Sleep, Sleepiness and Health Complaints in Police Officers: The Effects of a Flexible Shift System

Transcription

1 Industrial Health 2007, 45, Original Article Sleep, Sleepiness and Health Complaints in Police Officers: The Effects of a Flexible Shift System Claire Anne ERIKSEN and Göran KECKLUND* National Institute for Psychosocial Medicine (IPM), Department of Public Health Sciences, Karolinska Institutet, Box 230, SE Stockholm, Sweden Received March 30, 2006 and accepted November 6, 2006 Abstract: The aim of the study was to study the effects of a flexible shift system (based on selfdetermined work hours) with respect to sleep/wake complaints and subjective health. The comparison group was a rapidly rotating shift system, with frequently occurring quick returns. A secondary aim was to examine the relation between work hour characteristics indicating compressed or difficult rosters (e.g. number of workdays in a row, frequency of quick returns and long work shifts) and subjective sleep and sleepiness, within the flexible shift system group. The sample of the analysis included 533 randomly selected police officers, of which 26% were females. The participants answered a questionnaire. The results showed that the flexible shift system group did not differ with respect to sleep/wake complaints and subjective health. However, the flexible shift group obtained more sleep in connection with the shifts, probably because of longer rest time between shifts. Thus, they worked less quick returns and long work shifts. The association between work hour characteristics and sleep/wake complaints was weak in the flexible shift group. Instead, sleep/wake problems were mainly associated with the attitude to work hours. Key words: Shift work, Night work, Sleepiness, Sleep, Subjective health, Self-determined work hours, Police officers Introduction It is well known that the design of the shift system has a strong impact on sleep/wakefulness, health and well being 1). Although there is no ideal shift system, ergonomic recommendations have been proposed in order to design shift systems that minimize the risk for sleep, health and social problems 1, 2). For example, the number of night shifts should be reduced to 2 3 in a row to avoid accumulating sleep debt, severe sleepiness and increased accident risk 2, 3). Other recommendations concern the length of the shifts, the rest time between shifts, and the timing of the shifts 4, 5). Individual satisfaction with the shift schedule is also important and seems to reflect tolerance to shift work. Axelsson et al. 6, 7) found that those that are dissatisfied with the schedule showed a higher frequency of sleep/wake *To whom correspondence should be addressed. problems, complaints of insufficient recovery and lower testosterone levels. Thus, individual working time preferences might also be an important factor for avoiding negative consequences of shift work. Working time control, operationalised as being able to influence starting and ending times of a workday, scheduling of days off, the opportunities to take breaks and deal with private matters during the workday etc, was an independent predictor of subjective health and sickness absence in a series of Finnish studies 8 10). A low level of control increased the risk of future health problems especially among women with families and the authors proposed that a high degree of work time control could help workers integrate their work and private lives successfully 9). In another study high control over working times could reduce the adverse effects of work stress on sickness absence 10). Baltes and coworkers found that flexible work schedules were associated with high job satisfaction and a more positive attitude to work hours 11). Consequently,

2 280 CA ERIKSEN et al. it is important that shift systems permit some flexibility and even be individualized 12, 13), especially since self-determined work hours may reduce the work-home conflict and improve the social life situation 13). On the other hand, the knowledge of how shift workers select work hours (where possible) is very limited. Some authors have claimed that they might give too much priority to short-term social needs, which result in compressed work hours, and that work hour variability increases 14, 15). As a consequence, self-determined shift systems might increase the frequency of quick returns, long work shifts and many work days in a row, which increase the risk for disturbed sleep, sleepiness, accidents, and poor health 16). The topic of the present study was to explore the effects of a flexible shift system used by police officers with respect to sleep, sleepiness and health. Police officers often have difficult shift rosters 17, 18) and sleep/wakefulness complaints seem to be the cardinal symptoms 19 21). Garbarino et al. 22, 23) found