Interactions between lean management and the psychosocial work environment in a hospital setting – a multi-method study

Lean Management, Lean Thinking or Japanese Production Management (below called Lean) was introduced by Toyota to improve both efficiency and quality on production lines. This was done by identifying quality problems early and eliminating wasteful aspects like unproductive waiting time and unnecessarily large stocks [1]. Following this, Lean has been widely adopted across the world in manufacturing and later also in health care [2‐4]. When the effects of Lean have been researched, the focus has often been on process outcomes [3]. However, since Lean often involves changes in fundamental aspects of work such as work process and work structure [3], Lean is also likely to affect the working conditions of the employees [5‐7]. In line with this, Lean has been criticized as being a way in which management increases the demands on the employees [8]. However, in research on psychosocial work aspects in relation to Lean to date, both positive and negative impacts have been seen [8]. For example, when Lean leads to downsizing [9], the effects are likely to be very different from those expected from an integrated Lean approach in line with the broader socio-technical environment [10]. The negative associations between Lean and psychosocial work environments have been reported to be particularly pronounced in the manufacturing industry, while a mix of positive and negative correlations have been found in other settings including health care [11]. With similar findings, a recent review suggested a move from a cause-and-effect model to a more comprehensive model that sees Lean as an open and ambiguous concept which may have either positive or negative effects, or even both, depending on the actual Lean intervention implementation [8]. That review also states that research into the effect of Lean on the working environment is still limited outside of manufacturing industry, and suggests that the effects of Lean should be studied not only based on the concept itself but also on the Lean practice and the context into which it is introduced. In line with this, one large scale industrial study [5] reported that the relationship between Lean and stress differed depending on how the psychosocial working conditions had changed as a result of Lean. In that study, control over work speed and work flow, employee involvement and influence on the change process, team-based organization and management support were related to reduced stress while an increase in work speed, removal of resources, increased working hours and employees’ feeling blamed for defects were related to increased stress levels. Another study showed that the active participation of employees in change processes was strongly associated with psychosocial aspects related to job control (influence, possibilities for development, meaning of work, freedom at work and commitment), social support (supervisors’ and colleagues’ support and quality of leadership, sense of community, role conflict and role clarity) and rewards (esteem) [12]. In fact, an employee-centered Lean intervention implemented with good employee participation has been considered as the most important means of preventing negative effects on the psychosocial work environment [3, 8, 12]. In summary, further research is needed into the interaction between work organization interventions such as Lean and the psychosocial work environment, in order to improve such interventions’ design so that the psychosocial work environment is improved [6]. This research is particularly needed in health care, given the steady growth of Lean application in this area [13].

Table 4 describes our predictions for changes in all the scales in all three settings along with motivations behind it. The results of regression analysis of expected versus actual changes were found to be significant in all three units (ED: F(1,17) = 5.88; p = .03; Beta = .507; Ward-II: F(1,17) = 7.68; p = .01; Beta = .558; Ward-I: F(1,17) = 5.50; p = .03; Beta = .494). Table 5 shows the COSPSOQ scales scores for T1 and T2 in three settings.

Since this study included one unit that successfully implemented Lean during the study period (Ward-I), one where previously implemented Lean changes deteriorated (the ED) and one which partially implemented Lean during the study period (Ward-II), we performed a comparison between the patterns for the three units (Additional file 1). Ward-I and the ED showed changes in different directions in 13 out of 19 scales; and Ward-II and the ED differed in 14 of the 19 scales. In contrast, Ward-I and Ward-II showed changes in the same direction in 12 of the 19 scales. Looking only at the domains ‘Work organization and job content’ and ‘Interpersonal relations and leadership’, as 5 out of the 8 scales expected beforehand to be most responsive to Lean were included in these two domains, 8 out of 11 scale changes were in the opposite direction when comparing ED and Ward-I and 10 out of 11 scale changes were in the opposite direction when comparing ED and Ward-II. Again, comparing Ward-I and Ward-II showed similar patterns: 9 out of 11 scale changes were in the same direction. In line with the Lean intervention outcomes in the three units, the results for ED are mostly contrary to the results for Ward-I while the results for Ward-II are mostly similar with Ward-I. This comparison is most pronounced for the domains ‘Work Organization and job content’ and ‘Interpersonal relations and leadership’. No clear pattern could be detected under the domain ‘Demands at work’ and ‘Values at the workplace’. Only one scale, ‘Horizontal trust’, changed contrary to the Lean intervention outcome, i.e. it deteriorated on Ward-I and Ward-II and improved at ED. Therefore, the overall change patterns for the three units are in line with the Lean intervention outcomes.

