Psychological distress in Spanish airline pilots during the aviation crisis caused by the COVID-19 pandemic and psychometric analysis of the 12-item general health questionnaire

Abstract

The aim of this study was to evaluate the psychological distress of Spanish airline pilots, a group of professionals undergoing an unprecedented work situation as a result of the Covid-19 pandemic. To do so, we administered the General Health Questionnaire-12 (GHQ-12). A total of 342 questionnaires were obtained, with the largest respondent age group being 41–50 years. The psychometric properties of the instrument were also evaluated, with an exploratory factor analysis revealing a unidimensional structure that explained 59.23% of the variance. The total score on the standard GHQ-12 was 4.54 ± 3.31 very close to the cut-off point established to determine psychological distress. The score increased among those unemployed and was also higher among pilots on furlough compared to those whose work situation was relatively normal. Furthermore, the reliability measured by the total Cronbach's alpha was above 0.8 of each across all employment status considered. These results show the desirability of conducting periodic psychological distress assessments of pilots so that effective measures can be implemented to ensure their psychological and socio-emotional well-being.

Practitioner summary: This article evaluates psychological distress in a group of professionals that has received scant attention in the field. Moreover, it does so against the background of an atypical situation, evaluating the psychological distress suffered by pilots in different employment categories during the current severe crisis in the aviation sector.

Abbreviations: CFI: comparative fit index; COPAC: Official College of Commercial Aviation Pilots; ECA: European Cockpit Association; EFA: exploratory factor analysis; GHQ-12: general health questionnaire-12; IATA: International Air Transport Association; KMO: Kaiser-Meyer-Olkin test; PA: optimal implementation of parallel analysis; ULS: unweighted least squares

Keywords:

• Psychological distress
• Spanish aviation pilots
• GHQ-12 questionnaire
• factor analysis

Introduction

As a result of the health crisis caused by COVID-19, governments in many countries have been forced to restrict both international and national mobility, triggering an unprecedented crisis in the aviation sector, with more than 70% of the world’s commercial aircraft having been grounded in April 2020 compared to April 2019 (Curley et al. 2020). In Spain, the number of flights has been reduced by 72% compared to the previous year (Eurocontrol 2020). A study published by the European Cockpit Association (European Cockpit Association 2021) on a large number of European Airlines calculates that more than 18,000 pilot jobs have been lost or are on the brink of disappearance.

The International Air Transport Association (IATA 2020) forecasts that passenger airlines will lose around USD 84 billion in 2020. These losses may be even worse depending on the length of the restrictions and the emergence of new outbreaks (Albers & Rundshagen, 2020). In addition, according to the forecasting model designed by Iacus et al. (2020), at the end of 2020, in the worst case scenario, the loss could be as high as 1.41−1.67% of world GDP, resulting in the loss of between 2.2 and 30.3 million jobs, which could have a significant impact on the mental health of the sector's professionals.

Classic studies, such as those by Warr (1987) and Jahoda (1981), have provided theoretical explanations of the reasons why unemployment has a negative impact on psychological well-being. Warr (1987) suggests that unemployment generates negative psychological outcomes because individuals that lose their job do not experience the nine benefits he associates with work. Similarly, Jahoda (1981) concludes that the deterioration of mental health found in unemployed individuals is because they do not benefit from certain latent functions of employment. A set of variables that moderate the psychological impact of unemployment has been identified (Paul and Moser 2009), which include length of unemployment and country of residence. Furthermore, job insecurity correlates negatively with mental health well-being. As suggested by Russo and Terraneo (2020), this is because the worker experiences different types of insecurity, namely cognitive, affective and labour market insecurity. Furthermore, workers experience a subjective loss of control as they are uncertain about whether they will lose their job, resulting in significantly impaired mental health (Vander Elst et al. 2016; Witte, Pienaar, and Cuyper 2016).

Both isolation and uncertainty in the work environment can increase levels of stress and psychological distress. A number of studies have examined the differences between employed and unemployed individuals during the COVID-19 pandemic. Ruffolo et al. (2021), in their study on participants in Norway, UK, USA and Australia, reported that individuals with a job showed lower levels of psychological distress than those unemployed, and also presented higher levels of psychosocial well-being and quality of life, and lower levels of overall, social and emotional loneliness.

