Process mapping in healthcare: a systematic review

Given there was limited literature available on the use of PM and little practical guidance on its use in healthcare, we recognised the need to draw on evidence from other relevant fields such as manufacturing and other service industries. We conducted a snowballing review of methodological literature on PM in both healthcare and non-healthcare settings (Fig. 1, online supplementary appendix 1). The objective was to identify recommended good practice methods for conducting PM [46]. We identified the most cited studies by searching online databases (Google Scholar, Scopus, Medline, Embase) with keywords derived from our research questions. We then screened citations and reference lists from these sources and included relevant studies. Given the high number of irrelevant articles derived from the online database search, the snowballing technique was very useful to identify the few available methodological studies on PM, as it allowed us to find grey literature, that might be missed by conventional online search methods. We then assessed methodological guidelines (online supplementary appendix 1) and empirical literature selected in the systematic review (2.2, Fig. 1) and through a process of inductive and deductive analysis we developed a conceptual framework identifying overarching quality criteria for each phase of the PM process (3.2, Fig. 2). These criteria were discussed by all authors and selected if they could be applied to a wide range of PM approaches and QI project types. Iterative development of the framework continued as new ideas emerged through discussions and feedback from experts and practitioners.

A systematic literature review of empirical research reporting on the use of PM in healthcare was performed following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) standards [49] and registered on PROSPERO (ID: CRD42017082140).

Six studies were identified in the snowballing literature review of methodological publications on PM in healthcare and other service and manufacturing industries to develop the conceptual framework (online supplementary appendix 1). The conceptual framework (Fig. 2) described below provides quality criteria for each of the five phases characterizing the process of PM taken from the methodological literature.

The service family under analysis and those who will use the process/service (e.g. patients/services users/customers etc.) should be clearly identified and representatives from these groups should be involved in the project. It’s also important that participants have the right skills required to participate in the PM exercise, which might vary depending on the type of project, modelling language and methods used for the analysis. Training should be provided to PM participants to fill skills gaps, if needed. For example, for simple PM exercises this might include a quick introduction to the PM method, while for PM exercises involving more sophisticated approaches (e.g. Six Sigma) and/or more structured modelling languages (e.g. Business Process Modelling Notation), a more technical intensive training might be appropriate.

Information should be gathered to inform the PM exercise. In addition to multi-disciplinary meetings, data can be collected using different approaches, such as direct observations, interviews, self-reported patient experiences, analysis of electronic health records or other relevant databases, literature or document analysis.

Different perspectives should be gathered by people having diverse roles in the process, each bringing their view and knowledge of the process under analysis.

The process map should be analysed to identify gaps in the systems and opportunities for improvement. The final process map should be checked for accuracy and validated by key stakeholders/experts. During the analysis phase it’s good practice to annotate the process map with information derived from the analysis (e.g. activity durations, resources involved) and transfer paper-based maps in an electronic format. Having a tidy electronic version of the process map supports the analysis and the documentation of the PM exercise and it’s also useful to disseminate and share the map with interested parties or those involved in the process for comments and validation.

Process maps should be used to guide process improvement initiatives. Improvement ideas and actions generated throughout the PM exercise should be implemented to improve current systems and practice.

The study selection process for the systematic review of empirical studies using PM is reported in Fig. 3. A total of 105 articles met the inclusion criteria and were included in the review (online supplementary appendix 3). Study quality was moderate-high with 31% study scoring 10/10, 43% scoring 9/10, 20% scoring 8/10 and 6% scoring 6–7/10 (online supplementary appendix 4). 86% were published in or after 2010 and 65% were conducted in the USA and UK (online supplementary appendix 5).

PM exercises were reported from a wide range of healthcare settings including, in-patient services (32%), multiple settings (29%), outpatient (11%), A&E (8%), community care (6%), care provided in other settings (5%), primary care (5%), prevention and health promotion (3%) and laboratory services (3%). The most common type of projects reporting the use of PM in healthcare were process improvement/QI initiatives (68%), which include the use of FMEA/FMECA (11%) and Lean and Six Sigma (12%) approaches. Use of PM in health information technologies (HIT) projects was reported in 10% of studies. A few studies outline its use to develop and share evidence-based recommendations and pathways (9%). Others reported using PM to identify care process steps within activity-based costing methodologies (6%) or to provide a visual representation of patient journeys (5%). Only 3% of papers described the use of PM in integrated care pathway (ICP) projects. (Fig. 4).

We assessed all empirical studies against the conceptual framework quality criteria for each phase of PM. The key findings are displayed in Fig. 2.