that sleep disturbances, sleepiness and fatigue related accidents were common among police officers, in particular for those that worked shifts. However, also work-related stress due to traumatic events contributes to the high frequency of sleep disturbances among police officers 24). Swedish police officers have a tradition of working extremely rapidly rotating shift rosters with quick returns. A quick return means that the rest time between shifts is only 8 or 9 h. It is well known that quick returns reduce sleep length and increase sleepiness 7, 25). A shift system intervention on police officers that increased the rest time between shifts improved subjective sleep, self-reported health and biological risk factors for cardiovascular disease 26). However, during the last 5 10 yr flexible shift systems have increased in the Swedish police force and are nowadays more common than the rapidly rotating rosters. The flexible shift system called period planned work hours involves a high degree of individual flexibility. The employer presents a list of the need for staffing during the shifts for a period of 4 6 wk. The shift workers then select their own pattern of work hours within these limits. If some shift lacks staff, the employer can order officers to work. There are few studies on flexible shift systems based on self-determined work hours and their effects on sleep and health, although they are quite common (in Sweden) in health care, transport and retail. A study by Lowden and Åkerstedt 27) showed no differences in subjective health when they compared a flexible shift system with a traditional shift system. However, the studied group did not work nights. One might hypothesize that being permitted to choose one s work shifts would facilitate a striving towards high alertness and good sleep. On the other hand, social or economical factors might invite promotion of such factors at the expense of optimal sleep, alertness and health. The aim of this study was twofold. First, to compare a rapidly rotating shift system, with frequently occurring quick returns, with a flexible shift system based on self-determined work hours on sleep quality, sleepiness and health. Secondly, to examine the relation between work hour characteristics indicating compressed or difficult rosters (e.g. number of workdays in a row, frequency of quick returns and long work shifts) and subjective sleep and sleepiness, within the flexible shift system group. Since satisfaction with the schedule has turned out as a strong predictor of sleep/wake problems in previous studies (e.g. Axelsson et al. 7) ), the variable attitude to work hours was added as an independent variable in the latter analysis. The intention behind the second aim was to identify whether certain work hour components related to compressed work hours could be related to sleep/ wakefulness problems. Materials and Method A questionnaire was distributed to a random sample of 2,000 employees within the Swedish police force. The response rate was 76%. Among the responders 619 subjects were three-shift workers. The mean age for this group was 41 yr, the mean employment time was 16 yr and 26% were females. The study was approved by the local ethical committee at Karolinska Institutet (# ). The questionnaire included two questions related to the shift system. The first question referred to whether they worked on a traditional three-shift roster schedule, or on a flexible shift system. Two types of flexibility existed: flextime and period planned work hours. Flextime means that only start and end times of the shifts can be determined by the individual within a certain limit (±1 h) but without the possibility to select work shifts. Period planned work hours usually involve a high degree of individual flexibility and means that the employee can determine scheduling of work days and days off. The second question referred to whether they had a rapid (compressed work hours with maximum 11 h of rest time between shifts) or a more slowly rotating (2 4 similar shifts in a row, with almost no quick returns between shifts) shift system. According to these questions, four groups could be identified a flexible group (having period planned work hours, n=363), a rapidly rotating (or compressed) group (n=170), a slowly rotating group (n=45) and an other (n=40) group that mainly included subjects that had a shift system including flextime. One Industrial Health 2007, 45,

