Building a Functional Resonance Analysis Method Model: Practical Guidance on Qualitative Data Collection and Analysis

Abstract

The Functional Resonance Analysis Method (FRAM) is a novel research methodology that uses qualitative data collection methods to map and model complex healthcare processes by identifying and depicting the cumulative activities required to produce an outcome. The FRAM aims to identify the variability that can emerge in a process when healthcare activities are performed under dynamic conditions. With this knowledge, health care system design, safety, and quality improvement recommendations can be developed with a greater understanding of everyday process functionality. Researchers interested in using the FRAM require both an understanding of the methodology itself, as well as an understanding of how to effectively plan and conduct qualitative research. The purpose of this paper is to provide practical guidance to researchers on planning and operationalizing qualitative data collection and analysis methods to inform the building of a FRAM model. A combination of literature and practical experience will be used to examine and suggest appropriate ways for researchers to carry out this work.

Keywords

functional resonance analysis method process modelling qualitative data collection

Overview of the FRAM

The FRAM was first developed in the early 2000s in the field of engineering to be used in research and development related to safety and accident analysis (Patriarca et al., 2020). The methodology has been used to map, model, and analyze complex processes and systems in a number of domains, such as aviation, maritime transport, industry, and healthcare (Salehi, Veitch, et al., 2021). The FRAM has emerged more recently as healthcare research methodology that is gaining recognition with most studies being published since 2017 (McGill et al., 2022). In a review of the FRAM literature in 2021–2022, there have been an additional seventeen healthcare related publications featuring the FRAM ((Bos et al., 2022; Buikstra et al., 2021; Damen & de Vos, 2021; Gustafson et al., 2021; Hedqvist et al., 2022; Jatobá et al., 2022; MacKinnon et al., 2021; Schreurs et al., 2021; Slater, 2022; Slater et al., 2022; Sujan, 2021; Sujan et al., 2022; Thude et al., 2021; Tresfon et al., 2022; van Dijk et al., 2021; van Dijk et al., 2022; Watson et al., 2022). Before using the FRAM, researchers are encouraged to familiarize themselves with the methodology’s background and principles. Providing a complete review of the FRAM is beyond the scope of this paper. Fundamental FRAM literature is available that can guide a novice FRAM researcher (Hollnagel, 2012; Hollnagel et al., 2014; Hollnagel & Slater, 2022).

A FRAM model depicts the interdependent activities of work that make up a process to produce an outcome (Hollnagel, 2012). The model created, allows clinicians and administrators to gain a greater appreciation of the complexity of a healthcare process that may otherwise be invisible when using more traditional or sequential methods of analysis (McGill et al., 2022).

Studies using the FRAM can provide an enhanced understanding of complexity for a numbers of purposes, including process optimization, incident investigation, guideline development and implementation, intervention development, and prospective risk management (Damen & de Vos, 2021). Currently, there is much discussion in healthcare around complexity with little dedication to researching it appropriately (Greenhalgh & Papoutsi, 2018). The key to gaining this understanding is to gather data from information-rich sources who are stakeholders in the process, such as healthcare professionals and patients and their families. Hollnagel & Slater, 2022, explain the people carrying out the work are best able to provide information about the activities being analyzed. The accounts of stakeholders are key to distinguishing between the concepts of work-as-imagined and work-as-done. Work-as-imagined is a description of work that often is conceived from how a healthcare process is supposed to happen according to the literature or a description provided by management. Work-as-done is a description of work as it actually takes place in everyday conditions provided by those who actually do the work (Hollnagel & Slater, 2022). The concepts of work-as-imagined and work-as-done are central to the FRAM and several studies using the FRAM have examined this phenomenon and have been able to provide input on how work-as-imagined and work-as-done can be better aligned (Clay-Williams et al., 2015; Damoiseaux-Volman et al., 2021; Schreurs et al., 2021; Tresfon et al., 2022; van Dijk et al., 2022; Watson et al., 2022).

Steps of the FRAM

The steps explained below have been adapted for the healthcare context from an original description (Hollnagel et al., 2014).