No study reached overall compliance for all 10 the criteria. Only 7 studies (7%) adhered to either 9/10 (2%) criteria or 8/10 criteria (5%). For five of these seven studies the criteria about involvement of those who would be using the processes (e.g. patient involvement) was not met. Most studies adhered to 7/10 (15%), 6/10 (32%) and 5/10 (35%) of the criteria. The remaining 10% of studies were compliant to 4/10 criteria.

Phase 1: Most projects clearly identified a service family and the patient/service user group (91%), but patient representative involvement was only reported in a small number of projects (15%). Team member training in the technique prior the PM exercise was reported in only 15% of projects. This was usually delivered throughout meetings and by using examples [44, 64‐70], while in some cases this included intensive QI training [71, 72].

Phase 2: 85% of the studies stated that data and information had been gathered to inform the PM exercise as a substitute for (55%), or in addition to (30%), the group knowledge generated in the facilitated process mapping sessions. This included evidence-based best practice recommendations [34‐155], interviews [64, 73‐148, 154, 156, 158, 160], and more detailed approaches such as observations, operational data collection, time-and-motion studies, and video footage [48, 78‐81, 141, 143‐146, 150, 159, 163‐169].

Phase 4: All 105 studies reported that the process map was analysed. How and at what point the analysis was carried out varied significantly, depending on the type of project being reported. Most studies (91%) reported that the map was created to represent current state practice. Four projects [82‐85] reported both current state and ideal or future state maps, while 5 describe the ideal or future state process [65, 86‐89]. For example, in TDABC projects the analysis is focused on process costs and is mainly completed after the current state process map is created [90‐95]. In projects implementing evidence-based recommendations, a process analysis was completed before the creation of the ideal or future state process map [89]. Most studies (78%) reported that additional information gathered during the PM exercise - such as delays, safety problems, or flow of information, resources and activity - is represented on the final map. Only 19% of studies reported the use of charting software to draw the process map or to make a tidy version of the paper-based map, and only 48% of studies specified that the PM exercise had been reviewed for accuracy and confirmed by key stakeholders or external experts.

We identified the benefits of using PM in improvement initiatives described in the reviewed empirical literature and grouped them into three areas: (i) understanding local systems, (ii) inform scope, design, development and evaluation of interventions and (iii) co-production and knowledge exchange. (Fig. 4).

Studies reported that QI teams gained a more realistic understanding of current practice because PM allowed them to gather knowledge from people directly involved in the process under analysis and provided a visual representation of current or enhanced processes [65, 83, 96, 142]. The studies show that PM is a tool to break-down the complexity characterizing healthcare, by providing improvement teams with a structured picture of complex processes, using information from process stakeholders holding different roles and perspectives [34, 88, 90, 96, 98‐100, 140]. Diverse views elicited during PM help improvement teams gain a shared understanding of local practices and underlying systemic issues. For example, PM has been found particularly useful to disaggregate care process and identify costs for each process step [90‐95] as well as to understand interactions between different parts of healthcare systems. For example, in ICP projects the use of PM to understand systems helped to improve coordination of care across different settings and networks [86, 87, 101], while in FMEA projects it helped to identify potential systems failures [48, 66‐68, 100, 102‐107].

The identification of actual constraints and opportunities within local systems helped assessment of problem areas and development of improvement solutions grounded in research evidence and local knowledge [65, 76, 79, 81, 92, 96, 108‐110, 149, 151]. The use of PM before the implementation of Information Systems (IS) has been shown to support project members with diverse backgrounds achieve a shared understanding of the system and has been reported as crucial for solving design challenges [88, 111‐115]. Some studies describe how PM has also been used to assess actual care processes against recognised evidence-based standards [79, 99]. Reviewed studies also show how if used throughout the entire project, PM can play a role in the success of improvement initiatives by supporting continuous improvement. In this respect, PM has been found to be particularly useful for clarifying the scope of projects, targeting the intervention and planning improvement actions [9, 79, 116, 161].

PM was reported to be particularly useful to engage and motivate project stakeholders in designing and implementing change. Some studies report that PM supported the effective design of HIT by enhancing the involvement of process stakeholders [88, 111‐115]. Studies also describe how greater understanding of different perspectives provided by PM encouraged a culture of ownership and responsibility for improvement work [34, 65, 66, 81, 92, 118, 119]. For example, within ICP projects, PM allowed the clarification and reassessment of the roles and responsibilities within the team [86, 87, 101]. Other studies highlight how participation in PM helped to establish sense of urgency in clinicians regarding patient safety issues, thus enhancing their engagement [81, 92, 157]. Reviewed studies also show that the capacity of PM to facilitate the dialogue between diverse stakeholders helps to smooth barriers and tensions occurring during improvement projects or reach consensus on solutions [65, 66, 71, 73, 81, 92, 103, 118, 119]. For example, some studies reported how PM helped to promote the integration of health services across different settings by developing clinical evidence-based recommendations agreed among different healthcare professionals [74, 86, 87, 89]. Finally, studies describe how PM can be a valuable tool for documenting a care process for further dissemination [170]. This is beneficial, for example, in helping to inform patients and carers about their expected journey [75, 99, 120‐122] or to support training and education of healthcare professionals [81, 101, 152, 153, 162].