3 FLEXIBLE SHIFT SYSTEM 281 subject did not respond to the questions and could not be classified. The slowly rotating group and the other group were excluded from the analysis due to the relatively small group size. The flexible group and the rapidly rotating group did not differ with respect to age (mean was 42 yr in both groups), gender (female; flexible: 25%, rapidly rotating: 28%) and civil status (cohabiting; flexible: 83%, rapidly rotating: 82%). Rated workload did not differ between the groups (1 very low 5 very high, the mean across shifts was 3.4 for both groups). The shift change time between the night and the morning shift was around h (±30 min) for both groups. A typical shift sequence for the rapidly rotating shift system is night, evening and morning shift with 8 11 h of rest between shifts. Approximately one third of the shifts were worked at night in both groups. The length of the shifts is usually between 8 and 10 h, and 12-h shifts are rare. The questionnaire included 19 pages. In the present analysis, questions related to work hours, sleep, sleepiness/ fatigue, and health were used. The questions (variables) that were used, including the response scales, are presented in Table 1. The questions concerning work hours included attitude to work hours and whether one was satisfied with the possibilities to influence work hours. There was also a question of whether one would like to avoid night shifts. The flexible group also had a separate question of whether they could affect their work hours (yes or no). There were 7 questions that described the characteristics of the work hours. These questions referred to how often they have quick returns ( 9 h of rest between shifts), long work shifts (>10 h), at least 6 work shifts in a row, weekends (both Saturday and Sunday) off, 4 d or more off, only 1 d off between work sequences, and a question referring to how many night shifts in a row they usually worked. There was also a question of having a second job. The sleep questions were based on the Karolinska Sleep Questionnaire (KSQ 28) ). The questions referred to symptoms during the last 3 months. Three indices were calculated: disturbed sleep (DSI: difficulties initiating sleep, repeated awakenings with difficulties going back to sleep, premature awakening and disturbed/restless sleep, Cronbach s alpha: 0.82), awakening difficulties (AI: feeling of exhaustion at wake-up, difficulties awakening and not being well-rested, Cronbach s alpha: 0.73) and sleepiness (sleepy at work, sleepy during leisure time, involuntary dozing off at work, involuntary dozing off at leisure time, and need to fight sleep to stay awake, Cronbach s alpha: 0.81). The subjects also reported bedtime, wake-up time and sleep latency in relation to different shifts and days off. These measures were used for calculations of sleep length prior to the morning and evening shift, and after a night shift. Two questions related to recovery were asked. The first one refers to whether they have enough time for recovery between work shifts and the second one refers to whether they have enough time for recovery on days off (between work periods). The health section of the questionnaire included the Hospital Anxiety and Depression Scale (HAD 29), Cronbach s alpha: 0.87), the Shirom-Melamed Burnout Inventory (SMBI 30), Cronbach s alpha: 0.95) and a question of selfrated health in general. There was also a multiple item question related to health complaints during the last three months. Three indices were calculated: muscular-skeletal symptoms (pain in shoulders, pain in the upper back, pain in the lower back and pain in the wrist and hand, Chronbach s alpha: 0.80), cognitive symptoms (difficulties learning, unusual forgetfulness, unable to make decisions, make rash decisions, and became empty in the head, Cronbach s alpha: 0.84), and gastrointestinal symptoms (poor appetite, stomach complaints, heart burn, diarrhea, and constipation, Cronbach s alpha: 0.80). The statistical analysis included t-tests and χ 2 for the group comparisons. Correlation analysis and multiple regression analysis were used to study the covariation between work hour variables and sleep and sleepiness complaints. All work hour variables and age were used as predictors. The three indices of KSQ were used as dependent variables. Results Comparison between shift systems The results for the comparison between the flexible shift system and the rapidly rotating system are presented in Table 2. The flexible group was more satisfied with the possibilities to influence work hours. They also had less quick returns (short rest between work shifts) and long work shifts, but more often 6 work days in a row and worked more often at least 2 night shifts in a row. Both long (4 d off) and short (1 d off) free periods occurred more frequently in the flexible group. The groups did not differ with respect to sleep/ sleepiness complaints and health symptoms, with the exception of muscular-skeletal symptoms that occurred somewhat more frequently in the flexible group. The flexible group showed a non-significant trend of having more second jobs (p=0.08). Figure 1 shows the results for sleep length. The flexible group reported longer sleep in association with the morning (t-value: 2.9, p<0.001, df=528) and the evening shift (t=6.6,

