Transformational leadership encourages residents’ job crafting in surgical training: A mixed-methods study of residents’ perceptions


 Background Supervisors’ leadership style can enhance resident performance, especially in terms of their ability to deal with the demands in the workplace and to take advantage of the available resources. Dealing with job demands and resources is known as job crafting, which has implications for the persistence of residents in training. The link between supervisors’ leadership style and residents’ job crafting, however, is not well understood.Methods This mixed-methods study sought to explore the relationships between a transformational (team-oriented), transactional (task-oriented), and laissez-faire (passive) supervisory style and residents’ job crafting and to explain these relationships. Residents filled out the Multifactor Leadership Questionnaire to rate their supervisors’ leadership style and the Dutch job-crafting scale to assess their own job crafting. We tested the relationships using linear mixed effects regression analysis. To explain the ensuing results, we subsequently conducted a thematic analysis of semi-structured interviews with residents.Results One hundred and sixteen residents from 7 surgical programs participated. A transformational leadership style had a positive effect on residents’ job crafting (b = 0.19; 95% CI 0.08–0.32; p = 0.009), whereas the transactional and laissez-faire styles did not. This could be explained by the fact that residents felt their transformational supervisors had a positive influence on the atmosphere for training and on the job resources available to them, and considered them positive role models for how to handle the demands of the environment.Conclusion In residents’ view, a transformational style is positively related to the ability to craft their jobs and therefore has implications for their persistence in training. Future research should explore supervisors’ perspective on this relation and the effectiveness of leadership training for supervisors with a focus on resident outcomes, such as job crafting and persistence in training.

work, diminish bullying and burnout, and foster work engagement (14,15,(17)(18)(19)(20). Conversely, a transactional leadership style (TrLS) is task-oriented and aims to ful ll objectives, ensure standards and monitor outcomes (14). Those who have a TrLS offer rewards or punishments to others according to their performance during tasks. They are likely to discourage and decrease empowerment, job satisfaction, and work engagement (21)(22)(23)(24). Finally, a laissez-faire leadership style (LfLS) is passive and aims to meet the expectations of the organization. Those who have a LfLS are rarely present and they diminish job satisfaction, productivity, and job effectiveness (14,25).
In this study, we hypothesized that a TLS in supervisors would be positively associated with residents' job crafting in surgical training.
We also hypothesized that a TrLS and an LfLS would be negatively associated with residents' job crafting, due to the aforementioned characteristics. The study addressed the following research question: To what extent are supervisors' leadership styles associated with residents' job crafting in surgical training and why, from residents' perspective?

Methods
We employed a sequential, explanatory mixed-methods research design (26). More speci cally, we rst measured supervisors' leadership style and residents' job crafting using questionnaires. Then, we conducted interviews with residents to gain a deeper understanding of why these styles were perceived to affect their proactive behaviors. Such a research approach responds to the need to integrate multiple sources of data that together, by drawing on the strengths of quantitative and qualitative methods, can help explain the complex relationship between leadership and job crafting (27)(28)(29). The Commission for Medical Education of the Universidad de la Sabana granted ethical approval.

Setting and participants
This study was conducted in Colombia where residents hold full-time positions in healthcare institutions during four years of surgical training. Annual tuition for training in private programs is close to 12,000 USD, which includes 66 hours of duty per week. Three out of 20 residencies in surgery are accredited as high-quality programs in accordance with national standards; two more are in the process of attaining that accreditation. Considering the number of a liated institutions, each program hosts more than 30 surgical supervisors.
Residents' burnout rate is 33% and almost 11.9% of residents have serious intentions to leave training (5).
We invited all the residents of seven surgical programs (a total of 136 residents) to participate voluntarily in the quantitative phase between October and December, 2018. We chose the ve programs that were either accredited or in the process of obtaining high-quality accreditation. We included two others that had the longest standing tradition in the country. We emphasized con dentiality and anonymity of data.

