworking people with excessive daytime or insomnia than among thosewithout such symptoms (Drake et al., 2004). Daytime sleepiness andinsomnia are major symptoms of the circadian rhythm sleep disorders,shift-work type (shift-work disorder), according to the InternationalClassification of Sleep Disorders (American Academy of Sleep Medicine,2005). If some portions of the highly depressed group suffered fromshift-work disorder, then their sleepiness might be elevated more thanthat of the low-depressed group. Yet, this assumption cannot be testeduntil the participants in the present study receive medical diagnosis forshift-work disorder. It is sometimes quite difficult to make a differentialdiagnosis for causes of excessive sleepiness (Doghramji, 2004), and thereis room to refine the concept and definition of shift-work disorder(Akerstedt, 2005). Although the present results show that increased symp-toms of depression might raise the likelihood of future high sleepiness forshift-working men, additional work is necessary to verify this hypothesis.

The levels of job satisfaction were found to be associated with daytimesleepiness at follow-up among the shift-working men, although the inter-action between the level of job satisfaction and study year was not statisticallysignificant, probably due to the small sample size of the high-job satisfactiongroup (n ¼ 15) compared to that of the low-job satisfaction group (n ¼ 40).A recent meta-analysis highlighted a positive relationship between job satis-faction and mental health (Faragher et al., 2005), though it is not clear as tothe role of job satisfaction in sleepiness of the shift workers. However, shiftworkers who are dissatisfied with their work hours are reported to scorehigher sleepiness on the Karolinska Sleepiness Scale during shifts thanthe counterparts who are satisfied (Axelsson et al., 2004). Besides, dissatis-faction with the given shift schedule is reported to be associated with alowered level of testosterone in the morning (Axelsson et al., 2003).

Psychosocial Predictors of Daytime Sleepiness 1419

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

Taken together, the present findings may suggest that attention should bepaid to low levels of job satisfaction in a shift-work setting, which mightincrease the likelihood of subsequent sleepy state.

Several aspects of the present study require due consideration. Thenumber of participants was small, and most were men. These facts mayrelate to the size and characteristics of the study worksite (of approximately100 day workers and 100 shift workers, men accounting for 90% ofthe employees), switching of work schedules, missing values, and non-respondents at follow-up. Consequently, the present findings from thereduced number of participants (55 day workers and 57 shift workers)might have been influenced by selection bias. The use of the ESS scoreas the dependent measure of sleepiness may be criticized because thisscale addresses daily life situations, not necessarily work situations(Akerstedt, 2005). In addition to evaluating on-the-job sleepiness,however, overall assessment of workers’ sleepiness would be relevant,given the recent finding as to the distinct nature of sleepiness at workand outside work (Akerstedt et al., 2002). The present study relied onself-reported data for both dependent and independent variables. Theresults obtained might thus be affected by common method variance(Singh-Manoux et al., 2005). It would be desirable to quantify sleepinesswith other methodologies, such as physiologic and/or neurobehavioraltesting.

In conclusion, the current study suggests a potential predictive value ofvariance in workload for day workers and job satisfaction and depressivesymptoms for shift workers in light of daytime sleepiness. Although thelimitations of this study should be recognized, the findings imply that rede-signing these aspects of the work environment would be one of the possiblemeasures of managing daytime sleepiness.

ACKNOWLEDGMENT

This study was supported in part by a Grant-in-Aid for ScientificResearch 160790332 from the Ministry of Education, Culture, Sports,Science and Technology in Japan.

REFERENCES

Akerstedt, T. (2003). Shift work and disturbed sleep/wakefulness. Occup. Med. (Lonon). 53:89–94.Akerstedt, T. (2005). Shift work and sleep disorders. Sleep 28:9–11.Akerstedt, T., Knutsson, A., Westerholm, P., Theorell, T., Alfredsson, L., Kecklund, G. (2002). Work

organisation and unintentional sleep: results from the WOLF study. Occup. Environ. Med. 59:595–600.

Akerstedt, T., Peters, B., Anund, A., Kecklund, G. (2005). Impaired alertness and performance drivinghome from the night shift: a driving simulator study. J. Sleep Res. 14:17–20.

M. Takahashi et al.1420

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

American Academy of Sleep Medicine (2005). Circadian rhythm sleep disorder, shift work type. InInternational Classification of Sleep Disorders: Diagnostic and Coding Manual. 2nd ed., Westchester,Ill: American Academy of Sleep Medicine, pp. 131–133.

