Overall, medical residents are engaged in their work [
]. Work engagement is defined as a positive, fulfilling, work-related state of mind that is characterized by vigour (high levels of energy while working), dedication (strongly involved in work), and absorption (fully concentrated and happily engrossed in work) [
]. Work engagement is related to being more committed to the organization, having less turnover intentions, more satisfied clients and better financial outcomes [
]. Within the medical sector work engagement is related to making fewer medical errors. Highly engaged residents reported fewer errors due to inexperience and/or lack of time than those who were less engaged [
]. Although job demands and work pressure are high in hospitals, and stress and burnout are common in medical residents, one-fourth of the residents can still be characterized as highly engaged [
]. Why are some residents more engaged in their work than others? How can we increase work engagement in medical residents? In order to stimulate work engagement in medical residents, we should gain a better understanding of potential contributors of work engagement in this specific population.
According to the Job Demands and Resources theory and previous studies, job resources are positively related to work engagement [
]. Job resources refer to those aspects of the job that are functional in achieving work goals, reduce job demands and the associated physiological and psychological costs, and/or stimulate personal growth and development [
]. Important resources are opportunities for development, performance feedback, autonomy, skill variety, and social support from colleagues and supervisors. However, the specific resources that predict work engagement may differ per organization [
]. In addition to job resources, job demands also seem to be related to work engagement [
]. Job demands refer to work characteristics that require physical and/or psychological effort and are associated with certain physiological or psychological costs [
]. Although demands are associated with costs and often related to negative work outcomes, job demands that are challenging such as job responsibility, workload, and time urgency might also be positively related to work engagement [
]. However, a study by Mache et al. [
] in 123 surgeons found that job demands were negatively related to work engagement, and that job resources had a greater impact on surgeons’ work engagement than job demands. Especially having influence at work, opportunities for development and social support seemed to contribute to work engagement in surgeons. Furthermore, several studies indicate that interaction between demands and resources is relevant as well, suggesting that job resources become more salient and gain their motivational potential when employees are confronted with high job demands [
Less is known about the potential influence of home characteristics on work engagement, but several studies indicated that one’s functioning at home might impact one’s functioning at work and vice versa [
]. A study by Bakker et al. [
] found that home resources and demands were associated with work engagement in a sample of 323 Dutch dual-earner couples, whereas a longitudinal study by Hakanen et al. [
] did not find an effect of home demands or resources on work engagement in Finnish dentists (
The interference between the work and home domain might also be related to work engagement [
]. This work-home interference and home-work interference can be experienced negatively when demands from the work and family roles are incompatible, such that participation in one role makes it difficult to participate in the other, and positive when positive experiences from one role make it easier to enhance participation in the other role [
]. A study by Montgomery et al. [
] found that positive interference between the home and work domain was correlated with feelings of dedication at work.
With the increasing percentage of women within the medical profession it is very relevant to examine gender differences in work engagement and their contributing factors [
]. Prins et al. [
] indicated that female residents report significantly less vigour than male residents. Furthermore, Verweij et al. [
] found gender differences in contributing factors to burnout. In females, social support from family or partner seemed protective against burnout. In males, social support from colleagues and participation in decision-making at work seemed more important. As burnout and work engagement are negatively correlated, we might suspect that there are gender differences in work engagement as well. In addition, Rothbard [
] found that a positive affect towards the home/family life was related to work engagement only among women. This could indicate that the potential contributors of work engagement might differ between male and female residents.
The aim of the present study was to test a model including all the potential contributing factors of work engagement as described before, in order to examine which factors are most important. This will inform us about how to enhance work engagement. In addition, we explored whether there are gender differences with regard to possible contributing or protective factors of work engagement. We formulated the following hypotheses:
We expect (a) the job and home demands to be negatively related to work engagement, (b) the job and home resources to be positively related to work engagement, (c) the job/home resources to be more strongly related to work engagement if job/home demands are high, (d) positive work-home/home-work interference to be positively related to work engagement, and negative work-home/home-work interference to be negatively related to work engagement.
We expect the model to be different for male and female residents; we expect the home demands and resources to be more strongly related to work engagement in female residents than in male residents.
We analyzed the data by means of path analysis, using the Mplus7 statistical software package [
]. Path analysis is a subset of structural equation modelling using only measured variables and no latent variables. We developed a model and examined the associations between the specific work and home demands and resources and the four types of work-home interference on the one hand, and the three work engagement subscales on the other hand. These associations or pathways in the models represent hypotheses, which are based on previous research and theoretical propositions as described in the introduction. We included work characteristics in our analyses first, and added home characteristics in a next step, in order to be able to examine if including home characteristics explained an additional proportion of the variance in work engagement, beyond the effects of work characteristics. To examine possible gender differences in these associations we employed multi-group path analysis in Mplus7 [
]. The improvement in fit of the models was assessed using the chi-square difference test, the root mean square error of approximation (RMSEA) and the comparative fit index (CFI). Values of 0.90 and higher (CFI) and 0.08 or lower (RMSEA) indicate an acceptable fit [
]. Because of the large dataset and the number of variables included in this model we chose to indicate
‑values below 0.01 as significant in order to decrease the chance of type 1 errors. Standardized path coefficients (Beta values) were calculated to determine the possible predicting factors of work engagement. These indicate the patterns of associations between the predicting factors and work engagement. Standardized path coefficients (β) with values of less than 0.10 can be interpreted as small effects, values of around 0.30 can be interpreted as medium effects and values above 0.50 can be interpreted as large effects [
Before testing our hypotheses, the items representing the constructs in the research model were subjected to confirmatory factor analyses using Mplus7. All measures we used were employed in previous studies and generally proved to be valid. The fit indices of the measurement model were near adequate, χ
2(3444) = 17,382.40, RMSEA = 0.04, CFI = 0.85, standardized root mean residual = 0.05. All items loaded significantly on the predicted factor. However, one item from the homeload construct had a low factor loading (<0.40). We decided not to remove this item because of its conceptual importance and because the internal consistency of the construct, expressed with Cronbach’s α coefficient, was considered sufficient (α = 0.75).