The health crisis has had a negative impact on people’s mental health, not only due to unemployment but also as a result of other factors associated with confinement. An example of work in this field is that by Pérez, Masegoso, and Hernández-Espeso (2021), who, in a sample of 1781 Spanish adults, evaluated psychological distress, depression, anxiety and somatisation symptoms, along with other sociodemographic variables, finding that between 25% and 39% of the respondents presented clinically significant distress.

There is increasing scientific evidence on the association between psychological distress and a higher mortality risk. For example, in a longitudinal study with an 8.2-year follow-up on 10 cohorts from the Health Survey for England, with 68,222 participants aged over 35 years, free of cardiovascular disease and cancer as study baseline, a dose-response association was found between psychological distress (measured on the GHQ-12) and the risk of mortality. It was observed that even low levels of psychological distress (except in the case of cancer death, which was only associated with higher levels), increased the risk of mortality, despite adjustment for confounding factors, such as lifestyle, age and sex (Russ et al. 2012).

Aviation pilots are responsible for managing air operations in different areas (transport of passengers, aerial work or air freight). The profession involves coping with heavy workloads and dealing with situations that generate high levels of stress. Consequently, pilots may suffer problems that have a negative impact on their psychological well-being. This has been evidenced in a number of studies. For example, Wu et al. (2016) found that 12% of pilots reported suffering periods of depression and 4% reported having suicidal thoughts. Additionally, Demerouti et al. (2019) found that 40% of commercial airline pilots had very high burnout symptoms. However, the crisis caused by the COVID-19 pandemic has placed pilots in a novel situation of job insecurity.

Aim

The aim of this study was to assess the psychological distress of Spanish airline pilots according to their job situation following the first repercussions of COVID-19.

Materials and methods

Instruments

The Spanish Association of Commercial Aviation Pilots (COPAC, in its Spanish acronym) designed an on-line survey that was sent to all its members. The employment data collected referred to age, current employment status (unemployed, furloughed or working) and the type of aviation activity for which they were, or had been, hired (commercial transport of passengers, aerial work, air freight, executive aviation, teaching or other activities not included in the previous categories).

We used the General Health Questionnaire (GHQ-12), which is a self-administered screening questionnaire designed to assess psychological well-being and detect possible cases of psychiatric distress in general population (Goldberg and Williams 1988). Thanks to its brevity, simplicity and reliability, the GHQ-12 is one of the most widely used tests, and is regarded as one of the most appropriate for use in mental health surveys (Rocha et al. 2011). The GHQ-12 comprises 12 items, 6 positive and 6 negative. Each item has four possible responses, which are scored on a Likert-type scale (0-1-2-3). These can subsequently be converted into a dichotomous score (0-0-1-1) (‘better’ and ‘usual’ are scored as 0; while ‘worse’ and ‘much worse’ responses as 1), known as the standard GHQ score. The total score can range between 0 and 12 points, where the higher the score, the greater is the psychological distress. As there are no established cut-off points for the GHQ-12, we used 5 as the cut-off point, corresponding to the 90th percentile of the standard GHQ score for a representative Spanish sample of individuals under the age of 65. Above this percentile is considered greater psychological distress (Rocha et al. 2011). We also calculated the Likert score for the total sample and independently for each employment status, in order to compare the data with other publications in which only this type of score was calculated.

Procedure

The participants were told that the aim of the questionnaire was to determine the emotional impact on pilots six months after the onset of the COVID-19 pandemic, and to propose, if necessary, measures to improve pilots’ emotional well-being to enhance operational safety. An email was sent to 500 members of COPAC on three occasions in October 2020, with a link from which they could access the survey. All the participants voluntarily and anonymously completed the questionnaire, after having been informed of the aims of the study and giving their consent to participate. They were also told that the confidentiality and processing of the data were safeguarded under Organic Law 3/2018 on Data Protection and Guarantee of Digital Rights (6 December 2018). Only one questionnaire could be completed from any device.

Participants

The GHQ-12 questionnaire was sent to a randomised sample of 500 pilots, of which 342 participants responded, representing a response rate of 68.4%, and the 95% of those who returned questionnaires were male. Age was distributed as follows: aged below 30 years (5.6%), between 31 and 40 (19%), between 41 and 50 (45.3%), between 51 and 61 (28.1%) and over 61 years (2%). Of the sample, 81.6% worked in the commercial transport of passengers, 5.8% in aerial work, 5.3% in executive aviation, 2.6% in teaching, 2.3% in air freight, while 2% worked in other types of aviation activity. As regards their employment status, 9.3% were unemployed, 66.1% were on furlough and 24.6% were still working.

