Assessment methodology for Lean Practices in healthcare organizations: case study in a Brazilian public hospital

Paper aims: This article aims at proposing a methodology for assessing lean practices in healthcare organizations that are undergoing a lean implementation. Originality: The proposed methodology evaluates healthcare organizations and help managers identify the relationships intensities between their problems and lean practices. This application occurs in a case study in an Emergency Department from a Brazilian public hospital. Research method: The proposed methodology provides a theoretical framework that addresses lean practices according to fundamental aspects of the organization, such as people, partnerships, results, processes and leadership, corroborating to a systemic view of lean implementation and facilitating its long-term sustainability. Main findings: Our findings provide guidelines for the public hospital leadership, allowing a clear focus on the improvement efforts that can mitigate problems. Implications for theory and practice: Due to contextual diversity, the understanding of the critical success factors for lean implementation under such conditions deserves attention. Hence, the proposed method allows practitioners and academics to better comprehend the lean implementation within healthcare organizations. Such enhanced comprehension allows to anticipate potential issues throughout the lean implementation, entailing more assertive efforts. Moreover, accounting for specific problems when designing the lean implementation avoids the utilization of a general and ineffective approach.


Lean healthcare practices
In healthcare, as in the manufacturing industry, LP need to be adapted to the specificities of each service (Drotz & Poksinska, 2014). For Cookson et al. (2011), the application of LP in healthcare is becoming increasingly common for the maintenance of an efficient service of high quality from the perspective of patients and employees. Moreover, if analyzed under the context of emerging countries, the opportunities for research are even greater (Souza, 2009). Daultani et al. (2015) affirmed that methods that address the selection process of the most appropriate LP for healthcare organizations are still lacking, which entails that this selection occurs through speculation and intuition. Fillingham (2007) reported that, through the use of 5S, kaizen events, value stream mapping, standardized work, pull systems and continuous flow, Bolton Hospitals in UK achieved significant improvements. Kim et al. (2006) described the case of the University of Michigan hospital where processes were redesigned, reducing and standardizing the setup activities.
In this sense, the selection of appropriate practices for the improvement of processes and identification of their applicability in the operational context represent an additional problem for managers and professionals (Shah & Ward, 2007). There is a large number of LP that must be applied simultaneously in order to make a leaner organization (Tréville & Antonakis, 2006). Regardless the fact that LP have been used for years, there is little evidence of generalizable steps of lean implementation (Marodin & Saurin, 2013;Tortorella et al., 2015). Pavnaskar et al. (2003) complemented that the main reasons for lean implementation failures can be associated with misinterpretation in the intensity and way LP are adopted.
From an analysis of the literature on LH, we have listed some of the most cited LP in healthcare as shown in Table 1. Among these, 'value stream mapping', 'visual management' and 'operation standardization' appear in Table 1. Citation frequence of LP in the literature.

Method
The proposed methodology, comprised of six steps, is shown in Figure 1. The first step comprises the construction of the relationship matrix between the 22 LP (Table 1) and the 5 CSF (Table 2) using the Delphi method. To this end, the selection of experts was comprised of two criteria: (i) minimum of 10 years of practical and academic experience in lean implementation; and (ii) participation as authors in specific journals of the 10 out of 26 research papers. According to Augusto & Tortorella (2017), these practices require few adaptations when used in healthcare when compared to manufacturing environments and, therefore, this would be one of the reasons for their wider dissemination in the literature. On the other hand, practices 'total productive maintenance', 'autonomation', 'workforce empowerment' and 'small batch size' presented the lowest frequency of citation.