The aim of this study was to examine the associations of work and home characteristics and work-home interference with work engagement in Dutch male and female medical residents.
The results partly confirmed our hypothesized model in that we found work demands and resources, home demands and resources, and work-home and home-work interference to be associated with work engagement. However, we did not find support for hypothesis 1c; demands did not moderate the association between resources and engagement. Furthermore, our data do not provide any evidence for possible gender differences in the associations between work and home characteristics and work-home interference, and engagement (hypothesis 2).
Examining these associations in more detail, we see that especially opportunities for job development, positive work-home interference, positive home-work interference and mental demands at work were important contributing factors to all three subscales of work engagement. Interestingly, ‘mental demands at work’ was positively rather than negatively related to work engagement, suggesting that mental demands are considered to be challenging rather than exhausting [
]. Emotional demands at work and negative home-work interference had an important negative impact on work engagement.
There was only a very small effect of home demands and resources on work engagement. The work-home interferences, both the negative and positive home to work interference and work to home interference, seemed more important for experiencing work engagement. This might indicate that for medical residents the home situation in itself is not that influential, but the interference between the home and work domain has a greater impact on engagement at work.
Furthermore, we did not find gender differences regarding the associations in the model. This is in contrast with our expectations, but it is in line with research in a heterogeneous occupational sample (
= 846) by Korunka et al. [
], who did not find gender differences either in the relationship between job resources and work engagement. The strengths of the associations between resources, demands and work-home interferences with work engagement in our research are similar for male and female residents. However, previous research in medical residents demonstrated that female residents experienced more home demands and more negative home-to-work interference, while male residents reported more social support from colleagues and supervisory coaching [
]. The circumstances seemed slightly better for male residents as they reported more of what is needed for being engaged, which might explain that male residents also reported more vigour compared with female residents [
Although, more insight into the specific contributing factors of work engagement might lead to a better functioning of the residents themselves, the organization and a higher quality of patient care, it is also important to be aware of the possible downsides of work engagement [
]. Bakker et al. [
] suggested that there might be a limit to engagement. Overly engaged workers may often work overtime, forget to rest or maintain their personal relationships and hence experience more work-family conflict. In a medical culture with its emphasis on commitment, the high responsibility in patient care and expectation to work overtime, this is important to bear in mind.
Strengths and limitations
One of the strengths of this study is the large and representative population. All medical residents in the Netherlands were invited to participate, which resulted in a large sample with residents from different regions of the country, different hospital settings and different medical specialties.
As any study, the present study also has limitations. First of all, the data were collected at the end of 2005. So, our findings might have been overhauled by social or cultural changes that have since taken place or by developments in the postgraduate training of medical residents. However, we believe that although in 10 years the absolute values of work engagement and its contributors might have slightly changed, the relationships between these variables and work engagement have not been affected by such changes. The Job Demands-Resources model, on which we based our hypotheses, assumes that the availability of job resources helps in dealing with job demands and plays a motivational role, which fosters work engagement [
]. There is no reason to believe that this process or the specific characteristics associated with work engagement in medical residents have changed during the past decade because the nature of the job has not essentially changed.
Secondly, the data were collected at one point in time, so the associations cannot be interpreted causally. For instance, it could also be that medical residents with high levels of engagement more easily identify positive work-home interference. However, longitudinal research by Hakanen et al. [
] did not find evidence for the reversed effect of work engagement on job demands and resources.
Thirdly, although considering the nature of the survey the response rate was quite high (41%), more than half of the invited medical residents did not respond. This could have led to a selection bias. Common reasons for not responding in those who took the trouble to send in their non-response form appeared to be a lack of time (22%) and a lack of energy (11%). This might have resulted in an overestimation of work engagement in medical residents. However, again, this might especially have affected the absolute levels of burnout and its contributors instead of their relationships. Furthermore, as all measures were based on self-reports, common method variance may have biased our findings. Ideally, future research should combine self-report questionnaires with more objective measures, such as computerize tasks, observations and organisation’s records.
Future research and practical implications
Future research should use longitudinal designs to better explore the predictive nature of the associations found in our model and the possible long-term consequences of work engagement in relationship with health. Furthermore, forthcoming studies about work engagement would benefit from a focus on interventions, and study the impact and effectiveness of these interventions.
With regard to implications for clinical practice, our findings support the improvement of resources at work, such as better opportunities for career development, personal development and creating a supportive culture. Furthermore, residents should learn how to deal with emotional demands at work, for example by facilitating regular supervision or peer supervision where junior doctors can exchange difficult situations they might have encountered. Alternatively, one might stimulate curricular initiatives that incorporate aspects of allowing and regulating emotions, such as narrative medicine or mindfulness training [
]. In addition to changing circumstances at work, there should be more attention to the benefits of the decreasing negative interference and stimulating positive interference between home and work. Hospitals could provide facilities supporting the work-life balance such as flexible working hours, self-scheduling, part-time work, and childcare. Secondly, and this might even be more important, hospitals should create a culture in which residents feel supported in using these facilities without experiencing a negative effect on their career opportunities [