Self-Rated Health (SRH), Recovery From Work, Fatigue, and Insomnia Among Commercial Pilots in Relation to Occupational and Non-Occupational Factors

Background: This study investigates associations between self-rated health (SRH), recovery from work, fatigue, and insomnia and work conditions, psychosocial work environment, lifestyle and sense of coherence (SOC) among commercial pilots. Methods: A standardized questionnaire was sent to all pilots in an airline company, 354 (61%) participated. Associations were analysed by ordinal and logistic regression with mutual adjustment. Results: Totally 21.8% reported low SRH, 13.0% did not recover from work during free time, 61.9% had fatigue and 70.6% had insomnia symptoms. SRH was associated with high work demands, SOC, exercise, and overweight/obesity. Recovery from work was associated with part-time work, being a co-pilot, low social support at work and SOC. Fatigue was associated with female gender, high work demands, SOC, and younger age. Insomnia was associated with low social support, SOC, and free hours after work. Moreover, insomnia was associated with ying MD80 and intercontinental ights with Airbus 330/340, using Boeing 737 as reference. Conclusion: The psychosocial environment at work is important for health among pilots and a high sense of coherence can be protective. Occupational conditions may have an inuence on recovery from work, fatigue, and insomnia. Moreover exercise, being t, working part-time and having more free hours after work could improve health.

Self-rated health (SRH), also called self-reported health or perceived health, is a widely used indicator of health (De Bruin 1996). SRH, measured by a single question, has been proved to be a reliable predictor of mortality (Taloyan, Leineweber et al. 2015) and the development of chronic diseases (Latham and Peek 2013). SRH in the general population is in uenced by social differences between countries , social-economic status (Alvarez-Galvez, Rodero-Cosano et al. 2013) and occupational factors (Niedhammer, Chastang et al. 2008). We found no previous study on risk factors for SRH among commercial pilots.
The concept of sense of coherence (SOC) was raised by Antonovsky. A high SOC is a personality trait re ecting the healthpromoting capability to cope with stress (Antonovsky 1979, Antonovsky 1987. SOC is constructed with comprehensibility, manageability, and meaningfulness, and has been demonstrated to predict various aspects of health (Flensborg-Madsen, Ventegodt et al. 2005). The SOC scale has been used in occupational studies (Lundberg and Peck 1995 The psychosocial work environment can in uence health and is usually studied by using the demands-control-support model (Del Pozo-Antunez, Ariza-Montes et al. 2018, Lecca, Campagna et al. 2018). The work condition with high demands, low control, and low social support is the most harmful (Johnson 1986, Johnson andHall 1988). Among commercial pilots, low social support has been reported to be associated with sleep problems (Runeson, Lindgren et al. 2011).
Recovery is an essential psychological process for detachment from work, and preparation for new work challenges. Recovery from work can be affected by psychological demands at work, sleep quality, leisure style (Zijlstra and Sonnentag 2006), and vacation (Bloom, Kompier et al. 2009, de Bloom, Geurts et al. 2010, de Bloom, Geurts et al. 2013). We found few studies on associations between psychosocial work conditions among commercial pilots and health (Lindgren, Andersson et al. 2002, Runeson, Lindgren et al. 2011, and no previous study on recovery from work in this occupational group. When investigating health risks among pilots it is important to have a holistic perspective, including occupational as well as non-occupational risk factors. The aim is to study associations between occupational and non-occupational factors and SRH, recovery from work, fatigue and insomnia among commercial pilots. Our hypothesis in this study is that these four health variables can be in uenced by work conditions (type of aircraft, type of ights, the psychosocial work environment) as well as SOC and socio-economic and life-style factors.

