Effect of bright light on tolerance to night work.

Fifteen young (mean age 23.4 years) female nurses engaged in a resuscitation unit and working on a fast rotating shift schedule comprising two consecutive night shifts were exposed to short periods (4 x 20 min) of bright light (2350 lx) during their night duty to test a possible positive effect on their tolerance to night work. Two nights with normal lighting (20-380 lx) and two nights with bright light were compared. The following positive effects of bright light upon psychophysical conditions and performance efficiency were noted: in particular, signs of better physical fitness; less tiredness and sleepiness; a more balanced sleep pattern; and higher performance efficiency (letter cancellation test). This result could not be attributed to shifts of the internal clock although the exact cause remains to be determined. In fact, hormonal excretion and body temperature did not show any effect from bright light. In addition melatonin excretion was not suppressed appreciably by the bright light used.

Night work is a stressful condition which interferes with the normal synchronization of body functions, as well as with social habits. The abnormalities are believed to give rise to negative effects on health and well-being (1)(2)(3).
Tolerance to night work may depend on several factors, including the organizational aspects of the work conditions, the psychophysiological characteristics and social conditions of the work force, and the coping strategies developed by the individuals concerned. As a result, several types of compensatory mechanisms have been proposed to help alleviate these difficulties (4)(5)(6).
One possibility is the use of bright light during shift work. There are two reasons why this use might be effective. First, it has been shown that bright light (of an intensity of~2500 lx) can adjust the body clock as inferred from the phase of melatonin (7) and rectal temperature rhythms (8). Second, and in this instance the mechanism of action might not involve an adjustment of rhythms, bright light can reduce the symptoms of seasonal affective disorders (9), and some of the negative effects of night work can be likened to a mild form of endogenous depression (10).
Therefore we have tested whether brief exposures to bright light could be considered a possible tool for 414 improving tolerance to night work among shift workers. We exposed a group of nurses, engaged in a resuscitation unit, to short periods of bright light during their work shift (insofar as duty allowed) in an attempt to obtain a positive effect on their psychophysical condition and performance.

Subjects
Fifteen nurses, all women in good health, were studied. They were aged 21-29 (mean 23.4) years and had 0.5-7 (mean 2.9) years of experience with shift work. All of the subjects were single except two, who were married but without children. Three subjects were smokers.
All of the subjects worked in an intensive care unit for patients who had just been operated on for heart disease, and they worked in teams of five nurses per shift.
The unit had 10 beds divided between three smaller rooms that were separated from the central control area by glass walls. A separate small kitchen was located beside the unit.

Shift schedule
The nurses were engaged in a continuous, fast rotating shift schedule on an 8-d cycle: The times of the shifts were 0700-1400: morning (M), 1400-2100: afternoon (A), 2100-0700: night (N). Rest days (R) were lived in the conventional way.

Work conditions
The nurses were examined on four occasions dur ing the winter season (December 1990-February 1991) whil e working two consecutive night s under normal light conditions and two consecutive nights under "bri ght light. " In order to interfere as little as possible with normal work acti vity and not to affect the patients, we decided to alter the lightin g inten sity in the nearb y kitchen (where the nurses took their break s during work ) from 100 Ix (normal condition) to 2350 Ix horizontally at the eye level ("bright light" cond ition). Thi s level , the highest pos sible with out glare or unnatur al sensations, was achieved by three qu artziodid e lamp s placed on the ceiling and oriented in such a way as to avoid glare, and three overhead projectors illum inatin g the wall in front of the table , where the nurses could sit, have a drink, or take a light meal.
The nurses were exposed to bright light for four periods of 20 min before starting work (2030-2 100) and then every 2 h during the night dut y (betwee n 2300 and 2400 , 0 I 30 and 0230, and 0400 and (500). Thi s timing was dependent on the rest pauses and was not aimed at ach ie ving a phase delay, as such adju stment need not be adva ntageous in rap idly rotatin g shift systems (8).
In the inten sive care unit the lighting ranged from 20 ("li ght s off') to 380 (" lights on") Ix. Durin g the night shift the lights were generally turned off, but the nurses used an additi onal de sk lamp that gave a light inten sity of 240 Ix.
Two of the teams (5 nurses per team ) worked night s with bright light before night s with normal light. The other team had the oppo site sequence.
One nurse from the fir st two team s did not complete the study because of inte rvenin g famil y problems. Therefore 14 nurses were used in the eva luation of the coll ected data.

