Article Text
PDF
Abstract
Objectives The purpose of this study was to use objective registry data to prospectively investigate the effects of quick returns (QR, <11 hours of rest between shifts) and night shifts on sick leave.
Methods A total of 1538 nurses (response rate =41.5%) answered questionnaires on demographics and personality and provided consent to link this information to registry data on shift work and sick leave from employers' records. A multilevel negative binomial model was used to investigate the predictive effect of exposure to night shifts and QR every month for 1 year, on sick leave the following month.
Results Exposure to QR the previous month increased the risk for sick leave days (incidence rate ratio (IRR)=1.066, 95% CI 1.022 to 1.108, p<0.01) and sick leave spells (IRR=1.059, 95% CI 1.025 to 1.097, p<0.001) the following month, whereas night shifts did not. 83% per cent of the nurses experienced QR within a year, and on average they were exposed to 3.0 QR per month (SD=1.6). Personality characteristics associated with shift work tolerance (low on morningness, low on languidity and high on flexibility) were not associated with sick leave, and did not moderate the relationship between QR and sick leave.
Conclusions We found a positive linear relationship between QR and sick leave. Avoiding QR may help reduce workers' sick leave. The restricted recovery opportunity associated with QR may give little room for beneficial effects of individual characteristics usually associated with shift work tolerance.
- Quick returns
- Sick leave
Statistics from Altmetric.com
What this paper adds
Association between shift work and sick leave is likely to be schedule specific, and current research has only found support for a positive association between fixed evening shifts and sick leave.
The present study is based on objective records of shift work and sick leave, and suggests a positive linear relationship between quick returns (<11 hours of rest between shifts) and sick leave.
Quick returns are usually not a necessity in order to maintain staffing, and should therefore be considered for elimination in shift schedules.
Introduction
Shift work is a staffing practice intended to maintain production or provide service across all 24 hours of the day. In particular, shift work is a way of organising the staff where the ‘workers succeed one another at the workplace so that operation hours exceed the hours of work of individual workers’ (p. 14).1 The way shift systems are designed can have a major impact on workers' sleep and health. Shift schedules that cause circadian misalignment (especially night shifts) and those that do not permit sufficient recovery opportunity during the shift cycle, are often highlighted as detrimental for the workers' health.2 The European Working Time Directive have stated that workers are entitled to a minimum daily rest period of 11.0 consecutive hours per 24.0-hour period.3 However, some workers are exempted from this requirement through local agreements in certain European countries. Thus, <11 hours of rest between shifts is common among healthcare workers in some countries,4 ,5 and these short rests between shifts are referred to as quick returns.6 Quick returns occur in rotating shifts schedules, and primarily in changeovers from evening to day/morning shifts.6 However, quick returns can also transpire in the changeovers from night to evening shifts and from day to night shifts, but these are less common.6
Studies have shown that quick returns have negative effects on sleep duration, sleepiness and fatigue; but it is less clear whether quick returns affects the occurrence of diseases and the ability to work.6 A commonly used measure of health is sick leave, which refers to absence from work that is attributed to sickness.7 Sick leave is by most professionals accepted as a global measure of physical and social functioning. We know of only one previous study that has examined the relationship between quick returns and sick leave, where nurses were assessed with questionnaires before and after a change in their shift schedule that eliminated quick returns.8 The study reported no effect of introducing more recovery time between evening and morning shifts on subjectively reported sick leave, although the intervention led to improvements on a range of other health-related variables.8
Moreover, the evidence linking shift work in general to sick leave remains inconclusive (for a review see ref. 9). This is likely due to the fact that shift work is often treated as a unified exposure, whereas specific schedule characteristics, such as fixed evening shifts, appear to be associated with increased sick leave.9 At present, the evidence linking night shift to increased sick leave remains inconclusive,9 which is somewhat surprising considering the fact that working night shifts is generally regarded as the most problematic shift work schedule.10 One possible explanation for this is that there may exist a greater selection of healthy and tolerant individuals to night shifts, where the individuals who are working nights have a better ability than their daytime working counterparts to work shifts without experiencing adverse effects. Shift work tolerant individuals will often score low on the trait morningness, low on languidity and high on flexibility.11 ,12 As far as we know, no previous study has investigated these personality characteristics against sick leave as an outcome measure.