Measures Supervisors' leadership styles (independent variables)
The variables we measured were three: TLS, TrLS and LfLS. We obtained permission to administer the Multifactor Leadership Questionnaire (MLQ-5X) to residents to evaluate these styles in their supervisors (15). We asked residents to rate their supervisors in general, hence not their individual supervisors, direct supervisor, or the program director. We chose this approach because, rst, residents interacted with multiple supervisors on a daily basis; second, the high number of supervisors per program limited the feasibility of each resident rating them all; and nally, program directors were not fully involved in clinical supervision. Residents were instructed to rate 'the standard surgeon who supervises residents (in the workplace) in the program that you are enrolled in...' on a 5-point scale (1 = not at all; 5 = frequently, if not always). This instruction followed the principle of standard reference by which decision-makers -residents in this case -assign intermediate values to their expectations, ranging from best to worst (30).

Residents' job crafting (dependent variables)
We measured ve variables that corresponded to each job-crafting domain: (1) increasing structural resources; (2)

Statistical analysis
We rst calculated the descriptive statistics for all variables. For each program, we clustered the individual answers on the MLQ-5X and DJCS to represent the overall scores for each leadership style and job crafting. In these analyses, we calculated means, standard deviations (SD), and 95% con dence intervals (CI) for the independent and dependent variables per program. We conducted one-way univariate analyses of variance (ANOVAs) to identify signi cant differences in scores between programs. We adjusted p-values of ANOVAs (Bonferroni correction) considering the number of tests. We computed Cronbach's α for questionnaires.
We used R (R Core Team, 2019) and lme4 (31) to perform a linear mixed effects analysis of the relationship between job crafting and leadership styles. As xed effects, we entered the leadership styles into the model (without interaction term). As a random effect, we had intercepts for the program. We therefore applied the following mixed-effect model to the data in which JC represented global job crafting (or each one of the four job-crafting variables): Visual inspection of residual plots did not reveal any serious deviations from homoscedasticity or normality. P-values were obtained by likelihood ratio tests of the full model with the effect in question against the model without the effect. For each xed effect we reported the b estimate, the 95% CI, and the chi-square (df = 1) p-value (signi cance level of < 0.05). In general, b estimates around 0.10, 0.25, and 0.40 can be interpreted as small, medium, and large effects (32). For the random effect we reported the variance, standard deviation (SD), the 95% CI, and a simulation-based p-value (signi cance level of < 0.05). We also calculated the interclass correlation for the random effect (ICC = variance (program) / (variance (program) + variance (residuals))). Finally, we assessed the model's goodness of t with the R2m (marginal R-squared) and the R2c (conditional R-squared) indices (33).

Qualitative phase
The qualitative phase took place between January and April, 2019. First, we developed an interview guide to explore the quantitative results (Appendix 1). Then, LCD and AS recruited 20 residents using strati ed sampling based on demographics (e.g., program and year of training). All interviews were conducted by phone. All participants gave verbal informed consent to be involved in the interviews after we had explained the mechanisms to ensure anonymity, con dentiality, and management of information. LCD conducted the individual, in-depth interviews in Spanish, using a non-technical language to guide the participants in each leadership style and formulate key questions. Upon completion, all interviews were immediately audiotaped and transcribed verbatim. Then, LCD and AS performed a thematic analysis of all transcriptions. With this method, we sought to identify themes within our dataset (34) and reach thematic saturation, that is, the stage in which no new categories appeared, previous data did not require any more modi cations, and no additional data were needed (35). After 10 interviews, LCD and AS had identi ed the main themes, but felt more information was needed to explain some aspects of these themes in depth. We therefore conducted 14 interviews in total, after which LCD and AS felt thematic su ciency was reached. All authors subsequently discussed these themes iteratively to reach consensus.
In this analysis, we acknowledged, through re exivity, that we as researchers add meaning to the ndings. The researchers have different backgrounds, experiences, and perspectives on leadership and residency training, which may have in uenced data collection and analysis. LCD, AS, and LS are surgeons, supervisors, and directors of surgical programs. DD and WdG have extensive experience as educational researchers in workplace-based learning in residency training. Ultimately, our interpretation of the ndings was in uenced by the concepts of JD-R theory (11). Our different perspectives combined with the said theoretical concepts may bene t the strength of the study and the transferability of its ndings. Finally, in all stages of the qualitative phase, we followed recommendations for the translation of information, in our case from Spanish into English language (36).

