Int Arch Occup Environ Health (2007) 80:404–411

DOI 10.1007/s00420-006-0149-9

ORIGINAL ARTICLE

Concentrations of cortisol, testosterone and glycosylated haemoglobin (HbA1c) among construction workers with 12-h workdays and extended workweeks

Anne Helene Garde · Anne Faber · Roger Persson · Åse Marie Hansen · Nis Hjortskov · Palle Ørbæk · Bente Schibye

Received: 20 February 2006 / Accepted: 4 September 2006 / Published online: 7 October 2006© Springer-Verlag 2006

AbstractObjectives Working on large scale construction siteshave been shown to have severe health consequencesin terms of increased risk of hospitalization and disabil-ity retirement compared to construction work in gen-eral. The aim of the present study was to investigatewhether large scale construction work involving 12-hworkdays and extended workweeks leads to insuY-cient recovery measured as increased catabolic anddecreased anabolic metabolism.Methods The study group comprised 40 male con-struction workers of which 21 had 12-h workdays andextended workweeks (56 h/workweek). The compari-son group consisted of 19 male construction workers,who worked regular hours (37 h/week, weekends oV).Measurements of concentrations of cortisol in salivaand free testosterone and glycosylated haemoglobin(HbA1c) in blood were made in a repeated measuresdesign during 2 workweeks for both groups supple-mented with 1 week oV for construction workers withextended workweeks.Results The diurnal proWle of concentrations of sali-vary cortisol for construction workers with extendedworkweeks diVered from the diurnal proWle of salivarycortisol for those with regular work schedules(P < 0.001). The construction workers with extendedworkweeks tended to have 15% [95% CI ¡3%; 37%]higher concentrations of free testosterone in serum

compared to construction workers with regular workschedules (P = 0.09). There were no diVerencesbetween the two groups with respect to concentrationsof HbA1c. There was no increasing trend in concentra-tions of cortisol or decreasing trend in concentrationsof testosterone during the extended workweek. Thediurnal proWle for concentrations of cortisol diVeredbetween workdays and days oV for construction work-ers with extended workweeks (P = 0.003).Conclusion In conclusion, we observed no indica-tions of insuYcient recovery in terms of increased cata-bolic or decreased anabolic metabolism in constructionworkers with 12-h workdays and extended workweekscompared to construction workers with regular workschedules.

Keywords Long work hours · Extended workweeks · “17 Hydroxycorticosteroids” · Testosterone · “Hemoglobin a · glycosylated” · Construction work

Introduction

Recently, two epidemiological studies have shown thatlarge scale construction work may jeopardize health(Hannerz et al. 2005; Tüchsen et al. 2005). Incidencesof hospital admissions due to e.g. cardiovascular dis-eases [age and social class standardized morbidity andmortality ratio (SMR): 157 (95% conWdence interval(CI) 130–189)], musculoskeletal disorders [SMR: 115(95% CI 108–123)], and infections [SMR: 156 (95% CI132–184)] were higher among construction workerswho had formerly worked at The Great Belt Bridge(n = 5,123) compared to workers in the constructionindustry in general (n > 117,000) (Tüchsen et al. 2005).

A. H. Garde (&) · A. Faber · R. Persson · Å. M. Hansen · N. Hjortskov · P. Ørbæk · B. SchibyeNational Institute of Occupational Health, Lersø Parkallé 105, 2100 Copenhagen, Denmarke-mail: ahg@ami.dk

123

Int Arch Occup Environ Health (2007) 80:404–411 405

The Great Belt Bridge workers also had an increasedrisk for disability retirement compared with other con-struction workers and with economically active men ingeneral (Hannerz et al. 2005). However, the reason forthis increased health risk is not clear.

Construction work, in general, is characterized byhigh physical work loads including lifting and carrying,pulling, and pushing of heavy loads (Koningsveld andvan der Molen 1997). In large scale construction workthe physical load may be even larger, because e.g. thesteel reinforcement is heavier and the casting forms arelarger. Further, construction of large traYc links suchas the Copenhagen Metro required more labour thancould be recruited locally. Thus workers were recruitedfar from the work place and consequently were in needof housing, e.g. building site camps, closer to the work-place during the working period. In order to reducecommuting time and ensure a fast and timely conclu-sion a work schedule with long workdays (12 h) andextended workweeks (56 h/week) was implemented.