4 282 CA ERIKSEN et al. Table 1. Presentation of the questions and the response scales Variable Attitude to work hours Possibilities to influence work hours? Would you like to avoid night shifts (work between 23.00h and 06.00h)? How often do you have: rest time 9 h (quick returns) work shift >10 h How often do you work at least 6 shifts in a row? How many free weekends do you have per month? Response scale 1: very negative, 2: negative, 3: neither, 4: positive, 5: very positive 1: very satisfied, 4: rather satisfied, 3: neither, 4: rather dissatisfied, 5: very dissatisfied Yes/No 1: almost never, 2: seldom, 3: sometimes (several times/ months), 4: often (once every/week), 5: always (several times/week) 1: almost never, 2: seldom (sometimes/year), 3: sometimes (some time/mo.), 4: often (several times/mo.) 1: none, 2: one, 3: two, 4 :three or more How often do you have 1: almost never, 2: seldom (sometimes/year), 4 d off (excluding vacation) 3: sometimes (some time/mo.), 4: often/several times/mo. How often do you have 1: almost never, 2: seldom (sometimes/year), 1 d off between work periods? 3: sometimes (some time/mo.), 4: often/several times/mo. How many nightshifts in a row do you usually work? Second job Sleep need Sleep length Napping (per shift) Do you get sufficient sleep? Sleep quality (how do you sleep in general?) during work days during days off How often do you have enough time to recover between shifts? How often do you have enough time to recover during free days? DSI AI Sleepiness index Self-rated health HAD SMBI Cognitive symptoms Muscular-skeletal symptoms Gastrointestinal symptoms 1: one shift, 2: two shifts, 3: three shifts, 4: four shifts, 5: five shifts or more Yes/No Hours Hours Yes/No 1: definitely sufficient, 2: sufficient, 3: somewhat too little, 4: clearly too little, 5: definitely too little 1: very good, 2: rather good, 3: neither, 4: rather poor, 5: very poor 1: almost every shift, 2: sometimes/wk, 3: sometimes/ mo, 4: sometimes/yr, 5: never 1: almost every free period, 2: most of the periods, 3: some period/mo, 4: some period/yr, 5: never 1: low (no problems) 6: high (problems every day) 1: low (no problems) 6: high (problems every day) 1: low (no problems) 6: high (problems every day) 1: very good, 2: rather good, 3: neither, 4: rather poor, 5: very poor 0: min 42: max (high anxiety and depression) 1: low burnout score 7: high burnout score 1: low (never) 5: high (always, almost every day) 1: low (never) 5: high (always, almost every day) 1: low (never) 5: high (always, almost every day) DSI: Disturbed Sleep Index, AI: Awakening Difficulties Index, HAD: Hospital Anxiety and Depression Scale, SMBI: Shirom Melamed Burnout Inventory. Industrial Health 2007, 45,

5 FLEXIBLE SHIFT SYSTEM 283 Table 2. Means (± standard error of the mean, SE) or per cent Variable Rapidly rotating Flexible t-value/ d.f. Mean ± SE/per cent Mean ± SE/per cent χ 2 Attitude to work hours (1 5 positive) 3.5 ± ± Possibilities to influence work hours (1: satisfied 5: dissatisfied) 3.3 ± ± *** 495 Avoid night shifts (yes) 44% 46% Quick returns, several times/month or more 78% 62% 10.4** 1 Long work shift (>10 h), several times/month or more 55% 43% 6.8** 1 6 work shifts in a row, sometimes/month or more 6% 21% 15.9*** 1 At least 2 free weekends per month 82% 84% days off in a row, at least once/month 27% 50% 21.6*** 1 1 d off between work periods, at least once every month 27% 61% 49.3*** 1 2 nightshifts or more in a row 32% 64% 43.6*** 1 Second job 7% 12% Sleep need (hours) 7.9 ± ± Insufficient (not enough) sleep 49% 50% 0 1 Poor sleep quality during work days 23% 19% during days off 8% 8% 0 1 Have time to recover between shifts, sometimes/month or more 32% 40% Have time to recover during free days, most of the free periods 36% 36% 0 1 DSI (1 6 poor) 2.6 ± ± AI (1 6 poor) 2.7 ± ± Sleepiness index (1 6 poor) 2.3 ± ± Poor self rated health 7% 10% HAD (0 42 max) 8.7 ± ± SMBI (1 7 max) 3.0 ± ± Cognitive symptoms (1 5 poor) 1.9 ± ± Muscular-skeletal symptoms (1 5 poor) 1.9 ± ± * 528 Gastrointestinal symptoms (1 5 poor) 1.8 ± ± *p<0.05, **p<0.01, ***p<0,001, DSI: Disturbed Sleep Index, AI: Awakening Difficulties Index, HAD: Hospital Anxiety and Depression scale, SMBI: Shirom Melamed Burnout Inventory, d.f.: degrees of freedom. p<0.001, df=456), and showed a tendency for longer day sleep after the night shift (t=1.9, p=0.06, df=437). There was no difference for sleep in connection with days off (t=0.7, p=0.46, df=500). Napping did not differ between the groups and was mainly found in association with the night shift (19% reported regular naps prior to night work). A gender analysis within the flexible shift group showed no significant differences with respect to the work hour characteristics listed in Table 2. In the flexible group 67 subjects (21%) answered no to the question of whether they could influence their work hours. This group was compared with those who answered yes (256 subjects, 40 subjects did not answer the question). The age and gender difference was non-significant in both groups, however, there were more subjects living alone (non-cohabiting) in the group that answered that they could affect work hours (21% vs. 7%,