Methods
This study is a part of a project on work conditions and self-reported health among commercial pilots. A self-administered questionnaire was sent to all Stockholm-based pilots (captains and co-pilots) on duty in a Scandinavian airline company (N = 585), 61% participated (N = 354). The study protocol was approved by the Regional Ethical Board in Uppsala, Sweden, and all participants gave their informed consent. All methods were carried out in accordance with relevant guidelines and regulations. Details about the questionnaire have been previously published (Runeson, Lindgren et al. 2011). The questionnaire included questions on demographic factors, work conditions, psychosocial work environment, lifestyle (like exercise habit, free hour after work), home environment, and SOC. Body mass index (BMI) was de ned as the body mass divided by the square of body length. BMI variable was categorized as: underweight (< 18.5), normal (18.5-24.99), overweight (25.00-29.99), and obese ( > = 30.00). The psychosocial model has been described before (Runeson, Lindgren et al. 2011), which included work demand, work control, and social support. These questions were validated by Theorell (Theorell, Michelsen H et al. 1993). A short version of SOC was adapted from Lundberg and Nyström (Lundberg and Peck 1995). There was one question for each of three SOC dimensions: a) Manageability: do you usually see a solution to problems and di culties that other people nd hopeless? b) Meaningfulness: do you usually feel that your daily life is a source of personal satisfaction? c) Comprehensibility: do you usually feel that the things that happen to you in your daily life are hard to understand? Details about categories the exposure variables are listed in Table 1. A total of 78.2% of the pilots reported good or excellent SRH, and 64.4% of the pilots reported quite or very often feeling recovered after several days off work. Fatigue (61.9%) and insomnia (70.6%) were commonly reported ( Table 2). The ranges of scores for high work demands, low work control and low social support at work were 1-15, 3-18, and 0-42, and the interquartile ranges are 6-10, 8-11, and 10-19 respectively. The prevalence of SOC dimensions are displayed in Table 3. The range of total SOC scores was 1-9, and the interquartile range was 5-6.
Age Saab 2000 7.9 a. Body mass index (BMI) was de ned as the body mass divided by the square of body length. BMI variable was categorized as: underweight (< 18.5), normal (18.5-24.99), overweight ( > = 25.00), and obese ( > = 30.00). Overweight and obese were merged into one group for the statistical model.
b. The prevalence of work year is presented in three ranges, but in the statistical models, it was treated as a continuous variable.

Assessment of the four dependent variables
There was one question assessing SRH: "In general, how would you like to describe your health?"(Idler and Angel 1990). There were four options: "excellent", "very good", "fair", and "poor". The question on recovery from work was adapted from Gustafsson: "Do you feel rested and recovered when you start working again after a couple of days off?" (Gustafsson, Lindfors et al. 2008). There were ve options: "very often", "quite often", "sometimes", "seldom", and "never". One question asked about fatigue during work or leisure time. Three questions on sleep disturbances were adapted from a previous sleep questionnaire by Åkerstedt (Akerstedt 1984). These sleep questions asked about the di culty to fall asleep, repeated awakenings during sleep, and too early nal awakening, with a three-month recall period. There were four options for each question: "most of the time", "sometimes", "seldom" or "never". Insomnia was de ned as reporting at least one of the three symptoms most of the time or sometimes. Both fatigue and insomnia were dichotomous variables.