Data distribution and comparision

To determine the distribution of the sociodemographic variables and the total standard GHQ score, the Kolmogorov-Smirnov test was used. It was found that the variables did not meet the assumption of normality. Accordingly, to compare the scores' central tendency parameters between individuals aged under and over 41 years and between groups according to employment status, we used the non-parametric Mann–Whitney U test. The chi-square (χ²) test was used to compare frequencies of qualitative variables (expressed as proportions from the total) between two or more groups depending on age and employment status and for the score on each item on the questionnaire.

Questionnaire reliability

To evaluate the reliability of the questionnaire administered in this particular sample of Spanish aviation pilots, the internal consistency of the GHQ-12 questionnaire was measured using Cronbach's alpha, which was also calculated for each of the deleted items. We also calculated the correlation of each item with the total of the 12 items on the instrument.

Psychometric properties evaluation

To assess the psychometric properties of the GHQ-12 questionnaire administered we performed an exploratory factor analysis (EFA), using the FACTOR program developed by the Rovira i Virgili University (Ferrando and Lorenzo-Seva 2017). This software applied the Kaiser–Meyer–Olkin (KMO) test to measure the sample adequacy and performed Bartlett’s test of sphericity to check that the factor model had an adequate fit. To determine the number of dimensions of the instrument, we applied the Hull Method to select the number of common factors (Lorenzo-Seva, Timmerman, and Kiers 2011) and the Optimal implementation of Parallel Analysis (PA) (Timmerman and Lorenzo-Seva 2011; Lorenzo-Seva, Timmerman, and Kiers 2011), regarded as one of the best methods to determine the number of factors to retain. The factors were extracted using the Unweighted Least Squares (ULS) method. To recognise the ordinal nature of the items, we used the polychoric correlation matrix (Gadermann, Guhn, and Zumbo 2012). We also calculated the Comparative fit index (CFI), which measures the fit of a model even when the sample is small. A score of ≥ 0.90 is considered an acceptable cut-off point (Hu and Bentler 1999).

The descriptive analyses and the comparison of proportions were performed using IBM SPSS statistics v.20, the EFA was conducted using FACTOR and the figures were produced using SPSS and Microsoft Excel.

Results

Sociodemographic variables and psychological distress

The mean total GHQ-12 score was 4.54 ± 3.31, representing absence of psychopathology, although the value is very close to the cut-off point (following Rocha et al. 2011). The total proportion of individuals presenting psychological distress risk was 43.3%.

Table 1 shows the means of the standard GHQ-12 scores for the complete sample, and by age group and employment status. It can be seen that the lower the age group, the higher is the mean GHQ-12 score, with the highest values being in the under 30 age group (5.98 ± 3.77), and in the 31-40 group. The two age groups comprising 41 to 60 years present similar scores. Despite the higher scores among the younger respondents, no significant differences were found between age groups for the GHQ-12 score and psychological distress (Chi squared: 8.77; p = 0.067). Nonetheless, fragmenting the sample between those aged below and above 41 years, the psychological distress tended towards significance (Chi squared: 5.99; p = 0.010).

Table 1. Characteristics of the sample by age and employment status, and mean scores and standard deviation for standard GHQ score, for the complete sample and for each age group and employment status.

Variable	Category	Total (N)	(%)	Standard GHQ score (mean ± SD)
Age group	≤30	19	5.5	5.98 ± 3.77
	31–40	65	19.0	5.15 ± 3.01
	41–50	155	45.2	4.17 ± 3.21
	51–60	96	28.0	4.42 ± 3.45
	≥61	7	2.0	4.86 ± 3.94
Employment situation	Unemployed	32	9.4	6.00 ± 3.61
	On furlough	226	66,1	4.85 ± 3.34
	Working normally	84	24.6	3.13 ± 2.60
Total	Total	342	100	4.54 ± 3.31

The total score is associated with employment status, being higher when the status is worse. As can be seen in Table 1, the mean standard GHQ-12 score for the unemployed participants was 6.00 ± 3.61, corresponding to psychological distress. The scores for those on furlough or working normally corresponded to absence of psychopathology, although the participants on furlough, and thus in a more vulnerable employment situation, presented higher scores. Significant differences were found in the participants showing psychological distress among the three employment situations being higher in the unemployed group, followed by those on furlough and lower among those working (62.5%, 47.3% and 25.0% respectively, chi-square: 17.77; p ≤ 0.001). As mentioned in the methodology section, our aim was to obtain the total Likert score (Mean ± SD) for the complete sample (14.60 ± 6.09) and according to employment status: unemployed (17.22 ± 6.87), on furlough (15.15 ± 6.18), and working normally (12.13 ± 4.64), to enable comparisons with other studies that only report this type of score.