6 284 CA ERIKSEN et al. χ 2 =6.8, p<0.01, df=1). The groups were compared with respect to all the variables presented in Table 2 (except the question possibilities to influence work hours ). The group that could influence work hours had a more positive attitude to work hours (3.8 ± 0.06 vs. 3.3 ± 0.12, t=3.9, p<0.001, df=321), but did not differ with respect to other work hour variables. There were no differences in sleep, sleepiness and health variables. Correlation between work hour characteristics, attitude to work hours and sleep/wake complaints In order to investigate the relation between work hour characteristics of the flexible shift system and sleep/ wakefulness complaints, three multiple regression analyses were carried out. The dependent variables were the three indices of the KSQ. The correlation coefficients between the indices and work hour variables (listed in Table 1) and age are presented in Table 3. A general observation is that attitude to work hours shows the strongest associations with sleep/wake complaints. However, a wish to avoid night shifts, long work shifts, many work shifts in a row and the number of night shifts in a row showed moderate to weak correlations with the indices. The multiple regression analyses are also presented in Table 3. Attitude to work hours showed significant relations with all dependent variables. A negative attitude to work hours was associated with more complaints of sleep and sleepiness. A wish to avoid night shifts was also associated with difficulties awakening. For the awakening difficulties index, working many shifts in a row became significant. In the next step, attitude to work hours, wish to avoid night shifts and possibilities to influence work hours were removed. The results yielded no significant association between the work hour variables and DSI and the sleepiness index. However, for AI, a significant association was found for working 6 shifts in a row (R 2 =4.4%, beta=0.14, p<0.001). Thus, awakening difficulties increased with increased frequency of working many (6) shifts in a row. In order to understand the relation between attitude to work hours and the other work hour variables a multiple regression analysis was calculated using the same model as in the previous analyses (and with attitude to work hours as the dependent variable). Possibilities to influence work hours (beta= 0.45) and wish to avoid night shifts (beta=0.32) were significant and the amount of explained variance was 34.4% (F=15.6, p<0.001, df=10/268). Thus, a negative attitude to work hours was associated with dissatisfaction with the possibilities to influence work hours and a wish to avoid night shifts. In the next step, the sleep/ Fig. 1. Mean (and SE) sleep length in connection with different shifts and days off. Asterisks (*: p<0.05, **: p<0.01) indicate a significant group difference. M: morning shift, E: evening shift, N: night shift. wake indices (DSI, AI and sleepiness) were introduced as predictors in the regression analysis. Possibilities to influence work hours (beta= 0.43) and a wish to avoid night shifts (beta=0.28) remained significant, and sleepiness (beta= 0.17) became significant. The amount of explained variance increased to 39.3% (F=14.7, p<0.001, df=13/263). Thus, frequent complaints of sleepiness were associated with a negative attitude to work hours. Discussion The main finding of the present study was the lack of significant differences between the shift systems with respect to sleep/wake complaints and subjective health. The slightly higher frequency of muscular-skeletal complaints of the flexible shift system group were probably only a coincidence and should not be interpreted as a sign of poorer health in general. However, the flexible shift system was associated with more sleep ( 17 min) in connection with the morning shift, and about 1 h more sleep in connection with the evening shift. There was also a trend (p=0.06) towards more sleep ( 15 min) after the night shift as well. The lack of findings with respect to sleep/wake problems and subjective health is probably related to the working hour characteristics of both groups. Compressed work hours may cause accumulating sleep debt and sleepiness, and should be avoided according to the ergonomic criteria for shift schedule design 1). The rapidly rotating group had a compressed shift system, with frequently occurring quick returns and long work shifts. However, the flexible shift system group also had rather compressed work hours. Although they had less quick returns and long work shifts, Industrial Health 2007, 45,