CSF for lean implementation
Several studies were carried out with the purpose of listing the CSF for lean implementation. Liker (2004) describes 14 principles that guide the Toyota Production System and divides them into four categories: (i) philosophy; (ii) processes; (iii) employees and partners; and (iv) problem solving. Poksinska (2010) lists five problems encountered when starting LH implementation: employee resistance, lack of trained employees, lack of a clear customer focus, highly verticalized hierarchical structure and exclusively departmental improvement focus. In a complementary way, Young & McClean (2008) identified three critical problems that impair the expansion of the LP application in healthcare: (i) uncertainty level inherent to healthcare (e.g. difficulty in demand predictability and determination of process times); (ii) defining activities that add value; and (iii) conflicting and disconnected metrics.
In this sense, SAE J4000 standard, which comprises 52 items, is an instrument that aims to identify and measure best practices in a lean implementation (Society of Automotive Engineers, 1999).  and Vinodh & Chintha (2011) used the AHP (Analytic Hierarchy Process) to evaluate a lean implementation based on a review of the literature and experts' opinions from manufacturing organizations. Achanga et al. (2006) identified four CSF for implementing LM in small and medium-sized enterprises; they are: (i) leadership and management; (ii) finance; (iii) skills and knowledge; and (iv) organizational culture. The model proposed by Dahlgaard et al. (2011) aims to evaluate healthcare organizations with regards to their problems and results. The model is composed of five criteria: (i) leadership; (ii) people management; (iii) partnerships and resources; (iv) processes; and (v) results. Finally, the evaluation of Shingo Prize (2014) is composed by 10 principles grouped into four dimensions: (i) cultural enablers; (ii) continuous improvement process; (iii) organization alignment; and (iv) results.
Based on the studied authors in Table 2, we consolidate the main CSF to be reinforced throughout the lean implementation. In total, five critical factors were consolidated: (i) people; (ii) partnerships/resources; (iii) services/results; (iv) processes and (v) leadership. Such consolidation will serve as a basis for establishing the relationship between the LP and CSF. correlated areas. Initially, 18 (eighteen) experts were invited to participate in the study, of which only 6 (six) effectively participated in all steps due to agenda availability. The definition of the number of rounds to be carried out depended on obtaining an acceptable consensus level. The consensus was measured based on the variation coefficient. According to Giovinazzo (2001), an acceptable value to denote consensus must be lower than 30% for this variation coefficient. Since 110 relationships were analyzed, the final consensus was calculated as the number of items with variation coefficient below 30% divided by the total number of items. A value greater than 90% was adopted as an acceptable final consensus. Consequently, two rounds of the method were performed, making it possible to reach a final consensus of 91%.
Subsequently, an initial evaluation of the relationship intensities between each pair of LP and CSF was requested by e-mail to the experts, accompanied by an explanatory letter clarifying the importance of participation in the research, the objectives of the research and the proposed model. In this round, experts should answer the following question: "for lean implementation in healthcare, how does practice i impact the critical factor j?". It is noteworthy that at the beginning of each email, a brief explanation providing a uniform definition of each one of the LH practices and critical success factors was undertaken. According to Kothari (2004), such explanation may avoid misinterpretations among experts, mitigating erroneous values and responses. The evaluation was measured based on a scale from 1 to 9, where 1 denotes a 'non-existent' impact and 9 an 'absolutely important' . The experts' responses were returned by e-mail and consolidated. In the second round, the same experts had the information on the absolute frequency distribution of the answers sent by the others, as well as their own answers, leading to a reflection on his answers. The comments for each question were also made available for all experts, without authors' identification.
From the experts' answers, a structure called relationship matrix (M), schematized in Figure 2, was established, which shows the relationship intensity r ij rescaled into an interval from 0 to 1 for each pair of LP and CSF at the intersections between lines and columns. The results of matrix M are measured through an importance index for LP (ip i ), which represents the relevance of the practice for the implementation of LH, considering the sum of the relationship intensities of relationship of the practice for each CSF by means of the following expression:  In step (ii), there is the identification of the occurrence frequency of problems in the healthcare organization through the application of a questionnaire. The proposed questionnaire is an adaptation of the method developed by Dahlgaard et al. (2011), which lists 50 issues grouped according to each CSF (see Appendix A). The questionnaire was answered by employees with at least 6 months of experience in the organization under study. For each CSF there are 10 questions sequentially associated and evaluated through a Likert scale from 1 to 5, where 1 designates a situation that almost never occurs and 5 a situation that almost always occurs. Based on the answers, a mean value of frequency is established for each question and these are rescaled into an interval from 0 to 1, represented by f k (k =1,...,50). Due to the way that the questions were elaborated, a high value of f k reinforces a desirable condition. Consequently, the occurrence frequency of problems is given by fp k through Equation 2. Therefore, the occurrence frequency of problems in each of the five CSF, represented by p j (j =1,...,5), is given by the mean of the respective fp k associated. (2) For surveying the implementation level of the 22 LP in the healthcare organization, a second questionnaire was applied with the same employees of the department under study. Before the beginning of each interview, a brief explanation providing a uniform definition of each one of the LP was undertaken. Respondents were invited to assess the implementation level of LP based on a Likert scale from 1 to 5, where 1 denotes a practice that is almost never applied and 5 a practice that is fully adopted. Based on the answers, an average level of implementation is established for each LP and these are rescaled into a 0 to 1 range, represented by n i (i=1,...,22). The gap for the full adoption of each LP in the healthcare organization, represented by g i , is given by the following expression: At step (iv) the values of p j and g i are used for determining the criticality indices of LP, whose values are inserted in matrix M. The degree of criticality of LP integrates both theoretical and contextual relevance for the lean implementation. The establishment of the criticality indices provides managers of the healthcare organization basis for improving their processes. Thus, the criticality index c i is given by the following expression: High positive values of this index indicate the most important and critical LP. For the purpose of this work, the value of 1.0 standard deviation above the average was considered as a threshold value to denote the relevance of the practice (Tortorella et al., 2017b).
Step (v) corresponds to the multicriteria analysis to prioritize the LP implementation efforts. The evaluation of the LP to be prioritized in the healthcare organization consists in investigating its characteristics under real conditions of time, space and information, and according to available resources. Therefore, a proper decision process under this scenario should be supported by multicriteria decision making methods. At this step, senior leaders of the healthcare organization are asked to participate in the decision-making process by establishing criteria and attributes that allow the company to take into account its cultural and organizational characteristics and urgency for improvement.
To evaluate the distinctions between the attributes, a Multiattribute Utility (MAUT) was used to consider opportunities for improvement. The first criterion assesses LP in relation to the importance of implementation and is subdivided into three main attributes (Tortorella & Fogliatto, 2014): (i) achievement of strategic goals; (ii) behavioral construction consistent with organizational values; and (iii) impact on organizational structure. The second criterion evaluates LP in relation to the implementation effort, and consists of three attributes: (i) technical risk; (ii) the need for training; and (iii) the need for investment. From the multi-attribute structure, the organization's managers are asked to assign importance to the attributes and criteria, on a scale from 0 to 100, with 100 being the maximum importance value. The declared amounts are then rewritten in percentage terms, generating the weights (pa m , m=1,...,6).
Then, to prioritize LP implementation efforts a prioritization matrix (Z) was used, as shown in Figure 3. The LP are listed in the lines with their respective criticality values (c i ). In the columns of the matrix there are the attributes for prioritizing LP, which were previously identified and weighted. Leaders were asked to determine the degree of relationship of each LP with the attributes, expressed by pd im (i=1,...,22; m=1,...,6). Thus, each company leader was asked the following questions: (i) for the attributes related to the criterion importance for organization, "what is the importance of practice i in the accomplishment of the attribute m?". The evaluation is performed through a scale of three values: 9 (strong relationship), 3 (moderate relationship) and 1 (weak relationship); and (ii) for the attributes related to the criterion effort for implementation, "what is the impact of the practice i on the attribute m?". Because the attributes of this criterion reinforce an undesirable aspect (e.g. greater technical risk, more training or higher investment) in the organization, and to maintain consistency in the attribute rating scale (higher-is-better), a high value is a favorable condition. Thus, specifically for these attributes, the evaluation is carried out by means of a scale of three values, being: 9 (low impact), 3 (moderate impact) and 1 (high impact). The absence of impact is represented by the value 0. Thus, the implementation priority (z i ) is given by: The last step (vi) is the ranking of practices for LH implementation, whose values of z i were ordered. To help visualize the prioritized LP and facilitate decision making, it is recommended to use graphical tools, such as the Pareto chart.

