Extraversion was positively associated with subsequent sickness absence when controlling for several covariates, including health, work factors and previous spells of sickness absence; neuroticism also showed positive associations with sick leave

Big five personality traits and physician-certified sickness absence. Gøril Kvamme Løset, Tilmann von Soest. European Journal of Personality, January 3, 2022. https://doi.org/10.1177/08902070211065236

Abstract: Although several studies show that personality traits are associated with absenteeism, few large-scale studies have examined these relationships prospectively, integrating survey data and register data on sickness absence. This study examines whether personality is associated with sickness absence, and whether health factors, gender, age, type of occupation and job satisfaction moderate this relationship. We combine survey data assessing the Big Five personality traits from a large sample of Norwegian employees aged 18–62 years (N = 5017) with register data on physician-certified sickness absence up to four years after. Negative binomial regression analyses showed that extraversion was positively associated with subsequent sickness absence when controlling for several covariates, including health, work factors and previous spells of sickness absence. Neuroticism also showed significant positive associations with sick leave; however, the association diminished when accounting for previous spells of sickness absence. Moderator analyses demonstrated that age and type of occupation affected some of the associations between personality and sickness absence. The findings indicate that – in addition to general health promotion measures – specific interventions targeting individuals high in extraversion may be beneficial in reducing sick leave. How socio-demographic and work-related factors moderate the relationship between personality and sickness absence may be an interesting future research area.

Keywords: sickness absence, sick leave, personality, health, occupation

Discussion

The main purpose of this study was to examine whether personality is associated with physician-certified sickness absence. Overall, the results suggested that after controlling for a variety of relevant covariates and excluding respondents with prior spells of sick leave, high levels of extraversion were prospectively related to an increased risk of sick leave. Furthermore, neuroticism was positively associated with sick leave even with control for covariates; however, the association diminished when excluding participants with prior spells of sick leave. Two interaction effects showed further the complex relationships between personality, age and work factors, and sick leave.

The prospective associations between personality traits and sickness absence

Neuroticism was related to an increased risk of future sick leave, also when health, family and work factors were included in the analyses. These findings are in line with studies that found cross-sectional (Störmer & Fahr, 2013; Vlasveld et al., 2013) and longitudinal (Blekesaune, 2012; Raynik et al., 2020) associations between neuroticism and sick leave with control for health factors. However, new in our study is the finding that the prospective association between neuroticism and sickness absence diminished into insignificance when removing respondents that had sick leave in the interview year. This result may thus suggest that the neuroticism-sickness absence association does not necessarily originate in causal mechanisms where neuroticism influences sickness absence, but that associations may be explained by confounding or reverse causal directionality. Such a notion is in accordance with research showing that major negative life events, for example, the onset of chronic disease, and associated mental distress and deterioration in life quality predict lasting increases in neuroticism (Jeronimus et al., 2014; Jokela et al., 2014).

The positive relationship between extraversion and sick leave corresponds with previous studies on extraversion and absenteeism (Furnham & Bramwell, 2006; Judge et al., 1997), but conflicts with the results of Vlasveld et al. (2013), who found a negative effect of extraversion on both shorter and longer sick leave spells (>2 weeks). This latter study explains its findings in terms of extraverts possibly being less likely to deal with, for example, work stressors by means of avoidance coping through sick leave. In contrast, the positive associations between extraversion and absenteeism in other studies were suggested to originate from extraverts’ tendency to prioritise leisure and social obligations outside the work sphere, especially when work tasks are mundane (Furnham & Bramwell, 2006; Judge et al., 1997). This explanation, though, seems less intuitive for predicting longer term physician-certified sick leave. Possibly, the tendency of extraverts to take risks and seek excitement may increase the risk of long-term sick leave, as extraverts are more prone to substance use and accidents (Booth-Kewley et al., 1994; Terracciano et al., 2008). Alternatively, in line with the concept of presenteeism, individuals low in extraversion may be less likely to seek sick leave even when being ill. More specifically, by having a tendency to be reserved and inhibited, these people may prefer to continue with their normal work activity in order not to attract attention and have to disclose themselves to the doctor and colleagues. Also, people with low levels of extraversion may refrain from going to the doctor and approaching colleagues and superiors for sick leave because of lower levels of self-worth and social skills compared to extraverted people (Ozer & Benet-Martínez, 2006; Robins et al., 2001). The present study did not provide data that allowed to test for these potential mechanisms of the association between extraversion and risk of sick leave. Future longitudinal studies with a more comprehensive assessment of extraversion, including facets such as excitement seeking and assertiveness, may provide the opportunity to test such mechanisms by disentangling the effect of specific facets of extraversion on sick leave.