As can be seen in Figure 1, the mean GHQ-12 score was higher among the unemployed pilots compared to their counterparts on furlough or working normally. Significant differences were found (p ≤ 0.001) between the unemployed pilots and those working normally, and also between employed and those on furlough (p ≤ 0.001). We found no significant differences on the total score (p = 0.085), but we did find significant differences between the unemployed pilots and their furloughed counterparts.

Figure 1. Total GHQ-12 score according to employment status.

The highest General Health Questionnaire-12 score (6 points) corresponds to unemployed pilots. It is intermediate for those on furlough (4.8 points) and lower (3 points) for those who work normally.

Figure 2 shows the percentages of individuals that scored high on each of the questionnaire items according to employment status. It is worth noting the high percentage (84.4%) of unemployed pilots that gave a high score on Item 5 (Have you felt constantly under strain?) compared to 47.6% of those working normally. We also found high percentages of unemployed respondents compared to those that had kept their jobs for Item 3 (Have you felt you were playing a useful part in things?), Item 6 (Have you felt you could not overcome your difficulties?), Item 9 (Have you been feeling unhappy and depressed?), in this case, the percentages of unemployed and furloughed pilots are similar, Item 10 (Have you been losing confidence in yourself?), and Item 12 (Have you been feeling reasonably happy, all things considered?).

Figure 2. Percentages on each item in the GHQ-12 according to the employment status of the pilots in the sample.

A histogram compares one by one the answers to the 12 questions of the General Health Questionnaire as a function of employment status. The largest differences corresponded to items 3, 4, 5, 6, 8, 10, 11, and 12. In all of them, the highest percentages of favourable (or healthy) responses corresponded to pilots working. The lowest percentage of favourable responses always corresponded to unemployed subjects.

Table 2 shows the significant differences between pairs for each item according to their employment status. We found significant differences on some of the items between individuals working normally and those on furlough or unemployed. No significant differences were found, however, between unemployed and furloughed pilots for any of the items. The items that showed significant differences between individuals working and those on furlough or unemployed were numbers 3, 5 and 6. The items that yielded significant differences between unemployed pilots and their working counterparts were numbers 10 and 12. A significant difference was found between those working and those on furlough on item 9, while the difference between unemployed and working pilots on this item was close to significance. Although not shown in Table 2, significant differences were found across the three employment status groups for items 3, 5, 9 and 10.

Table 2. Comparison of percentages between different pairs of groups according to employment status (Unemployed/Working; Unemployed/Furloughed; Working/Furloughed) for each item on the questionnaire, referring to the proportion of individuals that obtained 1 point for each item.