Statistical methods
Multiple logistic regression was used to analyse the association for fatigue and insomnia (yes/no variables). Ordinal regression was used to analyse associations for SRH and recovery from work. For all ordinal regression models, tests of parallel lines were done, to verify that ordinal regression could be used. Initially, health associations were analysed in models with one exposure factors, adjusting for age and gender (single-factor models). As a next step, mutual regression models for occupational factors were constructed, including factors with a p-value of less than 0.1 from the single factor analysis, with age and gender adjustment. The occupational mutual model was then further adjusted for SOC. The nally, associations from the occupation/non-occupational models were selected for the nal mutual analysis (p < 0.1 as inclusion criteria). Pearson's correlation coe cients were calculated for the independent variables. The correlation between age and year of employment was above 0.7. Therefore, only age was included in the nal mutually adjusted models. The psychosocial factors were all included if any of them had a p < 0.1. For the logistic and ordinal regression models, odds ratios (OR) with a 95% con dence interval (CI) were calculated. A p-value < 0.05 was considered statistically signi cant.
Calculations were done by IBM SPSS Statistics 21. a. The scales of the health variables were listed above. Some of the groups were merged because of small numbers in the group. The categories of these health variables in the statistic models are as following: SRH-poor or fair, good, excellent; Recovery-never or seldom, sometimes, quite often, very often; Fatigue-never or seldom, sometimes or often.
b. Insomnia represents either of the following symptoms were reported sometimes or most of the time during last 3 months: di culty to sleep, repeated awakenings with di culty falling back to sleep, and too early nal awakening.  (1,5,6,9) a. For SOC: There were ve scale level for each dimension: never, seldom, sometimes, quite often, and very often. Some groups with a small number were merged to one, including "seldom or never" for manageability and meaningfulness, and "very or quite often" for comprehensibility. Table 4 shows associations between the four dependent variables and single occupational and non-occupational factors, adjusted for age and gender. Part-time pilots had better recovery (p = 0.021). Pilots operating Saab 2000 (p = 0.041) had less fatigue and those who operating MD 80 series (p = 0.012) and Airbus 330/340 (p = 0.007) had more insomnia, compared to those who operated Boeing 737 (reference). Higher demands and lower social support at work were associated with all the four dependent variables. Lower control was associated with less recovery from work. Total SOC was associated with higher SRH (p < 0.001), better recovery from work (p < 0.001), less fatigue (p < 0.001), and less insomnia (p < 0.001). Those with a higher sense of manageability, higher meaningfulness, or higher comprehensibility reported higher SRH, better recovery from work, less fatigue, and less insomnia, respectively. Overweight or obese pilots reported poorer SRH (p < 0.001), and more fatigue (p = 0.026). Ex-smokers (p = 0.003) and current snuff users (p = 0.017) reported poorer SRH as compared to non-smokers and non-snuff users, respectively. The pilots with a weekend couple relationship had lower SRH (p = 0.037), and less recovery from work (p = 0.021). Those with pre-school children at home (0-6 y) had less recovery from work (p = 0.009). Pilots doing frequent exercise had higher SRH (p < 0.001) and better recovery from work (p = 0.009). Those with more free hours after work had higher SRH (p = 0.040), better recovery from work (p < 0.001), and less insomnia (p = 0.005). The pilots who slept less than 5 h per night (short sleepers) had lower SRH (p = 0.042) and less recovery from work (p = 0.014) (  b. The associations between dependent variables and years of employment were only adjusted by gender, because of the strong correlation between years of employment and age.
c. The associations between dependent variables and the SOCtot and psychosocial variables were calculated by their interquartile range.
d. Some people are in a stable relationship, and they live together with their partner. They are de ned as couple. Some people are in a relationship, but they live separately from their partner, and they usually meet on weekends. They are de ned as weekend couple.
e. NA: Not available. Table 5 presents the results from the nal mutually adjusted models. The psychosocial work environment was associated with all health outcomes. High demands reduced SRH (p = 0.044) and increased fatigue (p < 0.001). Low social support reduced recovery from work (p < 0.001), and increased insomnia (p = 0.006). The type of aircraft was related to fatigue (p = 0.032) and insomnia (p = 0.003). Pilots operating MD80-series aircraft (OR = 2.27, 95%CI = 1.21-4.29) and Airbus330/340 (OR = 4.16, 95%CI = 1.98-8.74) had insomnia, using Boeing737 as reference. Co-pilots reported less recovery from work than captains (p = 0.008) and older pilots had less fatigue (p = 0.005). Part-time work increased recovery from work (p = 0.001). Overweight/obese pilots had lower SRH (p < 0.001) and pilots with doing frequent exercise had higher SRH (p = 0.004). More free hours after work reduced insomnia (p = 0.021). b. The associations between dependent variables and years of employment were adjusted by gender, BMI, smoking habit, and oral tobacco (snuff) use except age, because of the strong correlation between years of employment and age.
c. The associations between dependent variables and the SOCtot and psychosocial variables were calculated by their interquartile range.
d. Some people are in a stable relationship, and they live together with their partner. They are de ned as couple. Some people are in a relationship, but they live separately from their partner, and they usually meet on weekends. They are de ned as weekend couple.
NA: Not available