The lack of any main effect of conscientiousness in our study was surprising, given earlier findings that conscientiousness is negatively related to work absence and the well-documented health- and task-directed nature of conscientious individuals. Nevertheless, the other previous study on personality and register-based sickness absence also did not find an effect of conscientiousness (Blekesaune, 2012); neither did a study based on longitudinal survey data (Raynik et al., 2020). This could indicate that the potentially buffering effect of conscientiousness on work absence primarily appears for absenteeism and more short-term sick leave. In fact, conscientiousness has been positively linked to emotional exhaustion (Armon et al., 2012). Thus, although conscientious individuals are committed and motivated in their work, which would generally yield negative associations to work absence, these characteristics might entail aspects that over time could predispose them for longer absences due to burnout as well (Armon et al., 2012; Woods et al., 2013).

Less surprising, however, was the non-significant main effect of openness on sickness absence, as it is in accordance with most previous studies and thereby supports the notion that this personality dimension does not appear to be an overall decisive predictor for work absence. Still, a moderator effect of openness with age on sickness absence was found, indicating that openness increased the risk of sick leave for older employees compared to younger employees. Possibly, for younger employees, openness may promote adaptation to shifting work demands and integration in new workplaces which are of importance early in an occupational career. In contrast, high levels of openness may impede job performances and increase risk of sick leaves among older employees when extensive experience makes work less challenging and more monotonous.

For agreeableness, the lack of a main effect did not support our hypothesis of a negative association with sick leave. Yet, the moderator analyses suggested that agreeableness might have some bearing on sick leave in more intricate manners. The identified moderator effect of agreeableness with type of occupation lends some support to our assumption that personality may have differential associations with sickness absence when contrasting occupations that are not physically demanding with manual occupations. Perhaps, highly agreeable persons may be less willing to be absent from work in manual occupations because they perceive the additional burden on colleagues due to their own absence to be greater in physically strenuous occupations compared to non-manual occupations. This seemingly complex association between personality, type of occupation and sick leave, may be an interesting area for future research.

Strengths and limitations

The present study is the first to examine in a nationally representative sample the longitudinal relationship between personality and physician-certified longer term sick leave using register data on sickness absence. The use of a large, representative survey sample combined with highly reliable register-based data on sickness absence is a major strength. Using register data on sickness absence is advised because self-reported sick leave is prone to recall bias and social desirability (Thorsen et al., 2018). Moreover, several previous studies relied on cross-sectional survey data, thereby providing limited information about the temporal relationship between personality and sick leave, which further clouds causal inference.

However, the study also has limitations. First, the available register data on sickness absence only provided information about spells of sick leave that lasted more than 16 days. Shorter periods of sick leave, including both self- and physician-certified spells were not assessed, and the present study provides no information about how such spells are related to personality. Furthermore, data on sickness absence only provided information about the accumulated number of weeks of sickness absence within a calendar year for each of the four years (2008–2011). It was thus not possible to differentiate between several shorter term spells of sickness absence and one long-term spell within the year.

We used a short version of a widely applied personality measure. Yet, with four items measuring each trait, the breadth of coverage of this measure is limited, and thus this may also have limited its potential in predicting the outcome. With a more comprehensive instrument, we would be able to capture more nuances and assess narrow personality facets, which may provide more detailed information, in addition to broad traits in predicting sickness absence (Judge et al., 1997; Lounsbury et al., 2004). Personality facets or items tend to outperform the broader personality traits in predicting a wide range of behavioural outcomes (Paunonen & Ashton, 2001; Seebooth & Mõttus, 2018). However, studies show that the Big Five seems sufficient for predicting work-related behaviour (Judge et al., 1997; Woods et al., 2013). Still, discrepancies in findings between studies may be the result of different representations of items of personality traits used. Also, a more comprehensive measure of personality would provide more information about different mechanisms that can operate at lower trait levels and disentangle whether facet or item level associations with sickness absence go in different directions.

We chose a rather conservative level of significance with p < .01; however, we acknowledge that interaction effects would be considered not significant with a more strict correction for multiple comparisons when conducting moderation analyses. The two identified interaction effects should therefore be considered preliminary.

Finally, the generalisability of the study results may also be limited in other ways. Although the study sample was stratified to be nationally representative, the respondents were overall higher educated and healthier than the general population (Slagsvold et al., 2012), which could lead to somewhat biased study results. Nevertheless, the use of survey weights did not significantly change the results compared with unweighted analyses, thus indicating that such biases in the sample do not appear to be a major concern. The findings may further not be generalisable to other countries that have less generous sickness benefit schemes, a less inclusive working life, higher unemployment, or other labour market conditions that are different from the Norwegian context and that may affect the sickness absence rate.