Items	Unemployed	Working	p-value	Unemployed	On furlough	p-value	Working	On furlough	p-value
	(%)	(%)		(%)	(%)		(%)	(%)
Have you been able to concentrate on what you were doing? GHQ1	56.2	44.0	NS	56.2	56.6	NS	44.0	56.6	NS
			p = 0.166			p = 0.557			p = 0.032
Have you lost much sleep over worry? GHQ2	62.5	60.7	NS	62.5	64.2	NS	60.7	64.2	NS
			p = 0.518			p = 0.5			p = 0.333
Have you felt you were playing a useful part in things? GHQ3	46.9	8.3	p ≤ 0.001	46.9	34.5	NS	8.3	34.5	p ≤ 0.001
						p = 0.122
Have you felt capable of making decisions about things? GHQ4	34.4	10.7	NS	34.4	21.7	NS	10.7	21.7	NS
			p = 0.004			p = 0.089			p = 0.018
Have you felt constantly under strain? GHQ5	84.4	47.6	p ≤ 0.001	84.4	68.1	NS	47.6	68.1	p ≤ 0.001
						p = 0.042
Have you felt you could not overcome your difficulties? GHQ6	43.8	9.5	p ≤ 0.001	43.8	28.8	NS	9.5	28.8	p ≤ 0.001
						p = 0.067
Have you been able to enjoy normal day to day activities? GHQ7	59.4	57.1	NS	59.4	53.5	NS	57.1	53.5	NS
			p = 0.499			p = 0.335			p = 0.331
Have you been able to face up to your problems? GHQ8	34.4	13.1	NS	34.4	27.4	NS	13.1	27.4	NS
			p = 0.011			p = 0.268			p = 0.005
Have you been feeling unhappy and depressed? GHQ9	62.5	31.0	NS	62.5	62.4	NS	31.0	62.4	p ≤ 0.001
			p = 0.002			p = 0.577
Have you been losing confidence in yourself? GHQ10	40.6	7.1	p ≤ 0.001	40.6	20.4	NS	7.1	20.4	NS
						p = 0.013			p = 0.003
Have you been thinking of yourself as a worthless person?GHQ11	18.8	1.2	NS	18.8	8.4	NS	1.2	8.4	NS p = 0.013
			p = 0.002			p = 0.070
Have you been feeling reasonably happy, all things considered?GHQ12	56.2	22.6	p ≤ 0.001	56.2	38.9	NS	22.6	38.9	NS
						p = 0.048			p = 0.005
Total	62.5	25.0	p ≤ 0.001	62.5	47.3	NS	25.0	47.3	p ≤ 0.001
						p = 0.078

The significance values are also shown in bold**. ^**p ≤ 0.001.

NS: Non-significant.

Internal consistency and factor structure analysis

The scale shows good alpha measures of internal consistency or reliability analysis. The findings on reliability allow us to affirm that the GHQ-12 scores were adequate in their internal consistency as measured by Cronbach's ordinal alpha (Table 3).

Table 3. Reliability analysis. Cronbach’s alpha if each item is deleted, item-total correlation across the 12 items on the GHQ-12 questionnaire and Cronbach’s alpha for the sample and subgroups.

Sample	Total		Unemployed		On furlough		Working
Item	Cronbach alpha if item is deleted	Corrected item-total correlation	Cronbach alpha if item is deleted	Corrected item-total correlation	Cronbach alpha if item is deleted	Corrected item-total correlation	Cronbach alpha if item is deleted	Corrected item-total correlation
GHQ1	0.845	0.444	0.861	0.360	0.845	0.448	0.779	0.449
GHQ2	0.846	0.426	0.861	0.356	0.846	0.425	0.766	0.548
GHQ3	0.843	0.460	0.855	0.456	0.847	0.415	0.779	0.452
GHQ4	0.836	0.567	0.846	0.582	0.835	0.575	0.780	0.423
GHQ5	0.836	0.568	0.853	0.480	0.836	0.557	0.764	0.566
GHQ6	0.827	0.686	0.835	0.737	0.827	0.687	0.774	0.515
GHQ7	0.845	0.448	0.846	0.574	0.840	0.502	0.794	0.325
GHQ8	0.833	0.613	0.840	0.670	0.833	0.603	0.771	0.533
GHQ9	0.835	0.576	0.839	0.674	0.837	0.547	0.769	0.520
GHQ10	0.837	0.560	0.848	0.552	0.836	0.567	0.784	0.386
GHQ11	0.849	0.373	0.862	0.308	0.848	0.374	0.794	0.297
GHQ12	0.835	0.574	0.838	0.693	0.835	0.573	0.783	0.391
Total Cronbach’s alpha	0.851		0.860		0.850		0.793

Our data show the internal consistency is high, with a value of 0.851 corresponding to the standard GHQ score. The internal consistency of the questionnaire did not increase when any of the 12 items were deleted. Table 3 shows the Alpha values for the total sample and for each of the items, and by subgroup according to the participants’ employment status. The Alpha values across the subgroups were high, with the exception of the group of pilots who continued to work, where the values varied between 0.779 and 0.794.

In the exploratory factor analysis, the KMO coefficient showed a value of 0.952, while Bartlett's statistic (χ² (66) = 3870.3, p ≤ 0.000) was significant, suggesting that the data matrix is appropriate for conducting an exploratory factor analysis. The parallel analysis (Timmerman and Lorenzo-Seva 2011) revealed just one dimension which accounts for higher variance than expected in random matrices. This factor overall explained 59.23% of the total variance of the psychological well-being measured by the 12 items on the instrument. Table 4 shows the psychometric properties and those of the 12 items.