Emergency Department (ED) of a Brazilian public hospital
The proposed methodology was applied in the Emergency Department (ED) of a Brazilian public hospital integrated into the healthcare network of the SUS. In this sense, there are great budgetary constraints regarding the implementation of new processes and technologies. The primary ambulatory care of the hospital is structured in four basic areas: medical, surgical, pediatrics and gynecology. These areas attend 250 patients per day in average and are supported by a determined infrastructure (Table 3). Historically, few formal LH implementation initiatives have been demonstrated in the hospital. However, some of the LP practices are developed in isolation even without their in-depth knowledge. Among them, the practices of poka-yoke, which is applied to differentiate materials that can generate risks to the employees, and jidoka, stand out, since most of the equipment is programmed with devices to stop in case of any problem. Overall, LH is still poorly understood and widespread among emergency and hospital staff.
In particular, the study was restricted to the context of the ED, due to the greater flow of patients (approximately 250 patients per day) and the range of specialties with which the department operates. The function of the ED is to provide treatment to those in need of urgent medical intervention, aiming to stabilize the patient and reach a minimally satisfactory level of care, so as to be able to safely address the patient to the next step in the process. ED's team is comprised of 21 doctors, 18 nurses and 71 technicians and nursing residents, totaling 110 employees, which are divided into three working shifts, being the first shift from 7:00 a.m. to 1:00 p.m., the second from 1:00 p.m. to 7:00 p.m. and the third from 7:00 p.m. to 7:00 a.m.
When patients arrive at ED, they undergo a first risk classification. The most critical cases are denoted as 'red-emergency' cases and receive immediate care. In these cases, family members are responsible for completing the initial attendance sheet. For other patients, it is necessary to fill out this form at the reception. After this step, the patient is directed to a risk assessment, in which a nurse performs an evaluation, collects his/her vital data and classifies it according to the urgency of medical care, following the Vancouver protocol. The existing classifications for prioritizing medical care are: orange (very urgent), yellow (urgent), green (not urgent) and blue (non-urgent). After the medical care has been performed, the patient can be directed to beds or other rooms according to the medical need, or even be dispensed.