7 FLEXIBLE SHIFT SYSTEM 285 Table 3. Correlational analysis and multiple regression analysis with DSI, AI and the sleepiness index as dependent variables. Only the significant predictors in the multiple regression analyses are presented in the table DSI (r) AI (r) Sleepiness (r) Age * 0.01 Frequency of quick returns Frequency of long work shifts 0.15** * Many (6) work shifts in a row 0.12* 0.16** 0.10 At least 2 free weekends/month Many (4) days off in a row Few (1) days off in a row * Number of night shifts in a row 0.11* 0.15** 0.07 Attitude to work hours 0.26*** 0.25*** 0.32*** Possibilities to influence work hours * Wish to avoid night shifts 0.16** 0.16** 0.18** Significant predictors of the multiple 1) Attitude 1) Attitude 1) Attitude to regression analyses to work to work work hours hours hours (beta= 0.31) (beta= 0.23) (beta= 0.20) 2) Many shifts in a row (beta=0.13) 3) Avoid night shifts (beta= 0.14) Explained variance 7.5% 11.0% 11.0% F-value 3.0*** 4.1*** 4.1*** DSI: Disturbed Sleep Index, AI: Awakening Difficulties Index, r: correlation coefficient, ***: p<0.001, **: p<0.01, *: p<0.05, df for the multiple regression analysis: 11/265. they reported more often long work sequences (with 6 shifts in a row), longer spells with night shifts, and having only 1 d off between work sequences. In addition, quick returns were relatively common in the flexible shift group. The flexible shift group also reported a higher occurrence of free periods with many days off. Thus, the latter finding suggest that social needs were given priority in the flexible group presumably at the expense of optimal sleep and alertness. The increased sleep length for the flexible shift group is probably related to the lower frequency of quick returns. Short rest time between shifts is a strong determinant of sleep length 5, 25, 31, 32). Some studies show that chronic sleep debt is a risk factor for cardiovascular disease, diabetes and fatigue-related accidents 33 35). Thus, in the long run, the longer sleep lengths for the flexible shift group might decrease the risk for disease development and accident involvement. However, it should be pointed out that the group differences in sleep length for the morning and night shift was small, which probably explains the lack of group difference with respect to sleepiness complaints. The groups did not differ in their attitude to work hours, although there was a trend (p=0.08) towards a more positive attitude in the flexible shift system group. However, there was a clear, and expected, difference in the possibility to influence work hours that was in favor of the flexible shift system group. This should be regarded as a positive factor that facilitates coping with shift work. The social life situation could benefit of increased possibilities to influence work hours and reduce the interference of work hours on leisure time and family activities 9, 13). Although a majority (79%) of the shift workers in the flexible shift system group reported that they could influence their work hours, a relatively large group (21%) reported that they could not. This group was more negative to their work schedule, but did not differ with respect to sleep/wake complaints and subjective health. This was somewhat

8 286 CA ERIKSEN et al. surprising since individual flexibility and influence on working hours has been regarded as a factor that facilitates coping with shift work, and possibly improves sleep and health 16). However, one should keep in mind that the groups did not differ with respect to work hour characteristics. This raises the question of whether work hour characteristics are more important than influence on working hours. The weak relation between work hour characteristics and sleep/wake complaints in the flexible group do not support this hypothesis. The frequency of long work shifts, many work shifts in a row, and many night shifts in a row showed low correlations (r<0.17) with sleep/wake problems. Moreover, these associations disappeared when attitude to work hours and wish to avoid night shifts entered as predictors in the multiple regression analyses. Thus, in accordance with previous findings 7) a negative attitude to work hours was associated with poor sleep and sleepiness, although it should be pointed out that the amount of explained variance was rather low. The low amount of explained variance is probably related to the low frequency of sleep/wake complaints in this sample, which resulted in a restriction of range and as a consequence low correlations. However, sleep complaints may also be related to work-related stress, e.g. due to traumatic events 24), and non-work related factors