Working 12-h shifts has been associated withincreased fatigue and poorer safety compared to 8-hshifts, but also with positive outcomes of self-reports oflower stress levels, better physical and psychologicalwell-being, improved durations and quality of oV dutysleep and improvements in family relations (Smithet al. 1998). Further, long weekly work hours mainlydue to overwork has been associated with severalhealth complications (van der Hulst 2003). However,the evidence of negative health consequences of 12-hshifts and extended workweeks not related to over-work are not clear.

One mechanism behind the pathophysiology of longwork hours and extended workweeks may be insuY-cient recovery between work shifts, particularly if thework is physically strenuous. Mental and physicaldemands are not thought to be a problem in itself, pro-vided there is enough time for recovery. In contrast,mental and physical demands are thought to promotethe individual adaptation to various environmentalchallenges (McEwen and Seeman 1999; Seeman et al.2004). However, it has been suggested that if energymobilization and inhibited anabolism is long lasting, itmay increase the vulnerability of several organs includ-ing muscles, immune cells and gastrointestinal func-tions (Theorell 1996). The Wnal pathology (“organselection”) depends on and environmental factors(Ursin and Eriksen 2004).

Several studies indicate that heavy physical activityat intensities and duration much higher than thoseexpected at the workplace combined with insuYcienttime for recovery is associated with increased catabolicand decreased anabolic metabolism. Examples are

intense military training or a 6-day ski race, which havebeen associated with increased levels of serum cortisoland lowered resting levels of serum testosterone (Fell-mann et al. 1992; Gomez-Merino et al. 2003). How-ever, little is known about combinations of recoveryand physical activity at intensities more relevant inwork situations.

The overall aim of the present study was to investi-gate whether large scale construction work involving12-h workdays and extended workweeks lead to insuY-cient recovery as reXected by increased catabolic anddecreased anabolic metabolism. Cortisol and glycosy-lated haemoglobin (HbA1c) were used as measures ofcatabolic activity, whereas testosterone was used as ameasure of anabolic activity (Theorell et al. 2005). Thehypotheses were:

1. Construction workers with 12-h workdays andextended workweeks have increased catabolic(measured as concentrations of cortisol andHbA1c) together with decreased anabolic (mea-sured as concentrations of free testosterone)metabolism compared to construction workerswith a regular work schedule.

2. Construction workers with 12-h workdays andextended workweeks have increased catabolic(measured by cortisol) and decreased anabolic(measured by testosterone) metabolism at the endof a 2-week work period compared to the begin-ning of the work period.

3. Construction workers with 12-h workdays andextended workweeks have increased concentra-tions of salivary cortisol during working days com-pared to days oV.

Methods

Study group

The study group comprised 40 male construction work-ers. Of these, 21 worked on a schedule including 12-hshifts and extended workweeks and lived mostly(n = 18) in temporary building site camps. The remain-ing three lived at home. In general, the agreed scheduleincluded 2 weeks of work (with one or 2 days oV duringthe weekend) followed by 9 days oV work. The com-parison group consisted of 19 male construction work-ers working on a regular work schedule (5 days/week,weekends oV) and living at home. All participantsworked on average approximately 37 h/week over3 weeks. Working hours included a paid lunch breakfor workers with 12-h work days, but not for the

123

406 Int Arch Occup Environ Health (2007) 80:404–411

comparison group. Self-reported characteristics of thework schedule are given in Table 1.

The volunteers were identiWed in collaboration withthe union and the employer. Inclusion criteria wereworking as construction worker with either extendedworkweeks or regular work schedules. Constructionworkers with extended workweeks were recruited fromthree building sites on The Copenhagen Metro. Con-struction workers with regular work schedules wererecruited from three building sites constructing build-ings or road bridges. A total of approximately 145 vol-unteers were informed orally about the study. Of these,53 showed further interest. They were thoroughlyinformed and invited to participate and 40 volunteeredto participate. In total 36 volunteers (19 with long workdays and 17 with regular work schedule) collectedsaliva samples during workdays, 9 volunteers collectedsaliva on days oV (all workers with long work days) and

37 volunteers gave blood samples (19 with long work daysand 18 with regular work schedule). Construction work-ers with regular work schedules had longer commutingtimes than construction workers with 12-h workdays andextended workweeks (Table 1). Further, 7 (37%) of theconstruction workers with regular work schedule spentmore than 2 h commuting per day. With respect to age,height, weight and body mass index (BMI), there were nodiVerences between construction workers with long andregular work hours, as shown in Table 2.