Quantitative results
We included 116 residents from seven programs (92.6% response rate). Table 1 presents the characteristics of the participants. The mean scores of the MLQ-5X were: TLS = 3.39 ± 0.72; TrLS = 2.93 ± 0.46; and LfLS = 2.07 ± 0.75 (1-5 range). The mean scores of the DJCS were: global job crafting = 3.50 ± 0.41; job crafting to increase 1) structural resources = 4.35 ± 0.52; 2) social resources = 3.75 ± 0.66; and 3) challenging demands = 3.50 ± 0.61 (1-5 range); and job crafting to diminish hindering demands = 2.59 ± 0.73. Table 2 shows the results of the ANOVAs for the main variables between programs. After Bonferroni correction (p < 0.006), we identi ed Page 5/17 signi cant differences in the scores for global job crafting to increase social resources and to increase challenging demands. Similarly, we identi ed signi cant differences in the scores for a TLS and a TrLS. The Cronbach's α of the MLQ-5X was 0.94. Cronbach's alphas for the subscales of the DJCS ranged from 0.63 to 0.78.  Degrees of freedom = 6 (in all cases) **Probability was adjusted to p < 0.006 (Bonferroni correction) Table 3 presents the linear mixed effects analysis of the relationship between supervisors' leadership styles and global job crafting. A TLS had a positive effect on global job crafting (b = 0.19; 95% CI 0.08-0.32; p = 0.009). In general, the difference between programs regarding the effect of leadership style on job crafting was relatively small (ICC ranged between 0.002 and 0.16). We found no signi cant relationships between a TrLS and LfLS and global job crafting.  Table 4 shows the linear mixed effects analysis of the relationship between the leadership styles and each job-crafting domain. In general, a TLS was positively related to all job-crafting domains. In three domains (increasing structural and social resources and diminishing hindering demands), the effect was signi cant (p < 0.05). We found no signi cant effect between a TrLS and any jobcrafting domain. An LfLS had only a signi cant effect on job crafting to decrease hindering demands (p = 0.0001).  (Tables 3 and 4).
Qualitative results: Exploring the associations between supervisors' leadership style and residents' job crafting Six participants were female (42.8%). The distribution of participants by year of training was as follows: Year 1 (n = 4), year 2 (n = 3), year 3 (n = 3), and year 4 (n = 4). Three predominant themes emerged from the interviews. Table 5 gives an overview of representative quotations. When we [residents] nd a person who is charismatic, who salutes, who is decent, who has manners, who gets along well with all people, and who has good results and relationships with patients, we identify with that type of leader. A leader is someone who not only knows how to manage people, but also shows other people how to solve problems. Almost always, this type of leader has the solution to problems, easily and without collateral results.
(Interview #14: 1st -year male resident) "In an environment of trust the resident has a better development, while he/she loses his/her fears of being judged by his/her opinion and decisions ... You feel a safe environment where you will not be judged negatively." (Interview #8: 4th -year male resident) "This supervisor supports good decisions [by residents], or appropriately corrects wrong decisions. This generates a good, safe, and positive environment, which leads to more calmness for the resident. One works better and has better results." (Interview #9: 3rd -year male resident) "A resident who has a good modelsomeone who inspires him/her more, [expresses] con dence and encourages his/her improvement-is a better example than the one who is just waiting for an excuse to punish." "That supervisor is not a good role model because he/she is an indecisive person, someone who confuses." (Interview #14: 1st -year male resident) Theme 1: Supervisors' leadership style in uences the atmosphere for training in positive or negative ways Residents mentioned that supervisors who had a TLS could promote a positive atmosphere for training and showed high standards of patient care. The personal strengths they attributed to such supervisors were altruism, integrity, resilience, and trustworthiness.
Residents valued this atmosphere because it made them feel free to discuss their fears and expectations of training, strengthening both their performance (e.g., decision-making and problem-solving skills) and readiness for practice. In residents' view, these supervisors stimulated them to stay in the program and pursue their training.
Conversely, residents characterized supervisors who had a TrLS as people who actively searched for errors, and were punitive and authoritative, which did not encourage them to take the lead in their own training on the job. In most cases, these supervisors created a hostile atmosphere for training where residents experienced fear and mistreatment, leading to defensive behaviors to hide errors and avoid punishment, and to more intentions to leave training.
Finally, residents mentioned that supervisors who had an LfLS showed a lack of commitment to patient care and residents' education and were perceived to help create a negative atmosphere in the workplace. Moreover, they contributed to more demands for the resident (i.e., more workload), resulting in unsafe care for patients. At the same time, however, the rare presence of these supervisors in the workplace encouraged residents to take more care of patients and to deal with workload and pressure.
Theme 2: Supervisors' leadership style in uences the availability of job resources Residents mentioned that supervisors who had a TLS offered the resident more job resources and challenges in the workplace (e.g., in the form of support, teaching, and feedback). Supervisors who had a TrLS and LfLS, by contrast, provided fewer of these resources and challenges, while creating more hindering demands (e.g., workload and pressure). For instance, supervisors who had a TrLS gave poor feedback and instruction and frequently punished residents by limiting opportunities to participate in surgical care and to take on new challenges (such as the opportunity to operate complex patients). In residents' views, these negative aspects of training led to psychological distress, a lack of autonomy, and more intentions to leave training.
Theme 3: Supervisors' leadership style serves as role model for how to handle the demands in the workplace According to residents, supervisors with a TLS were positive role models. More speci cally, these supervisors were capable of handling the work environment and nding solutions to di cult situations effectively, for instance when surgical complications or con icts arose at work. Supervisors with a TrLS or an LfLS, on the other hand, were perceived as negative role models, because they created more demands for the resident (in the form of con icts, ambiguity), while wielding ineffective strategies to solve di cult situations.