Comparing the expected and the actual changes, ten out of 19 actual changes were same as predicted for ED, 13 out of 19 actual changes were as predicted for Ward-I, and ten out of 19 actual changes were as predicted for Ward-II. Looking only at the domains ‘Work organization and job content’ and ‘Interpersonal relations and leadership’, as 5 out of the 8 scales deemed as most responsive to Lean were included in these, eight out of eleven actual changes were as predicted for ED, nine out of eleven actual changes were as predicted for Ward-I and only four out of eleven actual changes were as predicted for Ward-II. Looking at the results of four domains for all three units, nine out of 15 actual changes were as predicted for ‘Demands at work’, eleven out of 15 actual changes were as predicted for ‘Work organization and job content’, ten out of 18 actual changes were as predicted for ‘Interpersonal relations and leadership’ and only two out of nine actual changes were in the same direction as predicted for ‘Values at the workplace’. For Ward-I, there was only one scale (‘Justice and respect’) which was expected to increase but decreased and two scales (‘Tempo at work’ and ‘Role Conflicts’) which were expected to decrease but increased. There were four scales which were expected not to change, of which two decreased (‘Horizontal trust’ and ‘Vertical trust’) while the other two were almost unchanged. Importantly, there were only two scales (‘Possibilities for development’ and ‘Role Conflicts’) under the domains ‘Work organization and job content’ and ‘Interpersonal relations and leadership’ which did not change as expected but all the other scales changed as expected. For Ward-II and the ED, almost half of the scales changed as expected with varying domain wise results. With regards to the four domains, the ‘Work organization and job content’ domain showed maximum agreement between the expected and the actual changes. The ‘Interpersonal relations and leadership’ and the ‘Demands at work’ domains also showed good agreement but the quantity of change was higher in ‘Interpersonal relations and leadership’ than ‘Demands at work’. The ‘Quantitative demands’ and ‘Cognitive demands’ scales under the domain ‘Demands at work’, which were among the scales most relevant to Lean, changed as expected in all three units. The domain ‘Values at the workplace’ showed the least level of agreement and the scales are mostly not in line with what was expected. In summary, we can see the maximum agreement between the expected and the actual changes for Ward-I while for Ward-II and at ED, the agreement was medium.Focusing particularly on those scales that were considered most relevant to Lean implementation before analysis, all the scales except the ‘Vertical Trust’ changed as expected for the ED and the Ward-I; however five out of eight scales changed as expected for Ward-II. The ‘Vertical Trust’ was the only scale which didn’t change as expected for all the three settings. Based on these results, Figure 2 shows that the Ward-I and Ward-II graphs are partly similar in pattern; however, the ED graph pattern is almost opposite in direction. The ‘Cognitive demands’ scale value at the Ward-II decreased, although a little, similar to the ED than the Ward-I. The ‘Vertical Trust’ scale value at the Ward-I decreased, similar to the ED than the Ward-II. With regards to the expected versus actual changes for these scales, most of the scales changed as expected. Only one scale (‘Vertical trust’) did not change as expected at any unit. Looking unit-wise at Ward-I and ED, all the scales changed as expected except ‘Vertical trust’. For Ward-II, only 4 out of the 8 scales changed as expected.

In this study, we aimed to investigate how implementing Lean in health care affects the psychosocial work environment. When we contrasted the relationship between Lean and the psychosocial work environment in three units which differed in their implementation of Lean, differing patterns emerged. This was particularly pronounced for the psychosocial work environment factors that were considered most relevant to a Lean intervention. The results suggest that when Lean was implemented (more) successfully, it was related to improvements in the domains ‘Work organization and job content’ and ‘Interpersonal relations and leadership’. Concurrently, in the unit where Lean deteriorated, the psychosocial work environment also deteriorated, while the unit that implemented Lean to a limited extent showed some limited improvements in the psychosocial work environment. In summary, this suggests that Lean, if implemented successfully, may be positively related to psychosocial work conditions.