We unpacked the black box of PM as a QI method and outlined quality criteria to guide its systematic use and reporting. Improving the quality of reporting of PM exercises would enhance transparency, encourage appropriate use of PM in practice, and support the definition of a common language to describe the process of PM [24]. We encourage practitioners and researchers to use and test the validity of our conceptual framework when implementing or reporting PM. We also suggest further development of reporting guidance for PM exercises and their use as a starting point in the design of prospective studies exploring the effectiveness of the method. Our findings show that improvements in reporting are required not only to systematically describe the “process” of PM but also for representation of the process map, as we found that many articles report only a partial or sample representation of the process map developed. Online versions of published articles or online supplements [48, 66, 68, 69, 90, 132‐136] could provide more detailed process maps as these are often difficult to display in printed versions of journals. Improvements in the way process maps are represented and reported might increase the effectiveness of PM as a key QI method. For example, annotating the process map with operational (e.g. waiting times, activity durations, waste/ value), cost (e.g. resources required to perform each activity), patient experience or other project data (e.g. areas targeted or changed by various plan-do-study-act cycles), can be helpful to visually identify gaps in the systems and document the process analysis throughout the project. Previous studies also demonstrated that successful implementation of QI initiatives depends not only on the conformance to methodological guidelines, but is greatly influenced by contextual factors (leadership, organisational culture, etc.) [16, 137‐139]. Our study has not taken into account the influence of context on PM exercises, because these factors cannot be assessed by analysis of the literature. While the main contribution of this study is in identifying quality criteria to support a more rigorous use and reporting of PM, we encourage practitioners and researchers to consider the influence of contextual factors on the effective use of QI approaches.

There is a need for further empirical research to explore the impact of improvement initiative context on practical implementation of PM. We partially explored how PM is used in practice by improvement teams in the NHS in a previous empirical study investigating benefits and success factors of PM in a sample of QI projects [12]. However, most of the projects included in this study [12] used the same methodological approach to PM, (multi-stakeholder meetings to generate the process maps). Further empirical research is needed to test whether our findings hold in QI projects developed by teams using different approaches to conduct the PM exercise, as identified in this literature review (e.g. when PM is used within Six Sigma or Lean approaches). Further literature and empirical research could also explore the representation of process maps in more detail. This would provide a wider perspective on how process activities can be represented and annotated with a variety of information (e.g. value/waste, bottlenecks, constraints, patient experience) and how this can influence the effective use of PM within improvement initiatives.

There are some limitations due to the search process. The database search could have included other search terms such as “process model*”, “process design*” or “system design*”, but the authors agreed that the effort required to screen the resulting records was not justified by the purpose and boundaries of the present study.

A key limitation is due to the fact that the systematic review is based on PM exercises as described in the selected empirical literature and not on the analysis of actual practice. This implies that results might be affected by reporting bias and selection of studies, as well as publication bias. The content of publications heavily depends on what journals accept for publication and on the limited space allowed. Therefore, projects using specific approaches (e.g. TDABC, Lean or IS development) are less likely to present a detailed description of the PM process, compared to other process improvement projects. Successful projects are more likely to be published than studies reporting less successful interventions, which may be equally useful for knowledge generation. Bias could also arise because we only searched English-language papers. However, our objective was not to perform an exhaustive review of all the studies applying PM techniques in healthcare, nor to assess the effectiveness of PM, but to provide a representative overview of the use of PM as reported in empirical literature.

Another limitation is due to the fact that PM exercises were usually reported as a part of a wider project. Clearly distinguishing the component attributable to PM from that associated with the whole project was therefore not always straightforward. We addressed this limitation in the development of the data item sheet and the conceptual framework, as well as in the data collection and analysis phase. For example, we decided not to quantitatively assess the different roles involved in the PM exercise, because it was not always clear if and how all team members were involved in the PM exercise. Furthermore, we evaluated the actual implementation of the recommendations derived by the PM exercise, considering the improvement actions reported in respect of the whole project.

Finally, within the included studies we found three papers [120‐122] which seemed to derive from the same project. We addressed this bias in the analysis and summary phase by discounting the patterns emerging from common characteristics of these three studies.