Problems related to shift work. A field study of Swedish railroad workers with irregular work

and SWENSSON, A. P,roblems r,elated to shift worik: A field study of Swed~sh milroad workers with irregular work hours. Scand. j. work environ. & health 1 (1975) 254-262. A group of 132 engineers from the north of Sweden was included in the study. A subsample of about 50 subjects was selected f'Or fur,ther laboratory investigations during a light and warm summer period and a dark and cold winter period. The mean for the hours of sleep noted on the sleep records was significantly lower for night work than ]01' day wonk. The ·amount of sleep during night work was significantly less during ,the ,light period than duriing the daI1k pertod as was the amount of :sleep during the day off. Body temperature measured during work followed a daytime pattern and had a ,low ampmude. Potas sium excretion and the blood levelis of cortisol displayed a stable circadian ;r;hyrthm with a dayttme pattern. Many environmental faotors made ,the resuLts of catechola mine data difficult to interpret. The fr,equency of ,reported peptk ulcers was higher in the engineer group than in some other gr,oups.

Many different types of shift work, e.g., two~shWft and three-shift, and ,their influence on the health 'and well-being of workel1S have been studied. Shift wOl1k can be regarded not only as work during ,two or three 8...h periods in 24 h with a defined rotation ,of blocks of different shifts but also as regular night wovk and quite irregular work schemes in whirch night work is included. These types of work may give rise to special problems with medical, psychologioal, and social natures. Investigations of wor~eI1s with regular shift WOI1k are numerous. Not many studies of night wOIikers have been 'made, and only a few investigations of workern with irregular hours have been published (6,33).
In an early ,study of a group of Swedish I1ailroad workeI1s, Bjerner et aL (6) found a higher frequency of peptic ulcers than in the general population.'I1he increase in comparison with the general population was even greater than for gr,oups doing regular :shift NVlo,I1k 1 (1,3,5). The impo!1Janlce of:liactors 'DIther ,thian [shift work per se is, however, indicated by the fact that in some investigations no ,such increase in the frequency of peptic ulcers was demon-straJted (12,34). One important factor in this oonnection is sleep deprivation. Bjerner et al. (5,6,7) £ound th:aJt workers wirth sleep deprivation oomplained of many symptoms, suoh as \tiredness, irritability, difficulties in ooncenJtl'a'tion, ,and headache, when working at night. Wedderburn (36) suggested that a short WOI1k period would be more suitaJble for preventing the development of sleep deficiency. Sleep debts during shift work were also reported by Tune (35).
One of the factors discussed in connec- dustrial group of paper mill workers on a oon'tinuoUfi rbhree--shift :schedule (18) :and also in eXiperimental situaJtions (15,17). tion wtth the 1i11-eMe<its of !Shidit work and! or nighttime work and/or work at irregu-Lar hours ·is the dristJurbance of oircadian rhythms. In animals ,the change from light to darkness and fmm darkness to light is a very strong synchronizer (19) or "Zettgeber" (4). In man, on Ithe other hand, sociarl rellatoons !Seem <to be stiN more impontlant (6,27). A great number of social complication!S wel'e .reported in shift workers when the family and community followed a daytime pattern (32). Many biologic-al parameters, as body temperature (28,29), sodium and potassium excretion (11,25,26), and blood levels of 11-0Hcortisol (10), display rhythmic variations.
Urinary excretion of catecholamines also shows rhythmic variahons (18,24). However, when interpreting catecholamine levels, one must remember that many envillonmental factors affect the levels, e.g., smOiking, coHee and tea drinking, physical exercise, and a high or low ambient itempeI1a1Jure (2,16). Stress rf,actors and their rela1tionship to catecholamine excretion have been studied in 'an in-