such as health, physical environment (e.g. noise level) etc. Hence, it is possible that such factors may show stronger association with sleep quality than work hour characteristics. Another interesting observation was the weak association between work hour characteristics and the attitude to work hours. In contrast, the attitude was related to a wish to avoid night shifts and whether one was satisfied with the possibilities to influence work hours. Also sleepiness problems were associated with dissatisfaction of work hours. A positive attitude to a shift system probably requires some autonomy over the working hours. Hence, a shift worker could be satisfied with a shift system that is ergonomically poor if it fits with the individual s preferences. The association between avoiding night work and the attitude to work hours is intuitive since this shift is the most difficult one for most three-shift workers. This study has some limitations that might have influenced the results. Firstly, the cross-sectional design does not permit any conclusions about causality. Longitudinal designs or evaluation of shift change interventions are preferable when the effects of different shift systems are studied. However, the present study did have a random sample that was representative for the Swedish police force. This strength would not have been possible if the design had involved a shift system intervention. Another problem with crosssectional designs can be the influence of extraneous factors. However, the groups did not differ in workload, gender distribution, family situation and age. The amount of night work and weekend work was also similar in both groups. It would also have been advantageous to have several comparison groups. Unfortunately the more slowly rotating shift system (which was the schedule that showed the strongest correspondence with the ergonomic shift design criteria) included too few subjects. Another methodological problem concerns the questions about work hour characteristics. It might be difficult for the flexible shift group to remember their work hours across time, in particular if they vary a lot. Thus, we don t know whether the questionnaire technique was a reliable method for quantification of their working hours. The solution for this problem is to use the participants factual rosters. It may also be argued that a questionnaire is not the optimal method for quantifying sleep/wake complaints. On the one hand, it is well known from many diary studies that sleep and sleepiness show considerable variation depending on type of shift 7, 32). On the other hand, due to practical reasons a questionnaire is often the only feasible alternative. Furthermore, the KSQ is a validated instrument, and has shown strong covariation with diary measures of sleep quality 36). In conclusion, the flexible shift system did not differ from the rapidly rotating shift system with respect to sleep/wake complaints and subjective health. However, the flexible shift group obtained more sleep in connection with the shifts, probably because of longer rest time between shifts. They worked less quick returns and long work shifts, but also longer spells with night shifts, more workdays in a row and had more often only 1 d off between work periods. This suggests that the shift workers in the flexible shift system selected rather compressed work hours, and that long periods with days off were given priority. The association between work hour characteristics such as frequency of quick returns, long work shifts, and many shifts in a row, and sleep/wake complaints was, however, weak in the flexible shift group. Instead, sleep/wake problems were associated with the attitude to work hours. Acknowledgements This study was financed by the Ministry of Industry, Employment and Communication. We thank Claes-Eric Claesson at the National Police Board and Prof. Torbjörn Åkerstedt at IPM for their support of the study. References Industrial Health 2007, 45,

9 FLEXIBLE SHIFT SYSTEM 287 1) Knauth P, Hornberger S (2003) Preventive and compensatory measures for shift workers. Occup Med 53, ) Knauth P (1998) Innovative work time arrangements. Scand J Work Environ Health 24 (Suppl 3), ) Folkard S, Tucker P (2003) Shiftwork, safety and productivity. Occup Med 53, ) Rosa R (1995) Extended workshifts and excessive fatigue. J Sleep Res 4 (Suppl 2), ) Kecklund G, Åkerstedt T (1995) Effects of timing of shifts on sleepiness and sleep duration. J Sleep Res 4 (Suppl 2), ) Axelsson J, Åkerstedt T, Kecklund G, Lindqvist A, Attefors R (2003) Hormonal changes in satisfied and dissatisfied shift workers across a shift cycle. J Appl Physiol 95, ) Axelsson J, Åkerstedt T, Kecklund G, Lowden A (2004) Tolerance to shift work-how does it relate to sleep and wakefulness? Int Arch Occup Environ Health 77, ) Ala-Mursula L, Vahtera