Current research on the association between shift work and sick leave is encumbered with several limitations. More studies addressing the effects of specific shift work characteristics on sick leave are called for.9 The complex nature of sick leave warrants better control of confounders in predictor models than what is often the case in research on this topic. In this regard, many studies often fail to take into account sick leave at baseline, which may result in an exaggeration of the observed strength of the association. The vast majority of studies also rely on subjective measures of sick leave and shift work exposure. In cases where objective records of sick leave are included, it is often with an overemphasis on long-term absence (14 days or more).
The purpose of the present study was to compensate for many of the aforementioned shortcomings by using registry data on shift work exposure as well as on short-term and long-term sick leave. We investigated whether exposure to quick returns and night shifts could predict later sick leave, and to what extent personality traits associated with shift work tolerance predicted sick leave and/or moderated any such prediction by shift schedule characteristics.
Methods
Subjects and procedure
The present study was based on registry data of shift work and sick leave from nurses and nurses' assistants (89.2% and 10.8%, respectively; hereinafter referred to as ‘nurses’) working at Haukeland University Hospital, one of four public hospitals in western Norway. The registry data were retrieved from local records kept by the hospital. All nurses employed in a 50% position or more (>18 hours/week) were asked to participate in the study. A total of 1538 nurses (response rate=41.5%) answered questionnaires on demographics and personality and provided consent to linking this information with their registry data on shift work and sick leave. The nurses were first asked to complete the questionnaire electronically via a link sent to their work email (991 responders). Non-responders were sent at least two reminders by letter, together with a prepaid return envelope (547 responders). Nurses responded to the questionnaire during the period lasting from May 2013 to February 2014. The nurses worked at different departments in all parts of the hospital, including internal medicine, surgical care, oncology, emergency, psychiatry and maternal care. Participation was voluntary, and the project was approved by the Regional Committee for Medical and Health Research Ethics, Northern Norway (number 2013/526/REK nor) and by the Norwegian Data Inspectorate (13/00569-2/CGN).
Registry data and definition of variables
Registry data of shift work (payroll data) and sick leave history were retrieved from the local employee records at Haukeland University Hospital. Payroll data included actual date and start and stop times for every shift carried out by all nurses during the year 2013. In the current study, night shifts were defined as ≥3 hours of work performed between 23.00 and 05.00 hours. Evening shifts were defined as ≥3 hours of work after 18.00 hours. In cases where the evening shifts were overlapping with night shifts and the definition of night work applied, the latter got priority based on the assumption that night shifts are more disruptive for the workers. Quick returns were defined as changeovers between shifts in a rotating schedule that permitted <11 hours of rest. Changeovers between shifts of ≤1 hour were viewed as a break, and the two shifts were collapsed.
Records of sick leave included actual date of every full day of absence from work attributed to sickness for the individual nurses during the year 2013. The sick leave data thus included information about short-term and long-term absence. We distinguished between sick leave days (number of days of absence due to sickness) and sick leave spells (number of periods of absence due to sickness) as outcome variables.13
Questionnaire data
The questionnaire assessed demographic variables (sex, age, marital status and children living at home), and the nurses were also asked how many years they had worked shifts. In addition, the nurses were assessed on standardised questionnaires measuring the personality traits morningness (the Diurnal Scale (DS)),14 flexibility and languidity (the revised Circadian Type Inventory).
The DS comprises seven items designed to assess the morningness–eveningness dimension, where a higher score indicates a tendency towards morningness.14 Participants are asked questions pertaining to their preferred bedtime and rise time, and their preferred time for activities during the day. The DS has shown adequate convergent validity in relation to other scales measuring the morningness–eveningness dimension.15 In the current study, the Cronbach's α of the DS was 0.60.
The revised Circadian Type Inventory (rCTI) comprises 11 items. Five items assess flexibility and six assess languidity.12 High scores on flexibility reflect better ability to sleep and work at odd times, whereas high scores on languidity indicate difficulties overcoming drowsiness and feelings of lethargy following sleep loss. The rCTI has previously demonstrated good reliability and validity in a working sample.12 In the present study, Cronbach's α for the flexibility and languidity subscales were 0.83 and 0.72, respectively.
Data analytic approach
STATA/SE V.14.1 (StataCorp. Stata Statistical Software: Release 14. College Station, Texas: StataCorp LP, 2015) was used for all analyses. For the purpose of analysis, the day-to-day registry data throughout the year 2013 was restructured so that the number of specific shift characteristics, sick leave days and sick leave spells were aggregated over individual and calendar month. Each individual then had up to 11 pairs of prospective observations, with exposure to shift characteristics as independent variable (IV) in 1 month, and sick leave the following month as dependent variable (DV) (IVmonth1→DVmonth2; IVmonth2→DVmonth3… IVmonth11→DVmonth12). Aggregating over calendar months, as opposed to using shorter or longer time brackets, allowed us to study the cumulative effects of quick returns (mean number of quick returns per month was three, see Results section) while also having a reasonable chance of capturing the short-term effects of quick returns.