Identify the occurrence frequency of problems
Thus, it was sought to identify the occurrence frequency of problems and the implementation level of LP in this context. In this sense, different employees from different hierarchical levels of the ED were invited to answer the two questionnaires described in section 3. Data collection was performed over a period of two months. The sample of respondents comprises a total of 30 individuals, which represents approximately 27% of the total employees in the department (see Table 4). Most of the respondents were technicians and nursing residents, who presented an average working time of 6 years in the ED. Nurses, in turn, correspond to 27% of the respondents and have an average working time of 3 years. Finally, physicians represent 13% of the sample, with 10 months as the average working time in the ED.  To identify the occurrence frequency of problems, each employee answered the 50 questions and, based on their answers, an average value of frequency was established for each question. The incidence frequency of problems in each of the five CSF, represented by p j (j=1,...,5), is given by the mean of the respective fp k associated. The higher the value obtained for p j , the greater will be the occurrence frequency of problems related to the respective CSF. Thus, it can be verified that problems related to CSF 2 (People Management) are the most frequent according to repondents' perception in the ED.

Relationship intensities among the CSF and LP
Regarding the implementation level of LP, each employee assigned an implementation score to the 22 LP and, based on their responses, an average implementation value was established for each LP. After that, values were rescaled into an interval from 0 to 1, obtaining values for n i and values for g i (i=1,...,22), which represent the gap for the full adoption of LP in the ED. From the data collected, practice LP 15 (Kanban) appears to have the highest potential (g 15 =0.68) for implementation in the department (see Appendix B).
The relationship intensities among the CSF and LP were consolidated and the importance index of each LP was obtained according to the maturity matrix M (see Table 5). According to established threshold value, five LP showed differentiation indexes above 1.0; they are: lp 7 (Kaizen), lp 8 (Problem solving methodology), lp 2 (Visual management), lp 11 (Crossfunctional teams) and lp 18 (Education/training). Then, based on the occurrence frequency of problems and implementation level of LP, the criticality values for each LP was determined. The higher the value obtained for c i , the more critical is the LP for the LH implementation in the ED. Similar to the importance index, practices whose differentiation indexes for criticality presented values above 1.0 were lp 7 , lp 8 , lp 2 and lp 15 .