• Step 1 is concerned with identifying a clearly described purpose and scope of a FRAM analysis of a healthcare process.

• Step 2 is concerned with identifying and describing the activities required for a healthcare process to take place.

• Step 3 is concerned with describing how the activities in a process vary.

• Step 4 aims to show how the aggregation of variability in activities impacting one another early in healthcare process (upstream) may have an impact on activities later in the process (downstream).

• Step 5 is concerned with monitoring the process and identifying how any negative variability that emerges can be dampened and how any positive variability can be enhanced.

FRAM Terminology

The FRAM refers to healthcare activities in a process as “functions” (Hollnagel & Slater, 2022). Functions are continuously carried out in healthcare processes and are human, organizational, and technological (Ross et al., 2018). For instance, functions carried out by community-based pharmacists would include “obtaining a medical history” and “communicating with the prescriber.” For these functions to be carried out, they have temporal and resource requirements (human, technological, operational).

Additionally, functions require guidance, such as policies, procedures, and clinical guidelines. All these characteristics of functions are referred to as “aspects” (Hollnagel & Slater, 2022). The FRAM specifies how functions are characterized in terms of six aspects: Input (I), Output (O), Resources (R), Time (T), Control (C), and Preconditions (P). Clay-Williams et al. (2015) best defines the aspects that characterize functions:

1. The Input is what the function acts on or changes (what is used to start the function).

2. The Output is what emerges from the function (an outcome or state change).

3. A Precondition is a condition that must be satisfied for a function to happen.

4. The Resources are materials or people needed to execute a function.

5. Control is how the function is regulated or controlled (guidelines, protocols).

6. Time refers to any temporal requirements of the function.

Functions that are interdependent and impact one another in a process are “coupled” (Hollnagel, 2012). Functions that are coupled are connected through mutually shared aspects. A FRAM model is a visual depiction of all the functions and connections among the functions that can potentially take place within a health care process. The graphical depictions can be built, edited, and shared using specialized FRAM software (Hollnagel & Hill, 2020). The visualization of how a healthcare process can be carried outwith a FRAM model is a strength of the methodology as it allows clinicians and administrators to see all the potential ways a process can take place and to gain an appreciation the complexity that exists which may otherwise be invisible (McGill et al., 2022).

The authors of this paper are currently conducting a study examining the process of community-based comprehensive geriatric assessment. Figure 1 is an example of a FRAM model from the study in progress that depicts how a medication review with a community pharmacist can potentially take place. It shows the essential functions of the process, as well as how functions are connected and interdependent.

Figure 1.

A FRAM Model Depicting The Process of Community-Based Medication Review for Older Adults.

Researchers using the FRAM to examine and analyze complex health care processes must be aware that the resultant models, analyses, and recommendations are only as accurate and relevant as the quality of the data collected. When conceptualizing and planning a study using the FRAM, it is important for researchers to appreciate that simply appropriating qualitative methods to identify the functions, aspects, and interdependencies of a healthcare process does not suffice. A number of resources exist that provide a comprehensive overview of the tenets, methodologies, and methods of qualitative research (Clarke & Braun, 2013; Polit & Beck, 2020; Yin, 2016). This paper aims to provide practical guidance for researchers on how to operationalize data collection and analysis methods to inform the building of a FRAM model (steps 1 and 2 of the FRAM). The guidance in the fundamental FRAM literature related to these vital steps of the methodology is currently underspecified. Guidance on subsequent steps of the methodology is beyond the intended scope of this paper.

Designing a Study Using the FRAM – Step 1

The accuracy and validity of a FRAM model is dependent on the data collected and analyzed to build it. Qualitative research examines phenomena in a detailed and holistic way and aims to gain an understanding of and provide insight into real world issues (Moser & Korstjens, 2017). Gathering accounts of how everyday work is accomplished from those who provide healthcare or receive healthcare is essential to building a FRAM model that accurately represents the activities of a healthcare process essential to producing an outcome (Hollnagel, 2012). A clearly described purpose and scope of a FRAM analysis allows researchers to delineate boundaries of the process they intend to examine and provides direction on preparing a sampling plan and determining appropriate methods of data collection.