Owing to the clustered nature of the data (ie, observations nested within individuals) a multilevel model statistical approach was employed. Since the outcome variable was count data (sick leave), the appropriate choice for modelling was the Poisson distribution. However, the Poisson model assumes that the mean is equal to the variance. In order to take into account the overdispersion of sick leave (ie, variance greater than mean), a negative binomial model was used. A multilevel negative binomial model was used to investigate the predictive effect of exposure to quick returns and night shifts every month for a year, on sick leave the following month. Personality characteristics associated with shift work tolerance were investigated as predictors and possible moderators of the effects of the shift work characteristics on sick leave. Since quick returns are closely related to evening shifts, the same model was also run in an additional analysis where ‘quick returns’ was replaced by ‘evening shifts’ as independent variable.
Results
Of the 3706 nurses invited to participate a total of 1538 returned the questionnaire (response rate=41.5%). The majority of the nurses were women (87.7%). Their mean age was 42.5 years (SD=12.0) (range=22–73 years). Sixty-seven per cent of the nurses had at least one night shift, 86% had at least one evening shift and 83% had at least one quick return during the study year. Mean number of nights and evening shifts worked per month were about the same, with 4.5 (SD=3.5) night shifts and 4.6 (SD=2.4) evening shifts per month. The mean number of quick returns worked per month was 3.0 (SD=1.6). Additional descriptive data of shift characteristics are shown in table 1. In terms of sick leave, 72.2% had at least 1.0 sick leave day during the study year, the mean number of sick leave days and sick leave spells per month were 0.9 (SD=1.6) and 0.4 (SD=0.7), respectively.
Descriptive statistics on type and number of shifts worked by nurses during the year 2013 (N=1538)
Table 2 presents the result from the multilevel negative binomial model. The results show that the number of quick returns experienced in 1 month both significantly increased the risk for sick leave days (incident rate ratio, IRR=1.064, p=0.003) and sick leave spells (IRR=1.061, p=0.001) the following month. For every 1-unit increase in quick return, the expected log count for sick leave days the following month was 0.062 (ie, IRR=1.064); which furthermore tells us that for every 3-unit increase in quick returns (mean number of quick returns experienced by nurses per month were 3; SD=1.6) the expected log count for sick leave days the following month was 0.186 (ie, IRR=1.205). This suggests a 20.5% increased risk of being on sick leave the following month compared to those not exposed to quick returns. Number of night shifts experienced in 1 month was not significantly associated with sick leave the following month. Also, none of the personality characteristics (morningness, languidity, flexibility) were associated with sick leave. The personality characteristics were also examined as moderators of the relationship between quick returns and sick leave; however, no significant interaction effects were found. Prior sick leave significantly increased the risk for future sick leave, in terms of sick leave days (IRR=1.188, p<0.001) and sick leave spells (IRR=1.360, p<0.001). Number of years engaged in shift work (shift work experience) was associated with a significantly reduced risk for sick leave spells (IRR=0.994, p=0.017), but not for sick leave days (p=0.151). None of the other control variables (sex, age, marital status, children living at home, number of hours worked the exposure month) were related to sick leave.
Result from the multilevel negative binomial model of exposure to shift characteristics in 1 month on sick leave the following month for the year 2013 (N=1538, number of observations=17 169)
Since quick returns most often occur in succession to an evening shift in a rotating schedule, the abovementioned analysis was repeated with evening shifts as predictor instead of quick returns (including evenings both followed by and not followed by a rapid return to day shift). The results showed no significant effects of number of evening shifts experienced in 1 month on sick leave the following month, neither for sick leave days (IRR=1.021, 95% CI 0.988 to 1.056, p=0.213) nor for sick leave spells (IRR=1.021, 95% CI 0.994 to 1.050, p=0.126) (results not reported in table).