Discussion
The results from our survey demonstrated that supervisors' leadership styles and residents' job crafting differed signi cantly across the programs under scrutiny. Globally, however, we found the transformational leadership style to have a signi cant effect on residents' job crafting in both xed and random effect analyses (with the interaction of programs). Residents valued supervisors with such a style for their positive in uence on the training atmosphere and on the availability of job resources, and because they served as positive role models. Conversely, neither the transactional nor the laissez-faire style was found to have a signi cant effect on residents' job crafting.
During the interviews, however, residents argued that these supervisors had a negative in uence on the training atmosphere, the availability of job resources, and on role modeling.
We must view these results in relation to the existing research. Our ndings echo those of previous research pointing to the positive in uence of supervisors with a TLS on the atmosphere in the workplace (20, 37,38). In our study, a safe atmosphere -understood as a non-punitive and open environment for training -helped create favorable conditions for residents (e.g., trust in the supervisor, less power distance and less fear to discuss expectations), inducing them to search for more opportunities to participate in decision-making, to solve complex problems, and to cope with adversity. In other words, a positive atmosphere is one of the centerpieces of residents' job crafting. Our results, moreover, suggest that such a positive atmosphere may depend on supervisors' ability to create a deep connection with residents, which supervisors with a TLS do. These ndings tie in nicely with previous studies on the importance of supervisors' behaviors to a positive learning climate and residents' well-being (39)(40)(41).
To our knowledge, this study is the rst to explore the impact of a TLS in surgical education, as most studies have hitherto focused on its effects on clinical outcomes (patient safety and team performance) (42,43). Moreover, our study emphasizes the importance of a TLS to a crucial aspect of residents' education, that is, residents' job crafting, considering the complexity of the surgical work environment for training. Our results also suggest that supervisors must not only offer residents structural resources (e.g., autonomy and responsibility), social resources (e.g., feedback and coaching), and more challenging demands (e.g., participation in complex cases), they must also encourage them to seek these, so that residents can craft their jobs e ciently and improve their performance.
These results are in keeping with studies into the effect of a TLS on empowerment and autonomy in healthcare contexts (22,44). We found that supervisors who embrace a TLS can help residents to gain control at work, by demonstrating effective ways to handle hindering demands and stressors (e.g., con icts, frustrations of training). Few studies have considered judging surgeons who serve as role models by abilities other than their "surgical skills" and "mastery of technique." The ability to deal with con icting demands and to cope with adversity are examples of what residents expect to learn from their supervisors beyond the traditional dexterity competences (45-47). Our ndings indicate that supervisors should be aware of their modeling function with respect to these non-technical competences.
Our qualitative ndings, on the other hand, suggested that supervisors with a TrLS have a negative in uence on the atmosphere for training. Fear and power distance were important factors that explained such a hostile atmosphere. These factors, in turn, serve to illustrate how the supervisors who embrace this leadership style are disconnected from residents in the workplace: Ultimately, these supervisors negatively affect the availability of job resources and increase hindering demands (i.e., workload). Other studies have reported similar ndings with respect to an authoritative leadership style in the supervisors (20, 24,37,38,48). Nonetheless, contrary to what we expected, we identi ed a positive association between an LfLS and residents' job crafting to diminish hindering demands.