Table 4. Item and psychometric properties of the GHQ-12.

Item	Factor loading*	Mean	Confidence interval (95%)	Skewness	Kurtosis
GHQ1	0.588	0.535	0.47–0.60	–0.141	–1.977
GHQ2	0.579	0.632	0.56–0.70	–0.547	–1.699
GHQ3	0.665	0.292	0.23–0.36	0.915	–1.161
GHQ4	0.804	0.202	0.15–0.26	1.491	0.219
GHQ5	0.743	0.646	0.58–0.71	–0.613	–1.622
GHQ6	0.950	0.254	0.19–0.31	1.131	–0721
GHQ7	0.606	0.550	0.48–0.62	–0.200	–1.957
GHQ8	0.839	0.246	0.19–0.31	1.185	–0.596
GHQ9	0.794	0.547	0.48–0.62	–0.189	–1.962
GHQ10	0.829	0.190	0.14–0.24	1.585	0.506
GHQ11	0.711	0.076	0.04–0.11	3.209	8.269
GHQ12	0.768	0.365	0.30–0.43	0.560	–1.684

*Extraction method: Factor loadings using the extraction method: unweighted least squares.

Discussion

This study reveals the psychological distress conditions of a group of Spanish aviation pilots coping with changes in their job stability as a result of the COVID-19 health crisis. To the best of our knowledge, no other study has analysed psychological distress in this group of professionals according to their current employment status. The current worldwide pandemic is having severe repercussions in the world of work, as reflected by our data, where a large percentage of pilots (75.5%) have found themselves in the unprecedented position of being unemployed or furloughed, with significant job losses expected (Iacus et al. 2020). This socioeconomic impact and the effect on jobs find a correlation in individuals’ mental and psychological health. As mentioned (Paul and Moser 2009; McKee-Ryan et al. 2005; Murphy and Athanasou 1999), unemployment has been shown to have a significantly negative impact on psychological well-being.

Firstly, it can be seen that the largest percentage of participants of our sample are aged between 41 and 50 years (45.3%), with the next largest age group being that of 51–60 years (28.1%). Our data are reasonably similar to those in the work by Gander et al. (2015), using four cohorts to assess pilot fatigue, where the captains had a mean age of 55 years and first officers a mean age of 46 years. In our study, although the mean total score was very close to the cut-off point suggested by Rocha et al. (2011), the overall sample showed no evidence of psychological distress. However, a large percentage (43%), were found to be at risk of psychological distress. These data were not compared to other pilot samples, but with studies analysing other groups of professionals and general population. Future studies are needed to confirm our findings among Spanish airline pilots.

The psychological distress revealed in the present study is somewhat lower than that reported by Ruiz-Frutos and Gómez Salgado (2021) in a sample of non-health workers, where 65% presented psychological distress and anxiety. The cut-off point, however, in their study was lower, set at 3 points. If we had considered this more restrictive score, the proportion of pilots with psychological distress risk would have been 59.1%. Additionally, according to Ruiz-Frutos et al. (2020), psychological distress was higher among frontline non-health workers aged 43 years or younger (69.4%) compared to their older counterparts (60.4%).

Furthermore, Banks et al. (1980) reported the differences found on the GHQ-12 score between employees in an engineering firm, recent school-leavers and a group of unemployed individuals. The scores were significantly higher for those unemployed, regardless of age, job level and marital status. According to their standard GHQ-12 score, 17% of the workers self-reported psychological distress (labelled as mental health by authors), while this rose to 60.4% among the unemployed group, with the mean score in this group being 4.76 ± 3.66 points. These figures are lower than those in the present study (unemployed: 6.00 ± 3.61 points). Nonetheless, these numbers are similar to the scores obtained by the workers on furlough or those whose job situation was insecure.

Studies have shown that unemployment and job insecurity are associated with mental health, with job insecurity being related to somatic symptoms while unemployment is strongly associated with worse general health and mortality (Kim and von Dem Knesebeck 2015). Additionally, a review of longitudinal studies, primarily conducted in the European Union, especially in Scandinavia, and also including works from Canada and the United States, found that job insecurity consistently affects health and psychological well-being over time (Witte, Pienaar, and Cuyper 2016). Thus, as reflected in the results of the present study, it should be noted that not only does unemployment present a risk for psychological distress but also job insecurity and instability. This consideration may be important in the implementation of psycho-emotional interventions and/or actions to promote psychological well-being intended to safeguard the mental health of pilots and other groups of professionals.