Discussion
The purpose of the present study was to prospectively investigate the effects of night shifts and quick returns (<11 hours of rest between shifts) on sick leave. The findings showed that quick returns were a significant predictor of sick leave, whereas night shifts were not. Nurses who were exposed to one quick return within a month had a 6.4% increased risk of being on sick leave the following month, compared to those not exposed to quick returns. Eighty-three per cent of the nurses experienced quick returns within a year, and on average they were exposed to three quick returns per month (SD=1.6). Working three quick returns per month was calculated to correspond to a 20.5% increased risk of being on sick leave the following month, compared to the risk of those without quick returns. The frequency of quick returns has previously been shown to be positively related to sleep problems. One study found four quick returns per month or more to be associated with poor sleep quality, more frequent short sleeps and more problems unwinding.16 We know of only one previous study that investigated the relationship between quick returns and sick leave. That study investigated the effect of a change from backward to forward rotating schedule on subjectively reported sick leave (among other outcomes), where the authors found no effect of introducing more recovery time between evening and morning shifts on subjectively reported sick leave.8 However, we would argue that the design of the present study achieves a significantly higher level of precision in terms of teasing out the isolated effects of quick returns. We also used objective records of sick leave, which may help explain the difference in findings between the two studies. Our findings suggest that there is a positive linear relationship between quick returns and future sick leave, and that reducing or avoiding quick returns may help reduce sick leave.
The majority of quick returns experienced by the nurses in the present study were of 9 hours or less (63.6%), and few were of <7 hours (2.2%). The time for real recuperation between shifts is however shorter due to commutes back and forth from work, time to unwind before sleep and time for personal needs, before returning to work the next day. The restricted time for rest induced by quick returns has been associated with shortened sleep duration and symptoms of insomnia,4 ,16 and reports of sleepiness and fatigue during the second shift in a quick return.6 One study on nurses reported that quick returns were related to the occurrence of needlestick injuries.17 Some studies have also linked quick returns to poor work–family balance,16 ,18 although other studies have not found this.19 All things considered, poor sleep, poor performance and domestic strain are frequently reported residues of quick returns, all of which have been associated with increased sick leave in previous studies.20–23
Comparing different shifts, previous research has found fixed evening shifts to be most consistently associated with long-term sick leave (from >5 days).9 ,24 ,25 In the present study, sick leave was assessed from the first day of absence and evening shifts primarily occurred in a rotating schedule. However, we did not find that evening shifts per se increased the risk for sick leave. The proposed explanations for the higher sick leave following fixed evening shifts in previous studies, include adverse social consequences associated with evening shifts, difficulties unwinding after the shift (and subsequent sleep problems), working evening shifts also imply that workers have less contact with their supervisors and evening shifts are often associated with more repetitive and physical handling work tasks.9 ,24 The fact that evening shifts alone did not influence subsequent sick leave in the present study, suggests that the negative effects of quick returns may be greater than the negative effects of evening shifts in a rotating schedule. However, our analysis does not allow us to make conclusions about the effects of working fixed evening shifts, or several consecutive evening shifts, on sick leave as an outcome measure.
Previous studies investigating the association between night shifts and sick leave have shown inconsistent results.9 No significant relationship between night shifts and sick leave was found in the present study. Recent studies on nurses have associated quick returns with a range of sleep-related consequences, and at the same time reported relatively less detriments of night shifts.4 ,16 A proposed explanation for this is that quick returns may impose a greater strain on workers than that of night shifts.16 Also, the work load on nurses is often reduced during night shifts, as in-patients usually do not require the same level of care during nights compared to the rest of the day.26 Hence, the strain associated with higher work load on day and evening shifts, which are the shifts most often involved in quick returns, may in part account for the greater detriments of quick returns relative to night shifts.16 Another point worth noting is that quick returns imply a rotating schedule, while night shifts occur in fixed and rotating schedules. Rotating schedules may be associated with more detriments than fixed schedules; even though the evidence on the differential effects of the two schedules on sick leave overall remains inconclusive.9 The fact that the vast majority of night shifts in the present study were part of a rotating schedule (98.1% of night shifts) prevented us from investigating the differential effect of the two forms of night shifts on sick leave. It has been proposed that there may exist a greater selection of healthy and tolerant individuals to night shifts, compared to those exposed to quick returns.16 However, as pointed out above, the fact that the majority of nurses working nights in the present study did so in a rotating schedule, suggests that those who worked quick returns and those who worked nights largely constituted the same individuals.