Discussion
The psychosocial environment at work can in uence self-rated health, recovery from work, fatigue and insomnia and the operating type of airplanes can in uence insomnia. A high sense of coherence, avoiding overweight or obesity, regular physical exercise and working part-time were found to be health-promoting factors for commercial pilots.
Our study has some strengths. It is a unique study on risk factors and health-promoting factors for self-rated health, recovery from work, fatigue, and insomnia among commercial pilots. The study had a reasonable response rate (61%), and there were no differences in age or gender between participants and non-participants (Runeson, Lindgren et al. 2011).
Thus, selection bias after employment is not likely to have any major in uence on our results. However, since pilots are initially selected to be healthy, they are not comparable with the general adult population. Another strength is that we used statistical models with mutual adjustment for different occupational and non-occupational factors.
One limitation of the study is that we only used self-reported data, which could create information bias. However, we found associations between speci c dependent variables and speci c risk factors, rather than a general increase of many associations with the same order of magnitude. Thus, information bias is not likely. Another limitation is that we only included pilots from one airline, which limits the external validity of the study. Moreover, the cross-sectional study design limits the possibility to draw conclusions on causality.
We will limit our discussion to factors signi cant in the mutually adjusted models. Sense of coherence (SOC) and the psychosocial work environment were the most important factors. SOC was associated with all four health variables in single factor analysis as well as in mutually adjusted models. We found no previous study on SOC among commercial pilots. SOC represents a salutogenic coping ability and is considered to be relatively stable in adulthood ( In the single-factor analysis, the psychosocial work environment was related to all four health outcomes. In the mutually adjusted models, high work demands were associated with poor SRH and fatigue, and low social support at work was associated with poor recovery from work and insomnia. Our question on recovery from work was adapted from a previous publication among government employees (Gustafsson, Lindfors et al. 2008). We found that one third (35.6%) of the pilots did not feel recovered after several days off work. Copilots had less recovery from work than ight captains and those working part-time reported better recovery from work. The main differences between captains and co-pilots are rank and salary. The captains take responsibility for the ight, and the co-pilots takes the order from the captain. To our knowledge, our study is the rst on recovery from work among commercial pilots.
A majority (61.9%) of the pilots reported fatigue (sometimes or often). In our previous study, using a different question on fatigue,.29.9% of pilots reported weekly fatigue, and 82.8% of pilots reported any fatigue during the last three months (Fu, Lindgren et al. 2015) Female pilots had more fatigue which is in agreement with a previous study. A similar gender difference has been reported previously (Hughes and Kumari 2018). Moreover, older pilots had less fatigue. This is opposite results as reported in a large population study from the UK reporting that fatigue increases with age (Hughes and Kumari 2018). However, older pilots are more experienced and have a less tight duty schedule.
More than two-thirds of the pilots (70.6%) reported insomnia. Besides SOC and the psychosocial work environment, overweight or obesity and increased insomnia and free hours after work reduced insomnia. Moreover, type of operated aircraft in uenced the prevalence of insomnia. Airbus 330/340 was the only aircraft type used for intercontinental ights.
As expected, pilots on these intercontinental ights had more insomnia, as compared to Boeing 737, used for ights within Europe (reference group). Moreover, pilots who operated MD80s had more insomnia, as compared to those operating Boeing 737. Boeing 737 and MD80 are both narrow body aircraft operated for short/medium range. This difference could be caused by some other reason except ight duration.

Conclusion
In conclusion, for commercial pilots the psychosocial environment is important for a good health and a high sense of coherence can be an important health-promoting personality factor for self-rated health, recovery from work, fatigue, and insomnia. Avoiding overweigh or obesity and regular physical exercise are important health promoting life style factors.
Operating intercontinental ights can increase insomnia and working part-time can increase recovery from work. Having more free time after work can reduce insomnia. The airline companies should put more effort to improve the work environment for their pilots and the pilots should contribute by keeping a healthy lifestyle.
Abbreviations SRH-self-rated health SOC-sense of coherence Declarations Ethics approval and Consent to participate The study protocol was approved by the Regional Ethical Board in Uppsala, Sweden. All participants gave their consent to participate.

Consent for publication
All participants gave their consent for publication.

Availability of data and material
All the data included in this article is available by contacting the authors.

Competing interests
The authors declare that they have no competing interests. Funding