Ward-I showed improvements in three of the four domains of the psychosocial work environment. The greatest improvements were seen in the domains ‘Work organization and job content’ and ‘Interpersonal relations and leadership’. This is in line with a previous study [5, 12] relating improvements in the same domains to active employee participation in Lean implementation, a supportive leadership and regular morning meetings, all of which were also the case on Ward-I. Overall, the ED showed deterioration in psychosocial work factors. This deterioration was more prominent in the ‘Work organization and job content’ and ‘Interpersonal relations and leadership’ domains. This seems to be in line with the deterioration of Lean activities between T1 and T2. For example, there were less frequent morning meetings and decreased employee participation at the ED which may have resulted in the deterioration of scales related to job control and social support [12]. Ward-II, like Ward-I, showed major improvements in ‘Work organization and job content’ and ‘Interpersonal relations and leadership’. The concurrent deterioration in both ‘Role conflicts’ and ‘Social support’ from supervisors and colleagues may be related to the fact that the Lean-inspired ‘Problem Solving’ activity failed at this unit and the social support among the employees could not be promoted [26].

Due to the influence of multiple, concurrent changes (‘confounders’) and the heterogeneity in the Lean applications, a causal relationship between Lean and psychosocial work factors is difficult to establish. This has drawn the attention of researchers towards studying the changes in psychosocial work factors in the light of how and in which context Lean is implemented [8]. In the present study, we strove to take an approach inspired by realistic evaluation methods [27] by incorporating such information into the analysis. We did this by using a synthesis of the Lean literature [1, 5, 8, 23‐25] about the psychosocial work environment, qualitative data in context and Lean implementations to predict expected changes in psychosocial factors. A significant regression analysis between expected and actual results in all three units showed that the Lean literature synthesis and the qualitative information we had about the units were very helpful in foreseeing the actual pattern of changes. Some of the scales, which were predicted to be most relevant to Lean, implementation also behaved quite in line with Lean intervention results. However, the ‘Vertical trust’ scale was an exception which behaved oppositely. Even if this is not in line with what was predicted, it does seem to be in line with the leadership changes which occurred in the units. For example, the leadership change on Ward-II before T2, which was welcomed by the staff, could have caused an increase in ‘Vertical trust’. This could be one way of predicting change patterns in the psychosocial work environment caused by Lean intervention that are testable with a combination of quantitative and qualitative data. It shows the usefulness of combining theory and data to understand intervention impacts in a complex and changing reality.

In summary, the findings from the three settings are in line with previous research and suggest that Lean may be related to improved psychosocial work conditions, particularly when a high degree of participation is a central part [5, 7, 12]. Particularly, it adds to the literature by showing that issues relating to work organization, job content, interpersonal relations and leadership seem to be related to changes in Lean implementation. However, the findings from this study do not show any consistent pattern in how Lean is related to ‘Demands at work’ and ‘Values at the workplace’.

The content of a Lean intervention, its implementation and the context in which it is implemented may have a collective effect on the psychosocial work environment which is very difficult to capture. We tried to track a Lean footprint on the psychosocial work environment by combining longitudinal data on psychosocial work factors, multi-source data on Lean implementation and context. At the very least, we were able to capture a clear contrast between the footprints on Ward-I and ED which had opposite Lean implementation processes. Thus, although the design of this study does not allow causal inferences, the study does suggest that Lean interventions may be related to changes in the psychosocial work environment; positively when Lean is fully implemented and negatively when Lean changes deteriorate. Specifically, this study adds to the literature by suggesting that Lean implementation may be particularly relevant in relation to the ‘Work organization and job content’ and ‘Interpersonal relations and leadership’ domains of the psychosocial work environment. Also, health care managers should note that a failed implementation may actually worsen the work environment.

Waqar Ulhassan, PhD is a researcher at Medical Management Center, Karolinska Institute. His research interests focus on the implications of quality improvement programs on health care employees’ working and well-being, in particular teamwork and psychosocial work environment.

Ulrica von Thiele Schwarz is a psychologist and an associate professor in psychology. She is currently working at Medical Management Centre, Karolinska Institutet and the Department of Psychology, Stockholm University. Her main research theme is applying a behavioral perspective to interventions and continuous improvement within the work setting. Overall, the research is concerned with further the understanding on effective management of change in the workplace setting, and links between work conditions, health productivity and quality.

Johan Thor, MD, MPH, PhD, Director and lecturer; the Jönköping Academy for Improvement of Health and Welfare, Jönköping University (Sweden); research associate, the Medical Management Centre, Karolinska Institutet. Dr. Thor has devoted his career to promoting, and studying, healthcare improvement, covering process improvement, lean thinking, patient involvement, evidence-based medicine, and patient safety.

Hugo Westerlund is Professor of Epidemiology at the Stress Research Institute, Stockholm University. One of his main research interests is the impact of work, work organization, and work environment on the health and well-being of the employees. He mainly works with large cohort studies but has also conducted qualitative research.