All individuals signed an informed consent form,and the local ethical committee approved the studyprotocol (ref: 01-227/00).

Sampling procedure

Saliva and blood was collected during 3 weeks(2 working weeks and 1 week oV) for constructionworkers with extended workweeks and during 2 weeksfor construction workers with regular work schedules.Data collection was carried out between March-November 2001.

The volunteers collected saliva for measurement ofcortisol on Tuesdays and Thursdays by self-monitoringusing polyester Salivette® tampons. On each day foursaliva samples were collected: (1) at awakening (5:33AM (SD = 58 min) for construction workers withextended workweeks and 4:50 AM (SD = 45 min) forconstruction workers with regular work schedule), (2)30 min after awakening, (3) 45 min after awakeningand (4) at 19.00. Samples were stored frozen (¡20°C)until assayed. Blood samples were collected on work-day 2 (Tuesday) and 11 (Thursday) between 09.00 and10.00. Blood was collected from the antecubital vein byvenepuncture in Vacutainer® tubes (Becton Dickin-son, Rutherford, NJ, USA). Samples for HbA1c analy-sis were collected in tubes containing ethylene-diamine-tetra-acetic-acid (EDTA). Blood samples forfree testosterone were collected in plain tubes with noadditives and after 2 h at room temperature centri-fuged 10 min at 3,000 rev/min [928 relative centrifugal

Table 1 Self-reported characteristics of work schedules for con-struction workers with 12-h workdays and extended workweeksand those with regular work schedules

*Statistically signiWcant diVerence between groups (P < 0.05)a Based on data from volunteers reporting at least 10 workdaysper 2 workweeks

Extended workweeks

Regular workschedules

Number of workers 21 19Work hours per workday (h)Average § SD 11.5 § 1.2 8.0 § 0.7*Minimum–maximum 5.0–13.5 4.5–10.5Work hours per week (h)a

Avg per 2 workweeks 56.2 40.4Avg per 3 weeks 37.5 40.4No. of breaks

per workday (%)*·2 7.2 66.43 21.6 27.6¸4 71.2 6.0Total 100 100Total time in breaks

per day (%)*0–20 min 2.7 6.021–40 min 12.4 42.241–60 min 48.7 36.261–90 min 26.5 14.7>90 min 9.7 0.9Total 100 100Percentage of days

with reported overtime17.0 7.8*

Hours of overwork (on days with overwork)

Average § SD 3.3 § 1.3 0.7 § 0.3*Minimum–maximum 1.0–4.5 0.25–1Mean daily commuting time (min)

<30 75

Table 2 Demographic data

BMI body mass index = body weight (kg)/body height2 (m2 )

Construction workers withextended workweeks

Construction workers with regular work schedules

Number n = 21 n = 19Age in years (range) 42 (27–62) 39 (19–53)Height in cm (range) 181 (167–196) 179 (170–196)Weight in kg (range) 86 (62–105) 82 (60–106)BMI in kg/m2 (range) 26.3 (20.7–30.9) 25.5 (20.0–30.3)

123

Int Arch Occup Environ Health (2007) 80:404–411 407

force (RCF)]. Serum was separated from the sedimentand stored frozen (¡20°C) until assayed.

Further, self-reported sleepiness was assessed by useof a structured logbook Wve times every day for thewhole sampling period. The detailed analyses of the log-book data are reported elsewhere (Persson et al. 2005).