LP assessment
It is worth noticing that practices lp 2 , lp 7 and lp 8 stand out both in terms of theoretical importance (experts' opinion) and in terms of contextual criticality for the ED. The department under study is composed by diverse and complex flows that need more efficient ways of monitoring. All patient information is attached to the patient's chart and there is no method to track patient status or other kind of information. Besides supporting a better flow visualization, practice lp 2 allows to streamline the reaction process (Liker, 2004;Fillingham, 2007), enabling a more immediate identification of anomalies occurrence. Thus, the adoption of this practice helps to minimize misinterpretations throughout the processes, making the flow of value more efficient and assertive.
Practice lp 7 consists of continuous improvement activities inherent to lean implementation. Specifically in healthcare, this practice is more commonly understood as short-term events driven by crossfunctional teams (kaizen events), through which problems are identified and the necessary improvements are implemented (Souza, 2009). Since no initiative related to this practice was identified in the department under study, the high adoption potential perceived by the employees (g 7 =0.57) is justified. Thus, this result corroborates with the studies developed by Dickson et al. (2009) and Bush (2007), which exemplify the LH implementation through kaizen events.
Similarly, lp 8 has a wide scope of application and can be adapted according to the severity and type of problems identified. Previous studies focused on the LM (e.g. Spear, 2004Spear, , 2009 indicate that the adoption level of this practice may denote a differentiation factor among organizations undergoing a lean implementation, since both the organization and its employees develop as it is implemented. Spear (2005) adds that LH seeks to make employees able to solve problems by teaching them how to deal with root causes of problems. In this sense, according to Mazzocato et al. (2010), this practice should involve leaders in order to provide guidance to employees and the creation of stable structures for solving problems. Although the ED presents small initiatives, the adoption of this practice does not involve the whole team; only the nursury participates in problem solving meetings and they are optional. Thus, during the ED's daily activities, the vast majority of problems are solved in a reactive and superficial manner, without the direct involvement of all responsible parties, and the root causes of the problems are not usually addressed.
On the other hand, not all practices deemed as important presented high criticality values. For instance, lp 11 (Crossfunctional teams) (ip 11 =4.7) presents as main benefit the utilization of knowledge and complementary perspectives of the individuals during the solution of problems and continuous improvement of the processes. Such complementarity allows holistic solutions to be obtained, avoiding punctual gains in the value stream. Zarbo et al. (2009) andPoksinska (2010) emphasized that organizing employees in crossfunctional teams helps to analyze the flow of patients and identify what does not add value, engaging them in problem solving and process improvement. However, these teams tend to work on improvement projects with limited duration, leading to the dismemberment of the team after the project ends (Laganga, 2011). As the potential for full adoption of this practice is low in the ED, since there are already initiatives prior to LH, the criticality of this practice has been deflated by its implementation level.
The lp 18 (Education/training) appears as the fifth most important practice (ip 18 =4.7) for LH implementation in general. This result corroborates with Poksinska (2010), which indicates that the first step in LH is to train employees on the principles, methods, and practices applied in lean implementation. Similarly, Ben-Tovim et al. (2008) argued that employees training is critical to lean implementation by providing them the basic concepts and principles for establishing a critical mass of change agents. However, following the example of lp 11 , the interviewed team realizes that this practice is already being disseminated in the ED, which reduces the criticality degree for the context under study. This perception can be attributed to the fact that the hospital is linked to the federal university, favoring the dissemination of knowledge and praising the need for continuous training of its employees.
A few practices were considered of minor importance for LH implementation, such as lp 15 (Kanban), lp 22 (Small batch size), lp 14 (Workplace design) and lp 16 (Setup reduction). Augusto & Tortorella (2017) observed that the literature evidence of lp 15 and lp 16 in healthcare is significantly lower than in manufacturing environments, indicating the low frequency of application of such practices in healthcare. In turn, lp 15 has a high criticality value, although it has not been indicated as one of the most important by the experts. Since the stock control methods identified in the ED are typically manual and informal, the high value found for this practice's gap for full adoption (g 15 =0.68) is consistent. In addition, Mazzocato et al. (2010) identified only one healthcare study that approached improvements in setup time. Regarding practices lp 22 and lp 14 , the few evidences found in the literature portray their use in a limited way, being only adopted in supporting activities, such as laboratories of clinical and pathological analysis (Zarbo & D'Angelo, 2007;Zarbo et al., 2009).