METHODS
Questionnaires for infonmation concerning the sense of well-being, social adaptation, lodgings, sleep distuI1bances, and previous illness were completed at home after oral informaJtiJOn had been given by the investigators.
Sleep records wlere filled out at home for a 12-day working rpeI1iod. Hours asleep and ,aw:aike were noted Jor each day and night.
Body temperature was measured every other hour during a work shift with a theI1mometer pLaced under ;the tongue for 10 min. Measurements were not made immediately after physical exercise or coffee drin:king. 'Dhe resuLts were noted on la fonm by the subject himself.
Sodium and potassium in urine were determined by an atomic absorption method (Perkin Elmer).
Adrenaline and noradrenaline in urine were analyzed according to the method of Vlon Euler and Lishajl~o (14).
A self-rating of symptoms and feelings was made on a 10-grade scale. This was done in the same period as the urine sampling. The items rated were: irrita-tion, fatigue, restLessness, stress, boredom, depression, power of concentration, muscular pain, tension, headache, anger, nervousness, ca'lmness, and sleepiness.

SUBJECTS
One special group of Swedish railroad workers was selected, Le., r,ailroad engineem and engineer ·asst9tants Who transported iron ore between KiTuna!(a town in the very nor,1Jh of Sweden) and Vassijaure an: I11he NOTWeg~aJIl homer. This work is the same during the day and night, unlike other types of industrial work. Two examples of work rotation are shown in fig.  1. The questionnaires were given to 132 eng;iJneers and assistants aJnd, Tor ,comparison, to 51 pel'Sons doing 1:Jhree-shi:Dt work (railroad service personnel). From the total group of engineers a sample was ahosen for the laboratory investigation so thJalt work hours ,at differ·ent l1;imes:of the day and ni~t were covered.
' .{,' The investigation was performed during a warm ,and light period of the summer (A) and a cold and dark period of the winter (B). During the summer the midnight sun gives light for 24 h, whereas in December and J·anuary i1 is dark :£01' 24 h. Forty-nine persons paI1ticipated iln ,period A; .and 52 persons, in period B. Unfortunately, the same persons could not always be followed during these ,two periods, and therefore only group results from the various time periods could ,be evalual1;ed. The subjects served as their own controls during a test day when they were free from work.
A control p.eriod, When subjects had their day off, comprised the hours 6: 30-9: 30 a.m, 5: 30-8: 30 p.m, and midnight to 3 a.m. The subjec.ts stayed in the examination room but were allowed to walk about indoors, listen to the radio, read, smoke, and drink coffee.
An .attempt was made to standardize the effect of caffeine on catecholamine leve1s. Coffee was prepared in 'a standardized Wlay, ,aoo 150 m[ were ,giv'eJl to the subjects every hour both during work and during the control periods.
BeDore the start of the trip, on arrival at the border station, and after the return, a blood sample was drawn into heparinized tubes. Plasma was ,separated by centrifug,ation at 2,500 rpm and was then frozen at -20°C until the analyises ,could he made. Blood samples were ,taken at the start and at the end of each control period.
During a train run urine was co:llected in plastic bottles without a pr,eserva,tive. Unne was voided ililito hottle I immediately before the star.t of the run (the exact time was noted) so that the subjects would have empty -bladders at the e:x:act starting point. Bottle II was used in the middle of the run and represented ,an approximate time of 90 min; tthe ,eXladt ltJiJIne was noted. Urine was Vloi'ded iI1Ito bo1Jtle III !at the end of the tr,ip, and ,the exaci time was noted. During the 'Control periods, urine was collected for periods of 90 min in each of '1ihesubperiods described above. At the stations the laboratory staff received the bottles and recol1ded volumes, pH, and specific gravity. 'I1he samples for catechol J Kiruna K. Vassijoure Vj.  The train going from Kiruna to the border .carries the iron ore. A>£ter arriving in Vassijaure the engineers rest for 1-2 h and then return to Kiruna with another train wi'thout ore. In the figure the staI1ting point is marked K and the border station Vj.     amine -analysis were acidified to pH 3 and then frozen. The urine for the electrolyte analysis was fl10ren directly and stor~d irozetIl. ai -20°C. An anall'OgQus procedure was used for the control periOOs.