Chemical analysis

Saliva was retrieved from the tampons by centrifuga-tion and analysed for cortisol using a competitiveradio-immunoassay (RIA, Spectria) purchased fromOrion Diagnostica (Espoo, Finland) (Hansen et al.2003), while the RIA used for determination of freetestosterone in serum was Coat-a-count kit purchasedfrom Diagnostic Products Corporation (DPC, LosAngeles, CA, USA). Analysis was carried out accord-ing to the manufacturer’s speciWcations. A 1470 Wizardgamma counter (Wallac, Turku, Finland) was used formeasurement of radioactivity.

The high performance liquid chromatography(HPLC) system for the analysis of HbA1c consisted of aWaters 625 LC system together with a Waters photo-diode-array detector model 996 and a WISP 717 autosampler for automatic injection of the samples. Millen-nium chromatography software was used for calculationof concentrations (Waters Associates Inc. Milford,USA). A cation exchange column Mono S HR 5/5 fromPharmacia Biotech AB (Uppsala, Sweden) was used toseparate HbA1c from other components in the samples.

To show equivalence between diVerent runs refer-ence materials were analysed together with the sam-ples. Westgard control charts were used to documentthat the analytical method remained in analytical andstatistical control i.e., the trueness and the precision ofthe analytical methods remained stable (Christensenet al. 1993; Westgard et al. 1993). Con6 Immunoassaytri-level controls from DPC were used for free testos-terone, Lyphochek Diabetes Control from BioRad(Anaheim, CA, USA) for HbA1c, and natural salivasamples at two levels (6.6 and 16.6 nmol/l) were usedfor cortisol in saliva. The performance of the methodswas further evaluated by participation in inter-labora-tory comparison schemes (Garde et al. 2003a).

Statistical analysis

Statistical analysis was carried out by use of SAS® Sys-tem™, version 8.02 (SAS Institute). AMIQAS wasused for method evaluation and internal quality con-trol (Christensen et al. 1998). Concentrations of sali-vary cortisol and free testosterone in serum wereanalysed as dependent variables in variance compo-

nent models by use of the Mixed procedure. Modelswere Wtted with volunteer as a random factor (com-pound and autoregressive symmetry). For cortisol thevariables work-schedule (extended or regular, hypo-thesis 1), day (four levels, hypothesis 2), time-of-day(four levels) and the 2- and 3-way interaction termsbetween day, work-schedule and time-variables wereincluded in the initial model together with BMI andage. The eVect of the exact sampling time was evalu-ated by including the variable exact-time-of-samplingas a continuous, linear variable. DiVerences betweenworkdays and days oV (hypothesis 3) were evaluated ina model including the variables workday (workday orday oV), time-of-day (four levels) and the interactionterms between workday and time-variables. Variableswere successively omitted from the models if likeli-hood ratio tests produced probabilities larger than0.05. The initial model for concentrations of free tes-tosterone in serum included work-schedule (extendedor regular, hypothesis 1), day (two levels, hypothesis2), the interaction term work-schedule £ day, BMI andage. For comparison of concentrations of HbA1c inblood from construction workers with extended work-weeks and regular hours (hypothesis 1) a one-wayANOVA was used. Due to non-normal (skewed) dis-tributions and heteroscedastic variances (proportionalto level of measurements) for biomarker data, all con-centrations of biomarkers were analysed on logarith-mic scales. In tables the results are presented as meanvalues with 95% conWdence intervals.

For Table 1 diVerences between groups were evalu-ated by use of Student’s t test or Mann–Whitney.

Results

The diurnal proWles of concentrations of salivary corti-sol for construction workers with extended workweeksdiVered from the diurnal proWles of salivary cortisol forconstruction workers with regular work schedules(hypothesis 1, P < 0.001 for the interaction term work-schedule £ time-of-day, Fig. 1). Post-hoc analysis indi-cated that in the late afternoon construction workerswith extended workweeks had concentrations of corti-sol that were 69% [95% CI 55%; 88%] of the cortisolconcentrations observed in construction workers withregular work shifts (P = 0.004). Further, the diVerencebetween the cortisol concentrations in the morning andthe late afternoon tended to be larger (4.9 nmol/l, 95%CI 0–9.8 nmol/l) for construction workers withextended workweeks compared to the constructionworkers with regular work schedules (P = 0.054).There was a tendency (P = 0.09) for construction work-

123

408 Int Arch Occup Environ Health (2007) 80:404–411

ers with extended workweeks to have 15% [95% CI¡3%; 37%] higher concentrations of free testosteronein serum compared to the construction workers withregular work schedules, when combining the secondand the eleventh workday and controlling for BMI andage (Table 3). There were no diVerences between thetwo groups of construction workers with respect toconcentrations of HbA1c (Table 3).

