Effects of circadian typology on sleep–wake behavior of air traffic controllers


address: Dr Vincenzo Natale, University of Bologna, Department of Psychology, Viale Berti Pichat 5, Bologna 40127, Italy.
Email: natale@psibo.unibo.it


The effects of circadian typology on sleep–wake behavior in shiftworkers were investigated using wrist actigraph in 18 air traffic controllers (ATC), nine morning types and nine evening types, working in a backward 1-1-1 super rapid rotation shift schedule. The ATC wore a wrist actigraph continuously over 6 days (3 days on duty and 3 days off duty). Evening types presented more flexible sleep habits and slept significantly less than morning types. Regardless of circadian typology, the morning shift tended to reduce the amount of sleep whereas night shift produced a decrease in daily activity.


In many cases today work needs to be organized over a period of 24 h. The worker adapts himself to the work schedule in different ways, depending on a complex interaction between endogenous (age, gender, personality etc.), and exogenous (shift system, task load, social and environment conditions etc.) factors.

Studies have been carried out to identify individual differences (particularly circadian typology) that are able to modulate adaptation to shiftwork and, hence, to influence work performance indirectly. It seems that evening types adapt themselves more easily than morning types to changes in the sleep–wake habits and, in consequence, to working activities involving shiftwork systems.1 Folkard maintains that a possible way of attenuating problems stemming from shiftwork may be to select people who are naturally more inclined to tolerate shiftwork (i.e. evening-type individuals).2 The few longitudinal researches carried out on this hypothesis, however, have not produced convincing results.3

Air traffic controllers (ATC) belong to those professions that are exposed to the possible negative effects that bad work organization of shifts may have on circadian rhythms, and hence performance efficiency. Counterclockwise, rapidly rotating shift schedules are often employed to compress work during the week, minimizing the number of night shifts and maximizing the days off.4 These schedule types involve a very short daytime break before the night shift, usually the last shift of the working week.

The purpose of the present study was to investigate the effects of circadian typology on sleep–wake behavior by continuous recording of wrist actigraph for 6 days. Although in recent years actigraphy has become an essential tool in sleep medicine,5 only few researchers have used wrist actigraphs to study shift workers’ rest–activity cycles.6,7


The study was carried out at a medium-sized airport in Northern Italy and included 18 male ATC, aged between 22 and 49 years (mean: 34.00 years). All ATC gave written informed consent for their participation in the trial. Shiftwork was organized in a super rapid rotating shift system over a 6 day cycle with backward rotation: afternoon (1 p.m.−8 p.m.), morning (7 a.m.−1 p.m.), night (8 p.m.−7 a.m.) and three rest days.

Actigraph data (AMA Model 32, Ambulatory Monitoring, Ardsley, NY, USA) on the non-dominant wrist were continuously recorded during one shift cycle (6 days). The actigraph was initialized with a sleep–wake mode (zero-crossing mode, sampling frequency of 60 s, and 18 amplifier settings). Each ATC started the actigraph recording phase at 1 p.m., the beginning of the afternoon shift.

The actigraph data were analyzed using the action-w software program. Action-W helped to identify ‘up’ (high motor activity, i.e. the waking hours over 24 h) and ‘down’ (low motor activity, i.e. the longer sleep period over 24 h) intervals for each recorded day. Three ‘up’ intervals (A, the mean value of the up interval on working days; B, the up following night shift; and C, the mean value of the up intervals on rest days) and three ‘down’ intervals (D, the down before morning shift; E, the down following night shift; and F, the mean value of the down intervals on rest days) were determined for each subject. The following parameters were considered for each ‘up’ interval: mean activity (number of movements per minute), nap length (expressed in minutes) and the number of sleep episodes exceeding 5 min. Mean activity, sleep onset latency (expressed in minutes), the number of waking periods exceeding 5 min, sleep length (expressed in minutes), and sleep efficiency (the ratio between total sleep time and total bed time) were considered for each ‘down’ interval.

The subjects compiled an Italian version8 of the Composite Scale of Morningness,9 a circadian questionnaire composed of 13 multiple-choice items in which the subject is requested to indicate their own life rhythms and habits as far as going to sleep and waking up are concerned. Each item is given a score from 1 to 4 when the response patterns are limited to four and from 1 to 5 for all the items implying five response patterns. Because it is an added-score scale, the theoretical total varies from a minimum of 13 to a maximum of 55. Subjects were subdivided by median split criterion into morning (n = 9; mean CS score = 40.00 ± 7.83) and evening (n = 9; mean CS score = 29.44 ± 8.16) types (t16 = 2.80; P < 0.05). The median value obtained in the present sample (34.5) agrees with previous works.10,11


The data regarding ‘up’ and ‘down’ intervals are reported in Tables 1 and 2, respectively. A series of two-way anova was computed to examine shift (three levels) and circadian typology (two levels) effects on each parameter considered.

Table 1. Parameters for the ‘up’ intervals
 Morning types
(mean ± SD)
Evening types
(mean ± SD)
Mean value
for total sample
  1. A, mean value of the up interval on working days; B, the up following night shift; C, mean value of the up intervals on rest days; bold, significant differences.