Axelsson, J., Akerstedt, T., Kecklund, G., Lindqvist, A., Attefors, R. (2003). Hormonal changes insatisfied and dissatisfied shift workers across a shift cycle. J. Appl. Physiol. 95:2099–2105.

Axelsson, J., Akerstedt, T., Kecklund, G., Lowden, A. (2004). Tolerance to shift work—how does itrelate to sleep and wakefulness? Int. Arch. Occup. Environ. Health 77:121–129.

Briones, B., Adams, N., Strauss, M., Rosenberg, C., Whalen, C., Carskadon, M., Roebuck, T.,Winters, M., Redline, S. (1996). Relationship between sleepiness and general health status.Sleep 19:583–588.

Bultmann, U., Kant, I.J., Van den Brandt, P.A., Kasl, S.V. (2002). Psychosocial work characteristics asrisk factors for the onset of fatigue and psychological distress: prospective results from the Maas-tricht Cohort Study. Psychol. Med. 32:333–345.

Caruso, C.C., Hitchcock, E.M., Dick, R.B., Russo, J.M., Schmidt, J.M. (2004). Overtime and ExtendedWork Shifts: Recent Findings on Illnesses, Injuries and Health Behaviors. DHHS (NIOSH) PublicationNo. 2004-2143. Cincinnati, Ohio: U.S. Department of Health and Human Services, Public HealthService, Centers for Disease Control and Prevention, National Institute for Occupational Safetyand Health.

Cheng, Y., Kawachi, I., Coakley, E.H., Schwartz, J., Colditz, G. (2000). Association between psychoso-cial work characteristics and health functioning in American women: prospective study. BMJ. 320:1432–1436.

Costa, G., Akerstedt, T., Nachreiner, F., Baltieri, F., Carvalhais, J., Folkard, S., Dresen, M.F.,Gadbois, C., Gartner, J., Sukalo, H.G., Harma, M., Kandolin, I., Sartori, S., Silverio, J. (2004).Flexible working hours, health, and well-being in Europe: some considerations from a SALTSAproject. Chronobiol Int. 21:831–844.

Doghramji, K. (2004). Assessment of excessive sleepiness and insomnia as they relate to circadianrhythm sleep disorders. J. Clin. Psychiatry. 65(Suppl.16):17–22.

Doi, Y., Minowa, M. (2003). Gender differences in excessive daytime sleepiness among Japaneseworkers. Soc. Sci. Med. 56:883–894.

Drake, C.L., Roehrs, T., Richardson, G., Walsh, J.K., Roth, T. (2004). Shift work sleep disorder: preva-lence and consequences beyond that of symptomatic day workers. Sleep 27:1453–1462.

Eriksen, W. (2006). Work factors as predictors of persistent fatigue: a prospective study of nurses’aides. Occup. Environ. Med. 63:428–434.

Faragher, E.B., Cass, M., Cooper, C.L. (2005). The relationship between job satisfaction and health: ameta-analysis. Occup. Environ. Med. 62:105–112.

Folkard, S., Lombardi, D.A., Tucker, P.T. (2005). Shiftwork: safety, sleepiness and sleep. Ind. Health 43:20–23.

Fukasawa, K., Aikawa, H., Okazaki, I., Haratani, T., Takahashi, M., Nakata, A., Otsuka, Y., Kaida, K.,Hanada, T. (2006). Perceived sleepiness of non-shift working men in two different types of workorganization. J. Occup. Health 48:230–238.

Giebel, O., Janssen, Schomann, C., Nachreiner, F. (2004). A new approach to evaluating flexibleworking hours. Chronobiol. Int. 21:1015–1024.

Haratani, T., Kawakami, N., Araki, S., Hurrell, J.J., Jr., Sauter, S.L., Swanson, N. (1996). Psychometricproperties and stability of the Japanese version of the NIOSH job stress questionnaire. 25th Int.Cong. Occup. Health, Book of Abstracts, Part 2, p. 393.

Huberty, C.J. (2002). A history of effect size indices. Edu. Psychol. Measure. 62:227–240.Hurrell, J.J., Jr., McLaney, M.A. (1988). Exposure to job stress—a new psychometric instrument.

Scand. J. Work Environ. Health. 14(Suppl.1):27–28.Jansen, N.W., van Amelsvoort, L.G., Kristensen, T.S., van den Brandt, P.A., Kant, I.J. (2003). Work

schedules and fatigue: a prospective cohort study. Occup. Environ. Med. 60(Suppl.1):47–53.Jansson, M., Linton, S.J. (2006). Psychosocial work stressors in the development and maintenance of

insomnia: a prospective study. J. Occup. Health Psychol. 11:241–248.Johns, M.W. (1991). A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale.