There was no increasing trend in concentrations ofcortisol (Fig. 1) or decreasing trend in concentrationsof testosterone (Table 3) over the 2 week work-periodfor construction workers with extended workweeks(hypothesis 2).

The diurnal proWle for concentrations of cortisoldiVered between workdays and days oV for construc-tion workers with extended workweeks (hypothesis 3,P = 0.003 for the interaction term work-schedule £time-of-day, Fig. 2). On workdays construction workerswith extended workweeks had higher concentrations ofcortisol in the second (P = 0.002) and third (P = 0.047)morning sample, lower concentrations in the late after-noon (P = 0.0034) and a larger diVerence betweenmorning and afternoon (P = 0.0004), when comparedto days oV. Further, they woke up 60 min earlier on

workdays compared to days oV (P < 0.0001). Theobserved diVerences between workdays and days oVfor concentrations of cortisol in the second morningsample (P = 0.022) and over the day (P = 0.048)remained statistically signiWcant when controlling fortime of awakening.

Discussion

The present study investigated whether large scale con-struction work involving 12-h workdays and extendedworkweeks lead to increased catabolic and decreasedanabolic metabolism as a sign of insuYcient recoveryduring workweeks. No such signs were observedamongst construction workers with 12-h workdays andextended workweeks compared to construction work-ers with regular work schedules. With the growingneed for Xexibility, long work hours and extendedworkweeks are more and more widely used. Longwork hours and extended workweeks may have advan-tages for both workers and employers in terms of

Fig. 1 Mean cortisol in saliva. Open square Construction workerswith extended workweeks (n = 17) Cross construction workerswith regular work schedule (n = 17) SEM is in average 15%

1st weekTuesday Thursday

2nd weekTuesday Thursday

0

5

10

15

20

25

06:00 18:00 06:00 18:00 06:00 18:00 06:00 18:00

Time

Cor

tisol

in s

aliv

a(n

mol

/L)

1st weekTuesday Thursday

2nd weekTuesday Thursday

0

5

10

15

20

25

06:00 18:00 06:00 18:00 06:00 18:00 06:00 18:00

Time

Cor

tisol

in s

aliv

a(n

mol

/L)

Table 3 Concentrations of free testosterone and HbA1c

Construction workers with extended workweeks

Construction workers with regular work schedules

P value*

Concentration 95% CI Concentration 95% CI

Free testosterone (pmol/l serum)

Second working day

55 [28–108] 53 [33–84] 0.099

Eleventh working day

61 [38–97] 53 [29–97]

HbA1c (percentage of total HbA)

4.8 [4.1–5.5] 4.8 [4.0–5.9] 0.521*For diVerence between groups. Model for testoster-one was controlled for BMI and age

Fig. 2 Cortisol in saliva (avg § 2 £ sem) for construction work-ers with extended workweeks (n = 9) on a work day (open square)and a day oV (Wlled square)

0.0

5.0

10.0

15.0

20.0

25.0

30.0

04:00 07:00 10:00 13:00 16:00 19:00

Time

Cor

tisol

insa

liva

(nm

ol/L

)

Day offWorkday

123

Int Arch Occup Environ Health (2007) 80:404–411 409

increased possibility for several consecutive days oV-work, reduction in total commuting time and increasedproduction rates (Tepas 1985). But extended work-weeks and long work hours may also have negativeconsequences in terms of e.g. increased fatigue and riskof suVering an occupational injury or illness (Dembeet al. 2005; Smith et al. 1998). SpeciWcally, constructionworkers working around the clock on large scale con-struction work have higher incidences of e.g. cardiovas-cular diseases, musculoskeletal disorders, andinfections compared to workers in the constructionindustry in general (Tüchsen et al. 2005).