Analysis of the improvement opportunities
After presenting the improvement opportunities, the leadership of the hospital was interviewed in order to establish the importance of the evaluation attributes forming the decision profile for prioritization of the improvements. In this sense, the responses of ten senior management members were consolidated and the decision support structure indicated that the criterion 'importance of LP' is responsible for 45% of the decision, while 'effort for implementation' represents 55% of the decision. For the first criterion, the contribution of the attribute 'achievement of strategic goals' represents 36% of the importance of the criterion. Regarding the second criterion, the attributes 'technical risk' and 'need for investment' each represent 35% of the importance.
Then, leaders were invited to assess LP and determine the weights pd im for LP in relation to the m attributes. Both the head of nursing and medical were interviewed to establish consensus in the attribution of such weights. Once pd im weights were known and the weights of the attributes of the decision-making structure were determined, the final score of each LP for the ED was determined, as shown in the prioritization matrix Z in Table 6. In addition, Figure 4 organizes the practices in descending order of z k , allowing an immediate analysis of the improvement opportunities.
The results indicate that the practice lp 2 (Visual management) presents the highest priority value (z 2 =11.9). It is worth noticing that this practice also stands out in its values of importance and criticality. The leadership understands that its implementation has low technical risk and low investment need, hence justifying the use of visual management mechanisms to aid in the visualization of process flows and assertiveness. In addition, its adoption is characterized by low need for training, since it excels with simplicity of application and team's understanding. Although practice lp 21 (Workforce empowerment) has not been listed as important or critical, its value for implementation priority (z 21 =11.6) is the second largest among LP. This result can be particularly attributed to the fact that this practice has a high impact in the construction of a behavior consistent with hospital values. 3 3 3 9 9 3 1.2 11.9 lp 3 3 3 3 3 1 9 0.8 6.0 lp 4 3 3 0 1 9 0 0.8 4.1 lp 5 3 3 1 1 9 0 1.1 6.0 lp 6 1 3 0 9 9 9 1.1 11.4 lp 7 0 1 3 9 3 3 1.4 8.9 lp 8 0 3 0 0 3 0 1.3 2.5 lp 9 0 0 0 9 9 0 0.9 5.3 lp 10 3 3 3 0 0 0 1.2 3.6 lp 11 3 9 3 9 0 0 0.6 4.9 lp 12 9 9 3 9 9 3 0.8 11.2 lp 13 1 0 1 9 0 0 0.9 3.4 lp 14 3 0 3 1 9 1 0.8 4.3 lp 15 3 0 0 3 3 9 1.3 8.0 lp 16 9 3 1 3 3 3 0.9 6.8 lp 17 9 9 1 3 3 3 1.0 9.5 lp 18 9 9 1 9 1 3 0.8 8.8 lp 19 1 3 0 1 1 3 1.0 3.1 lp 20 9 3 0 3 3 1 0.9 5.9 lp 21 9 9 3 3 1 9 1.0 11.6 lp 22 0 1 0 3 0 0 0.9 1.2 The nurses and technicians are the professionals who hold the greatest amount of contact with the patient, which implies that their behavior and performance are closely linked to the patients' perception of value. Thus, an autonomous team allows greater efficiency in the execution of daily activities, without affecting the service level offered to patients. In addition, leaders understand that the investment needed to achieve the full adoption of this practice is relatively low and its application is widespread. Similarly, lp 6 (5S) does not have high values of importance or criticality, but was considered a priority mainly due to the low effort related to its adoption and potential for adapting to any area, such as equipment rooms and warehouse, which present a great diversity of critical items for the operation of the ER. Finally, lp 12 (Error-proof systems) appears as a priority due to its low technical risk, low investment need, high relation with the accomplishment of the strategic goals and the construction of a behavior consistent with the organization. This practice has easy adaptation and, through simple mechanisms, enables the reduction of the defect causes improving the quality of the services provided.
Considering the great complexity of the medical activity within the ED and the large number of interventions that a patient is subject to, there is a great risk of incidence of errors in these procedures, which can generate irreversible impacts on the patient.

Conclusion
The purpose of this study was to propose a methodology for assessing LP in healthcare organizations that are undergoing a lean implementation. Implications of this study are of considerable importance and relevance for both researchers and lean practitioners. The following sections describe in depth the contributions, and highlight the main study's limitations and future research opportunities.

Contributions to theory
In theoretical terms, the proposed methodology combines complementary concepts and methods aiming to evaluate healthcare organizations and help managers identify the relationships intensities between their problems and practices. Despite the number of studies about LH available in the literature (e.g. Miller, 2005;Kim et al., 2006;Ballé & Régnier, 2007;Fillingham, 2007;Mazzocato et al., 2010), there is still a lack of evidence regarding research containing methods to evaluate LP implementation in healthcare. Due to such scarcity, it is very difficult to assess whether an organization has actually implemented lean principles and practices and the level of its adoption in relation to other organizations, as pointed by Wan & Chen (2008). In this sense, the integration of the existing problems' occurrence, the adoption level of LP and the decision support criteria in the healthcare organization provides a theoretical framework that addresses LP according to fundamental aspects of the organization, such as: people, partnerships, results, processes and leadership. Further, this method allows a longitudinal assessment of the evolution throughout LH implementation, which is indicated as key by Bhasin (2011).
This framework consists of a normative theory for LH implementation, since it refers to the development of guidance on how actions will lead to desired results (Carlile & Christensen, 2004). The proposed guidance refers to the LP to be prioritized and implemented, taking into account peculiarities of the assessed system. While this prescription may appear obvious in hindsight, it seems to be typically neglected by both academics and practitioners. Hence, this framework allows LH implementation to be addressed in a broader way, corroborating to a systemic view of lean implementation and facilitating its long-term sustainability; which is a distinctive feature in comparison with other studies in the area.