More recently, an article using data from the Spanish National Health Survey analysed the relationship between unemployment and psychological distress (labelled by authors as mental distress), specifically in the construction sector, which was severely affected by the financial crisis that started in 2008. Although the authors used different scoring systems to assess the GHQ-12, in all cases they found that unemployment led to higher total scores. In the case of the standard GHQ-12 for the data analysed, the mean score was 1.29 ± 2.29 points (Farré, Fasani, and Mueller 2018), substantially lower than in the present study. Moreover, as the authors note, this major recession led to the collapse of the construction sector, resulting not only in the loss of jobs but also in low re-employment probability, triggering a rise in poor health and psychological disorders. A similar impact may be felt in the aviation sector, due to the effects of the current COVID-19 pandemic. This would affect not only the number of employees that might lose their jobs, but also the probabilities of future employment in the profession, leading possibly to increased psychological distress at both qualitative and quantitative level.

It is worth noting the considerable prevalence of psychological distress among the pilots working normally (25%). Pilots, like other professionals working under special conditions, may be exposed to additional psychological and physical stressors, such as spending many hours in small cabins or the burden of responsibility for the safety and lives of their passengers and their own. The mean total Likert score for the normally working pilots in the study was 12.13 ± 4.64. Higher psychological distress prevalence (14.1) has been found in other groups, such as submariners in the British Royal Navy (RN), where the working conditions and isolated environment are very unique. The study in question also showed the psychological stress rate among other RN personnel working on shore, ships and overseas, which varied between 25% and 30% (Brasher et al. 2010).

As regards the scores according to the different items, the highest scores were found among the younger respondents, particularly those below the age of 41. As mentioned, this is the age range with the largest number of unemployed pilots. In this group, the highest scores were found on Items 3, 4, 5 and 6. Although they were close to significance, no differences with the older age group were found below the threshold of p ≤ 0.001. This is due to having applied a strict threshold given that with a standard cut-off point of p ≤ 0.05 the differences would be significant. Accordingly, it would be necessary to look more closely at the younger groups and those whose future employment situation is more uncertain. However, the analysis of the items according to the pilots' employment status reveals that the items that scored highest among those furloughed and unemployed were Item 5 (85% among the unemployed), Item 2, where no significant differences were found with those working normally, and Item 9, with more than 60% in both groups (furloughed and unemployed) scoring the highest option. The scores on Items 5 and 9 revealed significant differences between unemployed and furloughed pilots and those working normally.

In contrast to the study by Lundin et al. (2016), the lowest scoring item in the sample of pilots was not 10 but 11, which is consistent with the findings of Farré, Fasani, and Mueller (2018). Despite Item 11 presenting the lowest score in both the sample and across the three employment situations, 18.8% of the unemployed pilots responded with the highest score, compared to only 1.2% of those working normally.

With respect to the psychometric properties, the results show high internal consistency for our data, with a Cronbach’s alpha of 0.851. Furthermore, this value did not increase when any of the items were deleted. Other studies conducted in Spanish population have also reported the high reliability of the GHQ-12 (Sánchez-López and Dresch 2008; Rocha et al. 2011). Our analysis of the theoretical structure of the GHQ-12 confirmed the unidimensionality of the 12-item scale in a Spanish airline pilot sample where the construct was found to be stable with a one-factor structure explaining 59.23% of the variance.