Oduyale et al. (2020) aimed to “explore the everyday practices surrounding co-administration of multiple IV medicines by Intensive Care Unit (ICU) nurses down the same lumen, the challenges encountered during the process of co-administration, and investigate how compatibility is assessed and managed in practice” (p.157). Focus groups were the chosen method of data collection for this study. The rationale for this decision was to allow ICU nurses with similar experiences to reflect on their everyday practice and provide in-depth responses about shared and common knowledge (Oduyale et al., 2020). The sample was purposive in that it aimed to include only qualified nurses from an ICU practice setting, with experience in the process of IV medication co-administration. A total of 18 ICU nurses were included in three focus groups, no rationale for the sample size was provided, but evidence supporting the size of each focus group was provided to readers. A strength of this study is the clear description of the purpose and scope of the FRAM analysis as well as the rationale provided for the study design choices.

Developing a Sampling Plan

A sampling plan can be broadly defined but should specify an approach to sampling and a rationale for the choices made (Moser & Korstjens, 2018). There is no requirement in qualitative research to explicitly state a sample size as this is often determined by data saturation. Saunders et al. (2018), defines data saturation as the point in the data collection process when additional data does not lead to any new or emergent information. With data saturation, the sample number emerges as the study goes on. In examining the most recent FRAM literature, the sample size for studies using interviews as the primary method of data collection ranged from 8 to 31 participants (Damoiseaux-Volman et al., 2021; Schreurs et al., 2021; Sujan et al., 2022; Thude et al., 2021; van Dijk et al., 2022).

The section in a study describing sampling is often not the most exciting or interesting one to read, but when it comes to a study using the FRAM it is one the most important sections to describe explicitly. Studies using the FRAM to examine complex processes aim to identify the functions of everyday work and the potential interdependencies and variabilities that emerge in a process under dynamic work conditions. The data needed to accomplish this goal is best gathered from stakeholders who have firsthand knowledge of how the system or process functions on an everyday basis (Hollnagel, 2012). A clear description of the eligibility criteria, steps taken to recruit participants, study setting, as well as who the participants are relative to the process of interest should be provided for readers. Qualitative researchers can choose from a variety of non-probability sampling methods to recruit participants (Polit & Beck, 2020).

Purposive Sampling

In reviewing the recent FRAM literature, several studies elected to conduct a purposive sampling approach in their studies (Bos et al., 2022; Buikstra et al., 2020, 2021; Damen et al., 2021; Gustafson et al., 2021; Kaya et al., 2019; Oduyale et al., 2020; Schreurs et al., 2021; Schutijser et al., 2019; Sujan, 2021; van Dijk et al., 2021). Yin (2016), claims in qualitative research the sampling approach is most often purposive to ensure the most relevant and plentiful data is obtained given the study topic with an emphasis on information rich sources. The rationale for selecting purposive sampling in studies using the FRAM is to ensure the sample can provide a work-as-done description to inform the building of a FRAM model that accurately depicts a healthcare process of interest under every-day conditions. In their FRAM analysis of the management of the deteriorating surgical patient, Sujan et al. (2022) explained purposive sampling was employed to identify participants who work on a surgical emergency unit or who would be involved in the wider system effort in caring for deteriorating surgical patients.

The study additionally provided a table for readers listing the number of participants by professional role. Schreurs et al. (2021) and Oduyale et al. (2020) also provided readers with tables describing participants in their studies by role and years of experience.

This is an important addition for any FRAM study because it provides readers with an overview of how broad or narrow the range of information and perspectives on the study focus will be. Health care processes are rarely completed by a lone professional, obtaining information and perspectives from a variety of workers is essential. Ensuring a variety of participants are included in a study is known as a maximum variation sampling, which is a variant of purposive sampling (Yin, 2016). Researchers who use the FRAM should also determine if the study focus would benefit from the inclusion of patients and their family caregivers. Laugaland et al. (2014) first voiced concern over the FRAM focusing solely on health care providers’ perspectives and advocated for future studies to include patients and family caregivers. Buikstra et al. (2020) echoed this sentiment by identifying this as a limitation in their study examining variability in the discharge summary process for older adults. Subsequently O’Hara et al. (2020) included patients and families as participants in their study which used the FRAM to examine the hospital to home transition process in older adults. The data from this study provided new insights that included upstream hospital functions (e.g., encouraging mobility and supporting patients to better understand their medications and medical conditions) leading to improved outcomes for patients following hospital discharge. These findings would not have been realized if only the perspectives of health care professionals were considered.