Questionnaire
The questionnaire forms were given to 132 engillleers or engineer assistanrtJs. The forms were fHIed out properly by 121,   Table 1 shows data on comfort at and attitudes towards work. In the engineer group, 40 % felt at ease, ,but the ,same percentage did not, ,and 37 % wanted to ohiange jIO'bS, but 50 % ,did not. In the thr,eeshift group, 41 % felt at ease, 47 010 did not, 69 010 wanted :to change to work with regular hours, laJlld only 29 010 wanted to continue witlJh the same work. Some negative effects of the work at irregular hours, as repol'ted by the two groups, are listed in table 2. Table 3 shows the frequency of previous and present illnesses in the two groups. None of the persons questioned had had anginal chest pain or a myocardial infarction. Peptic ulcers, verified by X-ray, were found in 13 010 of the engineers and in 8 010 of Ithe three-.shilt ,group. In addiJtion 40 Ofo of rtheengineers and 21 ,fJ/o of the threeshift workers had or thad 'had inoapacities due to jon disorders and hack trouble.
As pointed out in the introduction, problems ,often arise when a man has to sleep in the daytime. Table 4 gives the results of the questionnaire investigation A diagram of the mean body temperature during day and night for a work period is given in fug. 3. It oan be seen that la daytime ,paJttern is ,obtained. However, 'the a:mphtUJde iIs low. Bhysical 'ootivity may have had 'an influence ton the temper.ature during the night.

Boiy temperature
for the engineer .and the ,three-shift groups. Sleep duI'ling tthe day WIaS distuI1bed for 76 and 80 010, respectively, as against 27 and 33 010, respectively, for night sleep. The major disturbing factors were ,traffic in general, railroad traffic, and the noise from playing children.

Sleep records
It can be seen from tahle 5 that the mean of 'the sleep hours oalcU!lated from :the sleep recoI'ds kept during night wortk and work with the starting point late at night or early in the morning differed significantly (p < 0.001) from Ithe mean of ,the sleep houTlS recorded for daytime WOI'k or on a day off. It was also found 'thart; 1Jhe amount of sleep before s,tartmg night work was less than that before work in the daytime. Sleep during tthe day,time and dur,ing the day off in the light summer period was signifiaarutly less 'than that during the dark winter period (p < 0.001 and p < 0.001, respectively). It should be noted that there were no differences in the mean nUJmber of sleep ihoU!I'S Itretween dark and light periods when the wOI1kers did daytiime work Electrolyte excretion Fig. 4 shows Ithe 24-h ex;cretion curve for potassium~alculated for .all >the data (periods A and B), fol,lowed both for the periods of wo:vk and for the day off. A circadian rhyJthm is ,clearly seen in the data. This was not the case for the sodium and chloride excretion data.

Excretion of adrenaline
The results of the excretion of adrenaline are summarized in fig. 5, which gives the pooled data from periods A and B for the fact that r,ailroad workers also work during night hours a daytime pattern is retained.

Rating of symptoms and feelings
The only two parameters which showed measurable variations were sleepiness and fatigue. As eXipeoted, the subjeotive feeling of 'Sleepiness and d'a1tigue ,was highest at night from 12 p.m. to 5 a.m., whether rated during the Icontrol period or during work. The same was the case when the train started eaI'ly in the morning or returned after a night peri,od of work. Thus a diJscrepancy was found between the rated degree of fatigue and the excretion level of adrenaline in urine, whioh was generally high.
A comprehensive description and more details have been given in a Swedish report (23).
hours of work and the described periods of the day off. The most striking effect is the high level as a whole and the higher value £01' the corresponding ooI1ltrol perwd. However, the differences are not statistically signifioant. We could not find any difference in the adrenaline excretion between the summer and winter investigations. The mean ambient temperature was + 9.5"C, and -9.3°C, respective[y. Excretion of noradrenaline 'I1he results of ;the excretion of noradrena-10 line are recorded in tables 6 and 7 for periods A and B, respectively. Generally, the noradrenaline -levels were higher in the winter than in the summer period. There were generally no differences between the excretion values during the 5 work period and the day off.