A range of individual characteristics have been linked to shift-work tolerance; including low scores on the personality trait morningness, low scores on languidity and high scores on flexibility.11 Yet in the present study, neither of these characteristics were associated with sick leave. As far as we know this is the first study to examine the relationship between these personality characteristics and sick leave. Furthermore we found no interaction effect between these characteristics and quick returns on sick leave. The beneficial effects of the aforementioned traits, however, are likely shift specific. For example, eveningness personality (low on morningness) appears to be beneficial when working nights,11 whereas vigorous (low score on languidity) and flexible individuals appear to show better circadian adjustment to rotating (irregular) work schedules.12 In contrast to the other shift characteristics, quick returns, by definition, limit the recovery opportunity for all workers alike, which may give little room for individual differences to play a role in terms of shift work tolerance.
Prior sick leave emerged as a substantial risk factor for later sick leave in the present study, which is well-known from previous research.27 Workers who reported more experience with shift work had a slightly decreased risk for sick leave spells than those with less experience, which supports the notion of self-selection of vulnerable individuals out of shift work.28 These findings highlight the importance of adjusting for the history of shift work exposure and for sick leave at baseline in predictor models of future sick leave.
A limitation of the present study was the somewhat low response rate (41.5%) which can be considered a threat to the generalisability of the findings. To some extent, this was compensated for by the fact that repeated measurements from each individual throughout the study year were included. The sample comprised mostly women (87.7%), all being educated nurses or nurses' assistants, thus limiting the possibilities for generalising to men and other professions without reservations. The present study was also partly based on self-report, though, mostly through standardised well-validated scales assessing individual features that are subjective, still presumably stable, due to their very nature (eg, personality). One exception was self-reported shift work experience, which may be affected by recall bias. Reliability analysis revealed somewhat low internal consistency for the Diurnal Scale (morningness), which is likely due to the low number of items in this scale. Another point to take into consideration with the present study is that we decided to investigate exposure to shift characteristics and sick leave with time brackets of calendar months. Although this was performed after a careful evaluation of what would be the most appropriate approach, we cannot know whether looking at shorter or longer time brackets—which would increase the sensitivity to more acute or cumulative effects, respectively—would reveal stronger and perhaps also more effects of shift characteristics on sick leave.
Strengths of the present study comprised the use of objective records of shift work and sick leave and the prospective design, which enabled conclusions about directionality that is not influenced by recall or measurement bias. Accurate records allowed us to differentiate between specific shift schedule characteristics and to investigate their effects on sick leave from the first day of absence—which is a rare opportunity in sick leave research. Previous studies have thus mainly focused on long-term sick leave as outcome measure, which may be caused by other factors (eg, chronic disease) than short-term sick leave. Accordingly, as a matter of course, and to a greater extent, shift workers will report proximate issues such as sleep deprivation, fatigue and sleepiness,6 which are more likely to be mirrored in short-term sick leave. Another unique feature of the present study was that the objective records were combined with questionnaire data, which allowed us to examine the role of personality in shift work tolerance using high quality data. Essential for accurate predictor models of sick leave is proper control of confounders with potential explanatory value.9 Such confounders include sex, marital status, parental responsibility, sick leave at baseline and shift work experience; all of which were controlled for in the present study. To the best of our knowledge this is the first study that has prospectively investigated the effects of different shift-work characteristics on sick leave, both based on day-to-day registry data.
In conclusion, our results showed that quick returns, implying restricted recovery opportunities between shift intervals, increased the risk for future sick leave, whereas night and evening shifts (primarily in rotating schedules) were unrelated to future sick leave. Personality characteristics associated with shift-work tolerance (low on morningness, high on flexibility, low on languidity) had no main effects and no moderating effects on the association between quick returns and sick leave. Quick returns are usually not a necessity in order to maintain staffing, and should therefore be considered for elimination in shift schedules in an effort to reduce workers' sick leave and the associated economic costs.
Acknowledgments
The authors thank employees at the personnel department at Haukeland University Hospital for their assistance with retrieving the payroll and sick leave data.
References
Footnotes
Contributors ØV, AH, SP, BB, SW, ES and BS took part in planning the study. ØV, AH, ES and SW took part in gathering the data. ØV, AH, SP, BB, SW, ES, BS and EE took part in writing up the manuscript and reporting the work.
Funding This project was funded by a PhD grant at the Department of Psychosocial Science, Faculty of Psychology, University of Bergen, Norway.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Regional Committee for Medical and Health Research Ethics, Northern-Norway (number 2013/526/REK nor) and by the Norwegian Data Inspectorate (13/00569-2/CGN).
Provenance and peer review Not commissioned; externally peer reviewed.