The GHQ-12 was originally designed as a unidimensional instrument, but the number of factors or dimensions the scale is formed by has been a subject of much debate. Several studies have identified three factors: coping strategies, self-esteem and stress (Graetz 1991; Doi and Minowa 2003; Sánchez-López and Dresch 2008; Guan 2017; Picardi, Abeni, and Pasquini 2001). Others, however, have identified two: depression/anxiety and social dysfunction (Werneke et al., 2000; Glozah and Pevalin 2015). Finally, various studies have argued that the GHQ-12 should primarily be used as a unidimensional screening instrument to measure a single construct, general mental health (Hankins 2008; Ye 2009; Hu et al. 2007). This argument is grounded in the fact that an analysis of the differences between the factors or dimensions shows that the first factor includes all the positively phrased items, while the second contains all the negatively phrased ones (Rocha et al. 2011; Hystad and Johnsen 2020). Thus, although different factor analyses show an increase in the variance explained when more than one factor is used, this is explained by the response bias between positive and negative statements, as supported in various studies (Hankins 2008, Rocha et al. 2011; Hystad and Johnsen 2020; Ye 2009). In any event, what our data seem to confirm is the existence of a one-factor dimension in the sample of Spanish pilots surveyed during the COVID-19 pandemic. This one-factor structure is consistent with the initial version proposed by Goldberg (Goldberg and Williams 1988).

Rocha et al. (2011), in one of several studies on the psychometric properties of the GHQ-12, conducted in a representative sample of Spanish adult population comprising 29,476 participants, found, using the standard GHQ-12 score, that 73% of the variance could be explained by a single factor. The authors contend that the GHQ-12 can be used as a unidimensional screening instrument to measure mental health, especially when applying the standard GHQ-12 score. They also report an association between the questionnaire and individuals with depression symptom, anxiety or other psychological disorders, although it should be noted that this relationship was not validated using a gold standard mental health diagnostic test. Nonetheless, Lundin et al. (2016) also detected depressive disorder in general adult population, validating the GHQ-12 against psychiatric interviews designed to assess depression, considering the standard score performed acceptably in detecting depression.

A further study conducted in a sample of Spanish participants (N = 27,674) found the GHQ-12 to be a unidimensional measure containing spurious multicollinearity under the Likert scoring scheme due to the response categories in the negative items. Using the standard scoring method, as in the present study, eliminates this multidimensionality, avoiding the response bias (Rey et al. 2014). The results suggest that our study demonstrates the GHQ-12 is a twelve-item unidimensional scale with robust psychometric properties. It would be of interest, however, to continue research on the factor structure of the GHQ-12 with other population samples.

Among the limitations of this study is that, despite the novelty of assessing psychological distress in a typically inaccessible group of professionals, where there may be considerable stigma involved in admitting mental health problems, the sample is relatively small, especially as regards the unemployed group. A further limitation might be that a single method was used to determine psychological distress, without comparing with a gold standard method of mental health risk assessment as are clinical interviews by qualified practitioner. It is also worth noting this was an observational study and thus causality between variables could not be established.

Finally, we consider it of vital importance, given the current economic and healthcare crisis, and also looking to the future, to further analyse the psychological well-being of this group of professionals, using prospective studies that assess psychological distress in greater depth. Effort should also be made to break down the barriers to expressing mental health issues among pilots, and prevent stigmatisation in this profession, where significant stress factors exist. Attention should be focussed on developing preventive interventions to tackle psychological disorders in these times of high job insecurity, especially among the younger population. It is essential, now more than ever, to highlight the importance of pilots’ mental health to design effective and appropriate responses to safeguard the psychological and social-emotional well-being of aviation employees.

Conclusions

The GHQ-12 has been widely used in many studies with different population samples, but, to the best of our knowledge, had not previously been administered in a sample of Spanish airline pilots. In our sample, we found that the lower the age of participants, the greater was the psychological distress. However, we observed an inverse trend according to employment status, where the unemployed pilots presented the highest total scores, followed by those on furlough. As regards the items, six revealed significant differences between unemployed and working pilots. These items (3, 5, 6, 10 and 12) referred to feeling useful, feeling able to overcome problems, feeling sad or depressed, losing confidence, and feeling reasonably happy, all of which are related to poorer mental health, lower psychosocial well-being and greater symptoms of anxiety and/or depression. With respect to the psychometric properties, our findings for the GHQ-12 showed high internal consistency, and it is thus a reliable instrument to evaluate psychological distress in different job situations. The instrument was also found to have a one-factor structure that accounts for most of the variance.

We would like to underscore the importance of pilots’ mental health and the need to provide effective responses that can improve these workers’ psychological and social-emotional well-being. These times of health and economic uncertainty caused by the COVID-19 pandemic represent an opportunity to normalise and raise awareness of mental health problems, with the aim of seeking solutions. It is also of interest to delve deeper into this issue through exhaustive prospective studies.

Acknowledgements

We thank COPAC for their availability and effort in collecting the data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.