Managerial implications
Some managerial contributions must be highlighted. Our results provide hospital managers and leaders a clear focus on the improvement efforts that can mitigate the frequency of problems and strengthen the desired LH culture. Further, since most LP were originally conceived in traditional manufacturing environments, their understanding and, hence, adaptation to healthcare context features an additional challenge. The justification for that is twofold. First, personnel training in healthcare sectors does not emphasize the development of managerial skills. These professionals usually come with technical backgrounds that focus on aspects related to healthcare task performance rather than on process analysis from a horizontal perspective. Second, despite the increasing pressure for better performance of public healthcare services, the difference in the mindset of managers in private and public sectors is still large. While in the private sector managers must improve and optimize resources to financially sustain their businesses, public sector operations are usually supported by governmental institutions, whose primary objectives do not necessarily involve attaining financial profit or business performance. In this sense, an assertive selection of improvement initiatives allows to maximize results while minimizes the required efforts and investments related to LH implementation.
Also, the research method and the proposed improvement opportunities may help practitioners to solve real life problems related to the implementation of lean in public healthcare organizations. For the healthcare organization under study, our research worked as a basis for continuous improvement of the lean implementation process. For other organizations or departments, especially those sharing a similar context, the approach provides comparative parameters either in terms of problems occurrence or LP adoption level. However, it is worth to notice that, depending on the management decision criteria, the LP priorities indicated in our study are not valid. Therefore, further adaptations might be needed in order to replicate the proposed method.
Finally, by considering the proposed method to plan their LH implementation, managers can customize their design and prioritize improvement efforts that would actually mitigate relevant problems to the department/organization. Such approach allows a more assertive implementation which provides an increase in change acceptance and confidence level. This fact also enhances personnel engagement and buy-in, since a LH implementation that effectively addresses problems observed by most of the employees is more likely to be supported and, hence, sustained.

Limitations and future research
Some drawbacks should be highlighted. First, using a single case study (ED of a public hospital) for testing theory fundamentals could pose as a limitation to the validity and generalization of our results. To cope with that limitation, we follow a robust and structured research method that could be replicable to other case studies. That level of detail allows the proposed method to be carried on within organizations presenting similar background to the one addressed in this study, such as other healthcare organizations (public or private). As we stated before, the research method is also a contribution of our study, as it highlights priorities of improvements from a holistic point of view. We acknowledge that improvement opportunities should vary between different organizations, as lean is heavily dependent on context, but some opportunities may be very similar in other organizations. Nevertheless, since the encompassed improvement practices solely derive from LH approach, other improvement methods and management techniques available in the Industrial Engineering/Business Management spectrum are not considered. Such methods and techniques may either complement or even address issues that the single application of LH practices cannot. Thus, the alignment of the proposed method with other relevant methods and techniques characterizes another future research opportunity.
Further, implementing LH reinforces the need for re-orientation of public services regarding scope and strategic positioning. The evolutionary process of transferring the lean team-based approach to public healthcare organizations may enable that re-orientation. However, existing practices discourage the establishment of a transparency culture, where abnormalities and problems could be identified by anyone. Since the methodology is dependent on the opinion of employees involved in data collection, that fact may limit their perception with respect to problems occurrence and LP adoption level. Additionally, the observed results may be impacted due to the variation of understanding of LP and sensitivity to existing problems among employees. Thus, future research should address the underlying variances to respondents' profiles through the inclusion of more sophisticated statistical methods, allowing for more robust and reliable investigation and guidance for LH implementation. Further, these studies need to better understand the barriers and policies within different public sectors, enabling a proper adaptation of lean principles and maximize their acceptance by public healthcare organizations. It is worth noticing that an increase on the sample size would also benefit the research outcomes, allowing the establishment of more refined statistical analysis on the data (e.g. multivariate data analysis) and empirical validation of our findings.