Convenience Sampling

Convenience sampling is based purely on availability or accessibility of the sample (Polit & Beck, 2020). In qualitative research this type of sampling is not preferred because the sources providing the information may not be informative and produce an unwanted degree of bias (Yin, 2016). Watson et al. (2022) used convenience sampling in a study using the FRAM to examine the process of oxygen prescribing on inpatient units. The authors identified the use of convenience sampling as a study limitation because the precise level of understanding and experience of each participant related to the process of oxygen prescribing was not known. This may lead to the development of a FRAM model that does not truly reflect the everyday functions, interdependencies and variability that occurs in the process. Watson et al. explain the rationale for the convenience sampling approach was due to the practical challenges of data collection on a busy inpatient unit.

Snowball Sampling

Snowball sampling is an additional approach to sampling that could potentially assist researchers in overcoming recruitment challenges related to access. Polit and Beck (2020) define this approach as the sampling for a study through references from earlier participants in the study. Yin (2016) explains snowball sampling can be an acceptable sampling approach if it is purposeful and not done out of convenience. Purposeful snowball sampling aims to ensure that each referral meets predetermined eligibility criteria that ensures they possess the experience and knowledge specific to the process of interest. Arcuri et al. (2020) were challenged in accessing physicians for a study using the FRAM to depict the resiliency in the process of referral prioritization. A purposive snowball sampling approach was used to ensure the sample had the desired knowledge and experience to contribute to the study.

A weakness identified with snowball sampling is that it could limit the participant pool to a small number of acquaintances (Polit & Beck, 2020). This could be problematic in a health care setting where a snowball sampling approach may only capture workers who come from certain groupings of likeminded participants, such as 9–5 workers. Table 1 describes the different sampling approaches researchers can consider when designing a study using the FRAM.

Table 1.

Sampling Approaches Defined With Examples From the FRAM Literature.

Sampling approaches defined	Examples From FRAM Literature
Purposive sampling involves selecting participants based on their ability to provide the most relevant and in-depth data to analyze	Bos et al., 2022
	Buikstra et al., 2020
	Damen et al., 2021
	Damoiseaux-Volman et al., 2021
	Gustafson et al., 2021
	Kaya et al., 2019
	Oduyale et al., 2019
	Schreurs et al., 2021
	Schutjiser et al., 2019
	Sujan, 2021
	Sujan et al., 2022
	van Dijk et al., 2021
Maximum variation a variant of purposive sampling that aims to ensure a variety of participants with differing experiences, backgrounds, and knowledge are included (Yin, 2016).	Buikstra et al. (2020)
	Salehi et al., 2021a
Convenience sampling is when a sample is selected based purely on availability or accessibility to the researcher (Polit & Beck, 2020).	Watson et al. (2022)
Snowball sampling is when a sample is selected by way of referrals from earlier participants in the study. Snowball sampling can be purposive or out of convenience. (Polit & Beck, 2020)	Arcuri et al. (2020) [purposive]

Qualitative Data Collection Methods

There are several methods of data collection researchers using the FRAM can consider when designing a study. Polit and Beck (2020) state “it is often difficult to critically appraise the decisions researchers make in collecting qualitative data because details about those decisions are seldom spelled out” (p: 275). Efforts should be made to clearly describe how qualitative data was collected. Since FRAM studies are highly contextual, researchers will need to determine what methods of data collection will capture the data required and provide rationale for why the method(s) chosen were appropriate for the purpose of the study. The following section will review the most common methods of qualitative data collection used in healthcare studies that employed the FRAM with references to select examples from the literature of how researchers have approached data collection and analysis to inform the building of a FRAM model.