The Wrst hypothesis, which stated that constructionworkers with 12-h workdays and extended workweekswould exhibit higher concentrations of catabolic hor-mones (cortisol and HbA1c) and lower concentrationsof anabolic hormones (free testosterone) when com-pared to construction workers with regular workschedules, could not be conWrmed. On the contrary,there was a tendency for construction workers withextended workweeks to have higher concentrations ofserum free testosterone compared to the control groupwhen controlling for age and BMI. Both groupsshowed the expected diurnal cortisol proWle with rela-tively high morning concentrations and low eveningconcentrations. The observed concentrations of sali-vary cortisol were within reference intervals for ahealthy population although relatively high in themorning compared to white-collar workers (Hansenet al. 2003, 2006). This may in part be explained by theearly awakening in this study (Kudielka and Kirsch-baum 2003). Thus the results do not support that 12-hworkdays in combination with physically demandingjob assignments increase the activity in biological sys-tems, which promotes energy expenditure, and reduceactivity in systems involved in reproduction and anabolicmetabolism. There may be several reasons for this.

It is possible that although several days of heavyphysical activity at intensities observed in sports com-bined with insuYcient recovery is associated withincreased catabolic and decreased anabolic metabolism(Fellmann et al. 1992; Gomez-Merino et al. 2003), thismay not be true for the combinations of recovery andphysical activity at intensities more relevant in worksituations. Further, it can not be ruled out that it is the“healthy workers”, who have chosen to participate inthe present study. Workers with 12-h work days tendedto have more breaks during the day and the duration ofthese breaks was longer than for workers with regularwork schedules. It might therefore be that workerswith long work days compensate during the day.Another reason may be related to the time for recov-ery after the workday. The construction workers with

12-h work days may have relatively long time forrecovery despite their long working hours, becausethey lived in building site camps and spend relativelylittle time commuting. The group of construction work-ers with 12-h workdays were mainly housed in buildingsite camps and did therefore have very little familychoirs or other social relationships. This may inXuencesocial interactions negatively, but it may be individu-ally health promoting in that it reduces the total workload during a day and renders more time for sleep andrecovery. Further, the construction workers with regu-lar work schedules spent a considerable time commut-ing to and from work which may have led to shortertime for recovery. In fact, the average commuting timewas 80 min, and more than one third (n = 7) commutedmore than 2 h every day. Thus, when including com-muting time the total number of “work hours” forsome of the workers with regular hours was approxi-mately 9.5 h a day (or more) on Wve consecutive days aweek with only two days of recovery per week. Unfor-tunately, the present study group was too small tostudy the eVects of commuting time. However, the longcommuting times in the group with regular work sche-dules illustrate that there may be other conditionsrelated to work rather than long work hours per se,which is physiologically challenging. Taken together,construction workers with extended work hours livingin camps may have increased total daily time availablefor recovery compared to construction workers withregular work schedules, despite their long daily work-ing hours. A special group in this relation is the con-struction workers with 12-h workdays and extendedworkweeks, who did not live in camps, but commutedhome every day. It may be speculated that they wouldhave too little time for recovery on a daily basis. How-ever, there were not suYcient volunteers (n = 3) in thepresent study group to investigate this question.

The second hypothesis, that construction workerswith 12-h workdays and extended workweeks had anincreasing catabolic and a decreasing anabolic activityacross the 2 workweeks compared to the constructionworkers with regular work schedules, could not be con-Wrmed. The similar concentrations of cortisol and tes-tosterone at the beginning and the end of the 2-weekwork period suggest that the day to day recovery dur-ing the 2 extended workweeks appeared suYcient. Thiswas supported by another study on a correspondinggroup, and may in part be explained by a relatively lowrate of physical exertion (Hansen et al. 2004; Perssonet al. 2005). Perhaps, workers (unconsciously) adjustedtheir workload (or have more and longer breaks asmentioned above) in order to be able to meet the longwork hours, or the ‘objective’ workload was lower than

123

410 Int Arch Occup Environ Health (2007) 80:404–411

expected. The present study was carried out when con-struction of the Metro was in the Wnishing phase, andthe main work processes were stripping oV formworkand ironbinding rather than the more physically heavyconcreting.