Methods
Self-ratings. Subjective evaluations of work load were recorded at the end of the work shift by means of 100-mm visua l analogue scales, accord ing to the task load index of the National Aeronauti cs and Space Administration (NASA) ( I I).
Subjecti ve eva luations of the psychophysical condition s (20 items) experienced durin g each workshift wer e recorded at the end of the work shift by means of 100-mm visual analogue scales, according to Warr ( 12).
Subjective eva luations of mood and physical fitness, on a seve n-point scale according to Halb erg et al (13), and of overall fatigue (5 item s), accordin g to Grandjean et al (14), were made at the start, middle, and end of each shift. The five items comp rising overall fatigu e were stro ng-weak. relaxed-tense, rested-tired, energetic-exhausted, and awake-sleepy. Behavioral performance measures. A test of simple auditory reaction time (50 tones; 1000 Hz, 25 ms, 70 dB; with a random interva l between them ) was given at the start, middl e, and end of each shift. The mean reaction time to respond to the tone by pressing a switch was reco rded .
A search and mem ory (SAM) test, derived from Folkard et al (15), was give n in which the subj ects had to cancel one , three, and five target letters, respectively, in three groups of four lines of 30 letters. Thi s test was performed at the start, middle , and end of the shift . The tim e taken (s) and the percentage of correct answers were recorded. A score was derived by dividing the percentage of correct answers by the square root of the time .
Hormone sec retion . For the determination of plasma cortisol (g . dl 1) I ml of blood was collected from nine nurses at the start, middle, and end of the shifts. The serum was stored at -20°C for 7-20 d before the analysis, carried out by means of fluor imetri c enzyme immun oassa y (Baxter).
The urinary excretion of 6-sulfatoxymclatonin, adrenaline, and noradrenal ine was determined. All of the nurse s empt ied thei r bladder at the beginning of each shift and then collected their urine at the middle and end of the shift. Volum e and conc entration were determ ined for each half of the shift. 6-Sulfatoxymelaton in was determined by radioimmun oassay accordin g to Bojkowski et al ( 16), and the rate of excretion (ng . min-I) was calculated. Adrenal ine and noradre naline were determined by high-pressur e liqu id chro matography (Bio-rad") , and the rates of excretion (nm ol . min 1) were calculated.
Oral temp erature and acti vity logs. Th e nurses recorded their oral temp erature (digital thermometer, 1/20 Cel sius) on retiring and rising and approximately every 4 h du ring the workdays and the two rest days; they also kept a diary in which they reported time s of sleep and tim es of leisure, work and tra vel to work. A method of " purifyi ng" the data was used ( 17,18) to gain an estimate of the endoge nous component of oral temp erature. Briefly, the method consists of correc ting the raw temperature each hour for the fall produced by slee p and the rises produced by leisure, travel, and work. Thi s correcti on is bas ed upon the main type of activit y durin g the previous hour and is computed for each subject and da y. One can then use such purified data to estimate the phase of the end ogenous component of the temperature rhythm by calculating the phase shift of a standard set of " normative endog enous" data required to match the pur ified data set most accurately. As a control, raw (masked) data were assessed by the sam e techniqu e.
Statistical analysis. The data were analyzed by twoway (light, day) and three-way (light, day, occasion) analyses of variance (ANOVA) with repeated measures, Tukey tests, and paired Student t-tests by means of SYSTAT packa ge (19).