J, Kivimäki M, Kevin MV, Pentti J (2002) Employee control over working times: associations with subjective health and sickness absence. J Epidemiol Community Health 56, ) Ala-Mursula L, Vahtera J, Pentti J, Kivimäki M (2004) Effect of employee worktime control on health: a prospective cohort study. Occup Environ Med 61, ) Ala-Mursula L, Vahtera J, Linna A, Pentti J, Kivimäki M (2005) Employee worktime control moderates the effects of job strain and effort-reward imbalance on sickness absence: the 10-town study. J Epidemiol Community Health 59, ) Baltes BB, Briggs TE, Huff JW, Wright JA, Neuman GA (1999) Flexible and compressed workweek schedules: a metaanalysis of their effects on work-related criteria. J Appl Psychol 84, ) Kandolin I, Härmä M, Toivanen M (2001) Flexible working hours and well-being in Finland. J Hum Ergol (Tokyo) 30, ) Demerouti E, Geurst SAE, Bakker AB, Euwema M (2004) The impact of shiftwork on work-home conflict, job attitudes and health. Ergonomics 47, ) Giebel O, Janssen D, Schomann C, Nachreiner F (2004) A new approach for evaluating flexible working hours. Chronobiol Int 21, ) Jenssen D, Nachreiner F (2004) Health and psychosocial effects of flexible working hours. Rev Saude Publica 38, ) Costa G, Åkerstedt T, Nachreiner F, Baltieri F, Carvalhais J, Folkard S, Dresen MF, Gadbois C, Gartner J, Sukalo HG, Härmä M, Kandolin I, Sartori S, Silverio J (2004) Flexible working hours, health, and well-being in Europe: some considerations from a SALTSA project. Chronobiol Int 21, ) Knauth P, Rutenfranz MJ, Karvonen K, Undeutsch F, Klimmer F, Ottman W (1983) Analysis of 120 shift systems of the police in the Federal Republic of Germany. Appl Ergon 14, ) Vila B, Morrison GB, Kenney DJ (2002) Improving shift schedule and work-hour policies and practices to increase police officer performance, health, and safety. Police Quarterly 5, ) Peacock B, Glube R, Miller M, Clune P (1983) Police officers responses to 8 and 12 hour shift schedules. Ergonomics 26, ) Phillips B, Magan L, Gerhardstein C, Cecil B (1991) Shift work, sleep quality, and worker health: a study of police officers. South Med J 84, ) Vila B (2006) Impact of long work hours on police officers and the communities they serve. Am J Ind Med 49, ) Garbarino S, Nobili L, Beelke M, Balestra V, Cordelli A, Ferrillo F (2002) Sleep disorders and daytime sleepiness in state police shiftworkers. Arch Environ Health 57, ) Garbarino S, De Carli F, Nobili L, Mascialino B, Squarcia S, Penco MA, Beelke M, Ferrilla F (2002) Sleepiness and sleep disorders in shift workers: a study on a group of Italian police officers. Sleep 25, ) Neylan TC, Metzler TJ, Best SR, Weiss DS, Fagan JA, Liberman A, Rogers C, Vedantham K, Brunet A, Lipsey TL, Marmar CR (2002) Critical incident exposure and sleep quality in police officers. Psychosom Med 64, ) Kurumatani N, Koda S, Nakagiri S, Hisashige A, Sakai K, Saito Y, Aoyama H, Dejima M, Moriyama T (1994) The effects of frequently rotating shiftwork on sleep and family life of hospital nurses. Ergonomics 37, ) Orth-Gomer K (1983) Intervention on coronary risk factors by adapting a shift work schedule to biological rhythmicity. Psychosom Med 45, ) Lowden A, Åkerstedt T (2000) Einführung selbst gewählter arbeitszeiten im einzelhandel auswirkungen auf arbeitszufriendenheit, gesundheit und sozialleben. Z Arb Wiss 54, (in German). 28) Åkerstedt T, Knutsson A, Westerholm P, Theorell T, Alfredsson L, Kecklund G (2002) Sleep disturbances, work stress and work hours: a cross-sectional study. J Psychosom Res 53, ) Lisspers J, Nygren A, Söderman E (1997) Hospital anxiety and depression scale (HAD): some psychometric data for a Swedish sample. Acta Psychiatr Scand 96, ) Melamed S, Kushnir T, Shirom A (1992) Burnout and risk factors for cardiovascular disease. Behav Med 18, ) Lowden A, Kecklund G, Axelsson J, Åkerstedt T (1998) Change from an 8-hour shift to a 12-hour shift, attitudes, sleep, sleepiness and performance. Scand J Work Environ Health 24 (Suppl 3), ) Sallinen M, Härmä M, Mutanen P, Ranta R, Virkkala J, Müller K (2003) Sleep-wake rhythm in an irregular shift system. J Sleep Res 12, ) Partinen M, Putkonen PTS, Kaprio J, Koskenvuo M, Hilakivi I (1982) Sleep disorders in relation to coronary heart disease. Acta Med Scand 660 (Suppl), ) Ayas NT, White DP, Al-Delaimy WK, Manson JE, Stamper MJ, Speizer FE, Patel S, Hu FB (2003) A prospective study

10 288 CA ERIKSEN et al. on self-reported sleep duration and incident diabetes in women. Diabetes Care 26, ) Stutts JC, Wilkins JW, Osberg JS, Vaughn BV (2003) Driver risk factors for sleep-related crashes. Accid Anal Prev 35, ) Axelsson J, Kecklund G, Åkerstedt T, Ekstedt M, Menenga J (2002) A comparison of the Karolinska Sleep Questionnaire and the Karolinska Sleep Diary: a methodological study. J Sleep Res 11 (Suppl 1), 8. Industrial Health 2007, 45,