We observed no signs of insuYcient recovery interms of increased catabolic and decreased anabolicmetabolism amongst construction workers with 12-hworkdays and extended workweeks compared to con-struction workers with regular work schedules. Thus,insuYcient recovery due to long workdays and work-weeks does not appear to explain the negative healthconsequences observed for the construction workerson the Great Belt Bridge (Hannerz et al. 2005; Tüch-sen et al. 2005). A major diVerence between the GreatBelt Bridge studies and the present study was that theconstruction workers on the Great Belt Bridge workedday and night shifts, whereas the studied constructionworkers in the present study worked only daytime.Shift work including working at night is known to beassociated with impaired health, e.g. increased risk ofmyocardial infarction and peptic ulcer disease (Boggildand Knutsson 1999; Knutsson 2003). Hence it is con-ceivable that the higher relative risks for hospitalisa-tion and illness among the construction workers onThe Great Belt link, mainly were caused by physiologi-cal aberrations as a result of working around the clockrather than the long hours as such.

The third hypothesis that construction workers with12-h workdays and extended workweeks had higherconcentrations of cortisol on working days comparedto days oV was conWrmed for morning samples. Fur-ther, the diVerence between morning and afternoonwas larger on workdays compared to days oV giving thelowest afternoon concentration of cortisol on work-days. DiVerences in concentrations of cortisol betweenworkdays and days oV have been observed previously(Federenko et al. 2004; Garde et al. 2003b; Kudielkaand Kirschbaum 2003; Schlotz et al. 2004). The diVer-ences may partly be due to diVerences in time of awak-ening (Edwards et al. 2001), although the associationremained signiWcant after controlling for time of awak-ening in the present study.

In conclusion, we observed no signs of insuYcientrecovery in terms of increased catabolic and decreasedanabolic metabolism amongst construction workers with12-h workdays and extended workweeks compared toconstruction workers with regular work schedules.

Acknowledgements The project was Wnanced by the NationalInstitute of Occupational Health and a special bill “Satspuljen1999, pt.6” from Folketinget, The Danish Parliament: account FS17.21.05.30. Contractors for the bill were Centre for AlternativeSocial Analysis and Carl Bro A/S. Ulla Tegner, Anne Abildtrup,

Dorte Ekner and Dorrit Meincke are acknowledged for collec-tion of biological samples and chemical analysis.

References

Boggild H, Knutsson A (1999) Shift work, risk factors and cardio-vascular disease. Scand J Work Environ Health 25(2):85–99

Christensen SL, Anglov JTB, Christensen JM, Olsen E, PoulsenOM (1993) Application of a new AMIQAS computer pro-gram for integrated quality control, method evaluation andproWciency testing. Fresen J Anal Chem 345:343–350

Christensen JM, Olsen E, Kristiansen J (1998) AMIQAS PC-pro-gram. In: Method evaluation quality control proWency test-ing. National Institute of Occupational Health, Copenhagen

Dembe AE, Erickson JB, Delbos RG, Banks SM (2005) The im-pact of overtime and long work hours on occupational inju-ries and illnesses: new evidence from the United States.Occup Environ Med 62(9):588–597

Edwards S, Evans P, Hucklebridge F, Clow A (2001) Associationbetween time of awakening and diurnal cortisol secretoryactivity. Psychoneuroendocrinology 26(6):613–622

Federenko I, Wüst S, Hellhammer DH, Dechoux R, Kumsta R,Kirschbaum C (2004) Free cortisol awakening responses areinXuenced by awakening time. Psychoneuroendocrinology29(2):174–184

Fellmann N, Bedu M, Boudet G, Mage M, Sagnol M, PequignotJ, et al. (1992) Inter-relationships between pituitary-adrenalhormones and catecholamines during a 6-day Nordic skirace. Eur J Appl Physiol 64(3):258–265

Garde AH, Hansen ÅM, Nikolajsen TB (2003a) An inter-labora-tory comparison for determination of cortisol in saliva. Ac-cred Qual Assur 8(1):16–20

Garde AH, Hansen ÅM, Persson R, Ohlsson K, Ørbæk P (2003b)The inXuence of production systems on physiological respons-es measured in urine and saliva. Stress Health 19:297–306

Gomez-Merino D, Chennaoui M, Burnat P, Drogou C, Guezen-nec CY (2003) Immune and hormonal changes following in-tense military training. Mil Med 168(12):1034–1038