Results
Subjective evaluation of the work load and the psychophysical conditions at the end of the shifts The subjective evaluation of the work load did not differ at all between the four test nights, while some

Normal light
Bright light difference s emerge d concernin g the psychophysiological conditi ons perceived during the shifts. Of the 20 items considered, none of the nine "positive" items (alert, calm, energetic, enthusiastic, good mood, lively, optimistic, relaxed, satisfied) showed any significant effect of bright light exposu re, while two of the II "negative" items (annoyed, anxi ousworried, bored, depressed, demoralized, discouraged, fatigued, restless, tense, tired, wanting in energy) showed a posi tive effec t of the bright light. In particular, the nurses expressed feeling less "anxious-    I).
When the components of fatigue were con sidered separately, they all showed slightly, but not significantly, lower scores at the end of the second night shift with bright light , as compared with the other nights, whereas physical fitnes s was slightly higher (figure 2).  the second halves of the shifts, both having higher levels in the first half (adrenaline: three-way ANOVA F = 12.4 31, P < 0.0 I; noradren aline: ANO VA F = 12.807, P < 0.01) in connection with higher work activity, but no differences due to bright light exposure or between the first and second nights were detected . In all cases , the changes in concentration durin g the night shifts indica ted that the rhythms had not I I I  A significant positive effect of the bright light exposure emerged concerning the feeling of "restoredtired." The three-way ANOVA showed a significant "light x occasion" interaction (F =3.442; P =0.047) when all of the occasions of the four shifts were compared. Furthe rmore, considering the differences between the start and end of each shift, the two-way ANOV A showed a positive effect both for the factor "light" itself (F =4.333; P = 0.058) and for the "light x day" interaction (F = 3.442 ; P = 0.047), a finding indicating less tiredne ss dur ing bright light exposure, particularly durin g the second night.
As concerns sleepiness ("awake-s leepy") in particular, no significa nt differences emerged in the three-way ANOVA among the 12 occasio ns, but the lowest score of sleepiness was recorded at the end of the second night with bright light exposure, in spite of the fact that the group had the highest score of sleepiness before starting the same shift.

Behavioral performance measures
The auditory reaction times showed a progressive and significant increase from the start to the end of each shift (three-way ANOV A: F = 12.503, P < 0.00 I) with a slight improvement on the second night, either with or without bright light exposure (three-way ANOVA: F = 4. 132, P = 0.077). On both nights with bright light, the times were slightly, but not sig nifica ntly, lower (three-way ANOVA: F = 2.243, P =0.173) (table 2).
On the other hand , the letter cancellatio n test (SAM) showed a significantly better performance (higher index) during the two nights with bright light in comparison with the other two nights, particularly at the end of the shifts (three-way ANOVA: F = 6.869, P = 0.021). No significant difference was found betwee n the first and seco nd nights in either bright or normal light (three-way ANOVA: F = 2. 130, P =0.168).

Hormone secretion
Plasma cortisol did not show any effect of bright light exposure (F =0.673, P =0.459), but there were significant differences between the concentrations at the start, middle, and end of the shifts (F = 65.477, P < 0.00 I), presumably resultin g from circadian rhythm. In addition, there was a slight increase in concentration dur ing the second night under both normal and bright light (F = 11.342, P =0.012) (table 3).
Urinary excretion of 6-sulfatoxymelatonin showed a clear increase during the second half of every shift (three-way ANOVA : F = 18.921, P <O.OOI), and a slight effect emerged concerning the "light x day" interaction (three-way ANOVA: F = 4.53, P = 0.053). However, there was a lack of depression of melatonin excretion durin g exposure to bright light (figure 3).
Adrenaline and noradrenalin e excretio n also showed sig nificant differences between the first and Se ll/IIIJ Work Environ Health 1993. vol 19 , no 6 a Resul ts of t he t hree-way anal yses of variance are repor ted in t he text. adj usted fully to night work but, instead, appea red to have retained a phasing much more appropriate to day work .

Oral temperature
The mean oral temperatu res are shown in tab le 4. A three-way ANOY A indicated that there was a significant fall in oral tempera ture during night work , but that there were no significant effects of bright 418 • Results of the three -way analy ses of varian ce are report ed in the texl.
light nor sig nifican t differences between days. Thi s change in temperature was foun d with both the raw (F =3.66 1, P =0.04 1) and purified (F =7.120, P = 0.(04) data. When a two-way AN OYA was performed upon the temperature differences between the start and the  When the shifts in rhythm were considered (figure 4), there was a tend ency for the raw data to be shifted more in the presence of brig ht light (F = 1.898, P = 0.193). With the puri fied data this tendency was also found , and the amount of shift was slightly greater on the seco nd night (F = 2.633, P = 0.131). However, by far the clearest change was seen with the three-way ANOV A when any effects of purifi ed versus raw data were considered also. In this case the delays with the raw data were significantly (F = 32.535, P<O .OOI ) grea ter.