Sleep 14:540–545.Kalimo, R., Tenkanen, L., Harma, M., Poppius, E., Heinsalmi, P. (2000). Job stress and sleep disorders:

findings from the Helsinki Heart Study. Stress Med. 16:65–75.

Psychosocial Predictors of Daytime Sleepiness 1421

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.

Karasek, R.A., Theorell, T. (1990). Healthy Work: Stress, Productivity and the Reconstruction of Working Life.New York: Basic Books.

Kim, H., Young, T. (2005). Subjective daytime sleepiness: dimensions and correlates in the generalpopulation. Sleep 28:625–634.

Lockley, S.W., Cronin, J.W., Evans, E.E., Cade, B.E., Lee, C.J., Landrigan, C.P., Rothschild, J.M.,Katz, J.T., Lilly, C.M., Stone, P.H., Aeschbach, D., Czeisler, C.A. (2004). Effect of reducinginterns’ weekly work hours on sleep and attentional failures. N. Engl. J. Med. 351:1829–1837.

Nabi, H., Gueguen, A., Zins, M., Lagarde, E., Chiron, M., Lafont, S. (2006). Awareness of driving whilesleepy and road traffic accidents: prospective study in GAZEL cohort. BMJ. 333:75.

Nakata, A., Haratani, T., Takahashi, M., Kawakami, N., Arito, H., Kobayashi, F., Araki, S. (2004). Jobstress, social support, and prevalence of insomnia in a population of Japanese daytime workers.Soc. Sci. Med. 59:1719–1730.

Nakata, A., Ikeda, T., Takahashi, M., Haratani, T., Fujioka, Y., Fukui, S., Swanson, N.G., Hojou, M.,Araki, S. (2005). Sleep-related risk of occupational injuries in Japanese small and medium-scaleenterprises. Ind. Health. 43:89–97.

Nakata, A., Ikeda, T., Takahashi, M., Haratani, T., Hojou, M., Fujioka, Y., Swanson, N.G., Araki, S.(2006). Impact of psychosocial job stress on non-fatal occupational injuries in small andmedium-sized manufacturing enterprises. Am. J. Ind. Med. 49:658–669.

Ng, T.P., Tan, W.C. (2005). Prevalence and determinants of excessive daytime sleepiness in an Asianmulti-ethnic population. Sleep Med. 6:523–529.

Rosa, R.R. (2005). Toward better sleep for workers: impressions of some needs. Ind. Health 43:85–87.Rosekind, M.R. (2005). Underestimating the societal costs of impaired alertness: safety, health and pro-

ductivity risks. Sleep Med. 6(Suppl.1):S21–S25.Shen, J., Barbera, J., Shapiro, C.M. (2006). Distinguishing sleepiness and fatigue: focus on definition

and measurement. Sleep Med. Rev. 10:63–76.Singh-Manoux, A., Marmot, M.G., Adler, N.E. (2005). Does subjective social status predict health and

change in health status better than objective status? Psychosom. Med. 67:855–861.Takahashi, M., Nakata, A., Haratani, T., Fukasawa, K., Ogawa, Y., Fukui, S., Fujioka, Y., Nagai, C.,

Tachibana, N. (2003). Elevated daytime sleepiness in day and shift workers experiencing stressat work. Shiftwork Int. Newslett. 20:180. Available at: http://www.workingtime.org/images/7/7c/

Nov2003Full.pdf.Touitou, Y., Portaluppi, F., Smolensky, M.H., Rensing, L. (2004). Ethical principles and standards for

the conduct of human and animal biological rhythm research. Chronobiol Int. 21:161–170.van Amelsvoort, L.G., Jansen, N.W., Swaen, G.M., van den Brandt, P.A., Kant, I. (2004). Direction

of shift rotation among three-shift workers in relation to psychological health and work-familyconflict. Scand. J. Work Environ. Health 30:149–156.

Young, T.B. (2004). Epidemiology of daytime sleepiness: definitions, symptomatology, and pre-valence. J. Clin. Psychiatry 65(Suppl.16):12–16.

M. Takahashi et al.1422

Chr

onob

iol I

nt D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsita

t Aut

onom

a B

arce

lona

on

10/2

8/14

For

pers

onal

use

onl

y.