Hannerz H, Spangenberg S, Tüchsen F, Albertsen K (2005) Dis-ability retirement among former employees at the construc-tion of the Great Belt Link. Public Health 119(4):301–304

Hansen ÅM, Garde AH, Christensen JM, Eller N, Netterstrøm B(2003) Evaluation of a radio-immunoassay and establish-ment of a reference interval for salivary cortisol in healthysubjects in Denmark. Scand J Clin Lab Invest 63:303–310

Hansen AF, Garde AH, Strøyer J, Hjortskov N, Schibye B (2004)Physical performance among construction workers with longworking hours. In: Kristjánsson V (ed) NAM 2004. 50 Nor-diska Arbetsmiljömötet; Reykjavik, Island. Reykjavik: Ar-betarskyddsstyrelsen, 30 August–1 September 2004, pp 229–230

Hansen ÅM, Persson R, Garde AH, Karlson B, Ørbæk P (2006)Diurnal proWles of salivary cortisol on workdays of construc-tion workers versus white-collar workers. Scand J WorkEnviron Health (in press)

van der Hulst M (2003) Long workhours and health. Scand JWork Environ Health 29(3):171–188

Knutsson A (2003) Health disorders of shift workers. Occup Med(Lond) 53(2):103–108

Koningsveld EA, van der Molen HF (1997) History and future ofergonomics in building and construction. Ergonomics40(10):1025–1034

Kudielka BM, Kirschbaum C (2003) Awakening cortisol respons-es are inXuenced by health status and awakening time but

123

Int Arch Occup Environ Health (2007) 80:404–411 411

not by menstrual cycle phase. Psychoneuroendocrinology28(1):35–47

McEwen BS, Seeman T (1999) Protective and damaging eVects ofmediators of stress. Elaborating and testing the concepts ofallostasis and allostatic load. Ann N Y Acad Sci 896:30–47

Persson R, Garde AH, Hansen ÅM (2005) Fatigue and daytimesleepiness among construction workers. In: Jansen B, Kerk-hof G, Koopman M, Witmond.A (eds) 17th internationalsymposium on shiftwork and working time, program and ab-stracts; Hoofddorp, The Netherlands, 18–22 September 2005

Schlotz W, Hellhammer J, Schulz P, Stone AA (2004) Perceivedwork overload and chronic worrying predict weekend-week-day diVerences in the cortisol awakening response. Psycho-som Med 66(2):207–214

Seeman TE, Crimmins E, Huang MH, Singer B, Bucur A, Gru-enewald T et al. (2004) Cumulative biological risk and socio-economic diVerences in mortality: MacArthur studies of suc-cessful aging. Soc Sci Med 58(10):1985–1997

Smith L, Folkard S, Tucker P, Macdonald I (1998) Work shiftduration: a review comparing eight hour and 12 hour shiftsystems. Occup Environ Med 55(4):217–229

Tepas DI (1985) Flexitime, compressed workweeks and otheralternative work schedules. In: Folkard S, Monk TH (eds)Hours of work. Wiley, Chichester, pp 147–164

Theorell T (1996) Possible mechanisms behind the relationshipbetween the demand-control-support model and disordersof the locomotor system. In: Moon SD, Sauter SL (eds) Psy-chosocial aspects of musculoskeletal disorders in oYce work.Taylor & Francis, London, pp 65–73

Theorell T, Hasselhorn H-M, Music Norrtälje Study Group(2005) Endocrinological and immunological variables sensi-tive to psychosocial factors of possible relevance to work-re-lated musculoskeletal disorders. Work Stress 16(2):154–165

Tüchsen F, Hannerz H, Spangenberg S (2005) Morbidity amongbridge and tunnel construction workers who worked longhours and long days construction the great belt Wxed link.Scand J Work Environ Health 31(suppl 2):22–26

Ursin H, Eriksen HR (2004) The cognitive activation theory ofstress. Psychoneuroendocrinology 29:567–592

Westgard JO, Barry PL, Hunt MR, Groth T (1981) A multi-ruleShewhart chart for quality control in clinical chemistry. ClinChem 27(3):493–501

123