Sleep
Th e nurses' sleep patterns were slig htly different dur ing the two periods on night duty (as shown in figure 5), alth ough the total amount of sleep was equi valent.
Dur ing the shift with bright light the nurses showe d a more balan ced sleep pattern and a sig nifican tly shorte r slee p period bet ween the two night shifts ( t paired =-2.55, P =0.024). Thi s behavior was not significa ntly corre lated with any of the subjective feelin gs of tiredn ess, fatigue, or sleep iness.

Discussion
The res ults indi cated that there was a small positive effect of bright light upon some mood, psychoph ysical, and perform ance varia bles in our sa mple of nurses. Thu s with bright light the y felt less anxious and less wantin g in energy , and they performed the SAM test better. These effects were small , however, and did not extend to the other measures of mood, psychoph ysical variables, or performance. In addition there were no sig nific ant changes in the physiologica l variables, as judged by the phases of hormone rhythm s or pur ified temp erature data.
The result s suggest that the positive effects of the bright light used in this study cannot be attributed to shifts in the internal clock, although the exact cause remai ns to be deter mined. Mas king effects upon the temperature data were clear and could have led to misleading infere nces being dra wn. For thi s reason , the shifts obse rved in the phase of the masked , but not pur ified, temp eratur e data were probably caused by changes in activi ty.
Other accounts exi st which indicate far more pronoun ced effects than those measured in our study, and also a wider variety of varia bles, incl uding deep body temperatur e (20--24). Variations in protocol -part icul arly in the intensity, duration, and timing of the light expos ure -are likely to acco unt for many of the differenc es. Expos ures to light of high intensity and for prolonged periods of time (see, eg, references 2 1 and 22) were not feasible for our subjects while they were actively invo lved in intensive care nursing. Indeed , we noted that melatonin sec reti on was not suppressed appre ciably by the bright light used in our study. Even so, other resea rchers (20,24) have used light with an intensity similar to ours and observed shifts in severa l variables, including body temperature. Howe ver , these studie s were different, both with regard to the duration of light expos ure and with regar d to the details of timing.
Our light exposure schedule was determin ed not only by the constraints imposed by the work conditions in intensive care, but also by our objective to remove as many of the negati ve effects of nightwork as possible without cau sing a shift in the body clock. We did not wish to adj ust the body clock because 4 19 .~.. ---_._---Scand J Work Environ Health 1993, vol 19, no 6 the shift sys tem used was a rapidly rot ating one and the refore ph ase adj us tment of body rhythms was not re quired . In othe r words, ligh t of an int e nsity and durat ion sufficie nt to adj us t th e bod y c loc k (8, 25) might be of value whe n adj ustment to permanent nightwork or to a slow rotation of shi fts is involved, but it would be of mu ch less an ad vantage -and might even prove to be di sorienting -in a rapidly rot at ing shi ft sys tem.
We suggest that the use of bri ght light during ni ghtwor k need s to be tailored to whe ther or not adjustm en t of circadi an rhythms is required. For rapidl y ro tati ng shift sys te ms, a balance must be struc k betwe en using a light ex posure prot ocol that is too wea k and produces no effects and one that is too strong and produces large phase shifts.
Th e pr esent results sugges t that th e ki nd of light exposure used in thi s study might be ben eficial in rapidly rotating shift system s. This as sumption needs to be confirmed by oth er fiel d studies, but it is worth stressi ng that any support measure to inc rease tolerance to nightwork , as ex pos ure to bright light (26), sho uld be adopted only after the ad op tion of the best possibl e shift sche du le according to ph ysiol ogi cal , psych ol ogi cal , and social cri teria (27), and no t in spite of them.
Fin all y, our results obvious ly re late only to the immediate effec ts of brigh t light. Wh et her, and fo r ho w lon g, such po sitive effects w ould remain if th e use of br ight light were to continue is no t known. Nor is it kno wn if suc h short-term ad vant ages would al so redu ce the lon g-ter m negati ve effects upo n health and we ll-bei ng .