Mean activity (no. movements/ min)
A202.84 ± 21.04202.41 ± 14.53202.65 ± 17.54
B176.77 ± 30.09183.90 ± 19.34180.33 ± 25.03
C195.34 ± 15.84206.45 ± 12.98200.90 ± 15.17
Mean191.65 ± 22.32197.59 ± 15.62 
Nap length (min)
A21.59 ± 23.7657.32 ± 43.14 39.45 ± 38.46
B 47.04 ± 49.35 22.10 ± 23.78 34.57 ± 39.71
C 17.93 ± 19.3729.18 ± 31.40 23.56 ± 25.96
Mean 28.85 ± 30.83 36.20 ± 32.77 
No. sleeping bouts >5 min
A 0.72 ± 0.831.94 ± 1.84 1.33 ± 1.52
B1.55 ± 1.880.67 ± 0.50 1.11 ± 1.41
C 0.74 ± 0.830.86 ± 0.83 0.80 ± 0.81
Mean 1.01 ± 1.18 1.16 ± 1.06 
Table 2. Parameters for the ‘down’ intervals
 Morning types
(mean ± SD)
Evening types
(mean ± SD)
Mean value for total sample
  1. D, the down before morning shift; E, the down following night shift; F, mean value of the down intervals on rest days; bold, significant differences.

Mean activity (no. movements/min)
D 12.38 ± 4.93 12.88 ± 4.86 12.63 ± 4.76
E 12.58 ± 3.60 10.00 ± 4.59 11.29 ± 4.21
F 14.72 ± 5.88 12.42 ± 3.73 13.57 ± 4.92
Mean 13.22 ± 4.80 11.78 ± 4.39 
Sleep latency (min)
D  9.67 ± 4.69  6.94 ± 0.95  8.30 ± 3.57
E 13.22 ± 19.45  6.22 ± 3.69  9.72 ± 14.14
F 11.09 ± 5.69  9.51 ± 3.75 10.30 ± 4.74
Mean 11.32 ± 9.94  7.56 ± 2.80 
No. waking bouts >5 min
D  2.11 ± 1.36  2.39 ± 2.34  2.25 ± 1.86
E  2.33 ± 1.66  1.55 ± 1.42  1.94 ± 1.55
F  2.58 ± 1.59  2.28 ± 1.41  2.43 ± 1.47
Mean  2.34 ± 1.54  2.07 ± 1.72 
Sleep length (min)
D426.33 ± 43.12326.33 ± 33.67376.30 ± 36.71
E443.55 ± 131.01387.89 ± 110.78415.72 ± 121.13
F439.81 ± 82.79381.90 ± 74.71410.80 ± 82.12
Mean436.57 ± 85.64365.39 ± 73.05 
Sleep efficiency (sleep time/total bed time)
D 93.90 ± 3.69 92.81 ± 5.13 93.35 ± 4.37
E 91.69 ± 3.10 94.64 ± 4.17 93.17 ± 3.88
F 90.27 ± 6.19 91.83 ± 2.95 91.06 ± 4.77
Mean 91.95 ± 4.33 93.10 ± 4.08 

As regards the ‘up’ interval, a significant effect (F2,32 = 15.75; P < 0.0001) of shift on mean activity was found: the mean activity of the B (180.33) interval was significantly lower than the A (202.65) and C (200.90) intervals. This effect was the same for both typologies. A significant interaction between shift and circadian typology was found for nap length (F2,32 = 4.43; P < 0.05) and number of sleeping bouts (F2,32 = 4.66; P < 0.05); in particular evening types took more and longer naps in the A and C intervals in comparison with morning types.

Results regarding ‘down’ intervals showed a significant (F1,16 = 6.09; P < 0.05) effect of circadian typology on sleep length: regardless of shift condition, evening types slept less minutes (365.39) than morning types (436.57). With regard to the effect of shifts, a significant effect on sleep length (t = 2.16; P < 0.05) was found when the D and F intervals were compared: the sleep length of D (376.30) was significantly lower than that of F (410.80), the effect being the same for both typologies.

Further analysis was carried out on the first sleep episode after the night shift, which in some cases occurred immediately after the end of the night shift (daytime), in others the night after the night shift (night-time). Seven of nine evening-type subjects recovered sleep immediately after the end of the night shift, whereas seven of nine morning-type subjects waited till the following night (χ22 = 5.88; P < 0.05).


Circadian typology significantly influences shiftworkers’sleep–wake behavior. In particular, evening-type subjects prepare to go to bed immediately after the end of the shift, whereas morning types wait for the night after. Moreover, evening types tend to make greater use of naps as strategy for recovering sleep. The behavior patterns, differentiated per typology, once again seem to confirm the greater flexibility of the evening-type subjects’ sleep–wake cycle. In contrast, evening types sleep less than morning types regardless of shift. Such ‘sleep deprivation’, albeit slight, could derive from a difficulty of evening types to synchronize their sleep–wake cycle to work rhythms, and more in general to social rhythms.12 In this case the greater use of naps could imply worse, and not better, adaptability to shift-work in evening types compared to morning types. We feel it is important that further ad hoc experiments be carried out to better understand such key questions.

Finally, it is interesting to note that morning shiftwork prompts a steady reduction in sleep. As a result, working a counterclockwise rotating shift cycle can lead ATC to begin the night shift with a significant sleep deficit. In accordance with current research,13,14 we believe that this result should be taken into consideration when planning work schedules because over time the sleep reduction caused by a poor work schedule could have negative effects on cognitive efficiency, particularly attention.


Research supported by Ministry of University and Scientific and Technological Research ex 40% 1997.