Levels of cortisol (11-0R-cortisol)
The results are summarized in fig. 6. A circadian rhy,thm of a daytime pattern was found with the highest level at 6 a.m. and the 'lowest at midnight. No difference could be found between the values from work periods and from leisure time. In spite of o 6am 12 6 pm 12 hrs

DISCUSSION
A discrepancy in the frequency of the wish to change jobs and discomfort at work indicates ,1Jhat o1Jher factors might be components of satisfaction with this type of work. Furthermore, access to other types of WOI1k is limited in such a small town.
Economic faotors, such as extra pay for traveling work, might also compensate to some exteilit for negative factors.
Ooncerning the infjormation about previous or present illness, the frequency of joint disorders and back trouble is low when compared with that reported for other Swedish marterials (20,21). The engineers had more symptoms in the loco-  39.0 ± 2.6 41.3 ± 2.9 motor ,system than the three-shift control group. Draft, a common complaint from the engineers, is a possible explanation. There was no report of anginal chest pain or myocardial infarction. This result might be due to selection in the engineer group. The factors of selection could not be studied in the presenrt investigation. The same frequency of peptic disorders was reported by Jorgensen (22) lin an invesHgation of miners from the same town, Kiruna. There 20 Ofo of the persons had long-standing peptic disorders, and 10 Ofo had been away from work for this reason. We found peptic ulcers ,in 13 Ofo 'and gastritis in 26-29 11/0, respectively. 'I1his frequency its higher than what has been found for other groups, e.g., dock workers (30), but it is in accord with the findings of Bjerner et al. (6), who found that traveling railroad personnel had more peptic ulcers ,than Swedes in general.
The reported decrease of sleep during work at night may be a risk factor with respect to safety <rl work, as a decreased amount of sleep is expected to affect the alel'tness of ,the subject (8,9). Bjerner et all. (5) found longer reaction times on a multiple-choice reaction test and an increase iner.ror-ma~ingIa:f.ter sleep deprivahon.
The measurements for studying a circadian rhythm showed a daytime pa,ttern. This held true for body temperature, patassiUJrn. excretion, and ll-OH-oorttsol in blood. Comoy et al. (10) found a reversed pClJttern in whklh the highest v:alue occurl'ed ,at 1 p,m. just befure awakening in workers with constant night wor.k. Whether the daytime pattern of bLood cortisol is of ,any disadvantage in wOl'k dUl'ing rthe night i,s not known at present. Van Loon (29) found that night work for 1 week gave an inverse pattern a,t the end of the week and noted individual differences in such a response. Furthermore, he found that a very rapid return 'to a daytime pattern occurred ,after 1 or 2 days ,away from work. The pattern of potassium excretion in urine follows a stable circadian rhYJthm with a 'lower excretion during Ithe night than during the day (11,25,26). This rhythm is not so easily ,affected by ,the intake of food and beverages as is that for sodium and ohloride excretion.
Many environmental factors may affect oatecholamine excretion. Astrand etal. (2) showed that coldness evoked an increased adrenaline excretion in fishermen. An increase in adrenaline and to some extent noradrenaline, due Ito ,coffee drinking, was reported by Froberg et al. (16). Indubitably the effect of caffeine increased the adrenaline level in our invesrtiga:tion in both test situ<rlions. The unfamiliar test situation on the day off and the extra load of work on the train run under study might also have had a stressor effeot and might have raised the adrenaline values. In laboratory invesrtigartions, in which the same technique of analysis wa's employed, the adrenaline excretion level was 5-10 nanog/min (18); and under resting conditions, 2-3 nanog/ min (13). The same technique was used in the different test peniods,and it is :therefore unlikely that a method artifact would have caused abnormaUy high values. One critical point is that sampling should ideally be done art an acid pH. In the present investigation the urine was acidified ClJt the end of Ithe 'sampling period. (The delay was 1.5-2 h.) One effect of this delay, however, would be lower catecholamine values. A somewha,t 'astonishing fact was that the rated 'aJrnount of fattgue and 'the adrenaline leve:1s did not correlate. A correlation between 'inltenS'ity in emotions and adrena!line e~cret~on has been described by Froberg land his coworkers (17,18,24) and by F1~aOlkenhaeuser (15).