Hypothetically, this could be explained by the fact that residents were forced to take control of patient care and deal with clinical workload in face of poor supervision from the surgeons in charge. These observations deserve further investigation.
We acknowledge that the study has both strengths and limitations. A strength is that it adds information to the available evidence (conducted in non-healthcare settings from non-educational perspectives) supporting the positive relationship between transformational leadership and job crafting (28,49,50). Moreover, adding a qualitative stage provided more depth and was useful since few studies have focused on the qualitative dimensions of a TLS (23,51). A rst limitation is that the data we collected only represented residents' perspective and, consequently, the study lacks a supervisor perspective. Secondly, we did not study the role of moderators in the relationship between supervisors' leadership and job crafting. Possible moderators are the organizational culture at the level of departments and institutions, as well as residents' attributes (e.g., self-e cacy and grit).
This study has implications for practice and research. It is essential that supervisors become transformational leaders, as it will help residents to become skilled job crafters. As suggested by our results, a TLS in the surgical context typi es supervisors who reveal welldeveloped personal strengths (e.g., integrity and trustworthiness) and commitment to high standards of patient care. They contribute to a positive atmosphere for training in which they offer residents support aimed to strengthen their performance, motivation, and readiness for practice. Finally, they serve as role models for residents, by demonstrating effective behaviors for handling the demands of the work environment. We believe that organizations committed to strengthening a healthy workforce should implement formal training not only in this type of leadership for supervisors, but also in bottom-up strategies for residents to teach them how to optimize job demands and resources (e.g., job-crafting training as part of residency training) (12,20). Moreover, investing in transformational leadership development for supervisors could help strengthen residents' job t in surgical training and reduce burnout and dropout, as has been identi ed in other work contexts (38,52). Considering the limitations, we call for studies into the organizational in uences (e.g., culture and power) on the relationship between supervisors' leadership and residents' job crafting. Moreover, supervisors' perspectives on this topic deserve investigation. Similarly, we invite future studies to explore the effect of transformational leadership training and development for supervisors on residents' job crafting.

Conclusion
In conclusion, a TLS of supervisors in surgery is related to the extent to which residents are able to optimize the job demands and resources for training in order to gain control of the work environment. This relationship is rooted in the positive in uence of that leadership style on the environment for training, on role modeling, and on resources for the resident. A.S. contributed to the conception of the study, collection and analysis of data, funding acquisition, investigation, methodology, supervision, writing of the original manuscript, and to the revision and editing of the nal version.
L.S. contributed to the conception of the study, analysis of data, investigation, methodology, supervision, writing of the original manuscript, and to the revision and editing of the nal version.
All authors have read and approved the manuscript Competing interests: The authors declare that they have no competing interests Funding: Universidad de la Sabana (Colombia). The funder had no role in the study design, data collection, analysis, interpretation of data or in writing the manuscript.
Availability of data and materials: We as authors declare our commitment to provide all data of the present study by request.