Semi-Structured Interviews

Developing a Semi-Structured Interview Guide

Semi-structured interviews are the most common method of data collection in healthcare studies using the FRAM (McGill et al., 2022). DeJonckheere and Vaughn (2019) describe semi-structured interviews as an effective method to collect qualitative, open-ended data on a topic of interest from key informants and is guided by a flexible interview protocol that allows for follow-up questions, probes, and comments. Preparation for conducting semi-structured interviews is key.

The goal should be to develop well-planned interview questions and probes that can generate rich, detailed accounts while also ensuring a rapport has been developed with the participant (Polit & Beck, 2020). Developing a well thought out interview guide first requires an understanding of the healthcare process of interest. Gaining this understanding can assist researchers in determining the general information required to build a FRAM model.

Hollnagel et al. (2014) suggest consulting all available sources of information that can potentially form the basis of the interview questions and prompts, this may include document review of policies, procedures, and/or guidelines commonly used by workers in the process. Additionally, researchers may want to visit the work setting and meet with key stakeholders prior to conducting an interview. In the FRAM literature there are several resources that can assist in the development of a semi-structured interview guide. Table 2 provides an overview of the resources available in the FRAM literature that can assist researchers in developing a FRAM semi-structured interview guide.

Table 2.

Resources for Developing a FRAM Semi-structured Interview Guide.

Resource	FRAM Literature
Table of guided questions used to explore FRAM conditions	Clay-Williams et al. (2015)
Interview template with examples of questions	Hollnagel et al. (2014)
Interview template with examples of questions	Hollnagel and Slater (2022)
Topic lists	Shutisjer et al. (2019)
Topic lists	Damen et al. (2018)
Interview guide	Watson et al. (2022)

Conducting Semi-Structured Interviews

Hollnagel et al. (2014) suggest it can be useful if there are two interviewers conducting the semi-structured interview together with one interviewer asking questions and the other taking notes. Shutijser et al. (2019) used this approach to conduct semi-structured interviews in their study using the FRAM to examine and analyze the process variation between a protocol for double checking medication and the realities of everyday work by nurses. In this study, one researcher conducted interviews with a second researcher taking notes. One interviewer from the work domain who is not a manager or supervisor is also suggested so participants feel they can speak freely (Hollnagel et al., 2014). Relying solely on notes taken by a second researcher may result in losing the richness and detail of the interview (Clarke & Braun, 2013). Having an audio recording of the interview with transcribed notes is preferred to note taking since it provides a precise record of the interview that can be revisited by the researcher (Clark & Braun, 2013). To optimize participant responses, Polit and Beck (2020) suggest developing semi-structured interview guides that are flexible enough to allow the participant to feel they can speak freely about their knowledge and experiences, but also have direction with their sequence and reflect the broader research protocol.

Watson et al. (2022) use a semi-structured interview guide in their study exploring oxygen prescribing and administration on hospital inpatient wards. The guide begins by clearly stating to participants the purpose of the study and what is required of participants. This is an important starting point because it orientates participants to the specific process being examined and the information researchers are seeking. A subsequent question early in the guide asks, “Can you talk me through the process from your perspective?” (Watson et al., 2022: S2).

The question is an example of a “grand tour question” which is a type of question that aims to establish a broad setting or topic and does not focus on the specific item of interest or sequence of topics (Yin, 2016, p. 144). Providing the participant an opportunity to share their experience and knowledge of the process can potentially reveal a significant amount of information about functions, aspects, and potential variability, with opportunities for the interviewer to pose follow up questions. Polit and Beck (2020) also suggest “the ideal interview guide is often conceived of as an inverted triangle, moving from the general to the specific” (p: 96). The interview guide by Watson et al. goes on to pose several open-ended and closed ended questions as well as prompts regarding the process of oxygen administration on an inpatient ward. The questions and prompts focus on identifying the specific functions of the oxygen administration process and the aspects that characterize those functions, as well as how functions are potentially interdependent.

A key point to note in this interview guide is the authors did not use FRAM terminology to elicit details from participants about functions, aspects, and interdependencies, but rather language more familiar to clinicians. This approach allows for the interview to remain more conversational in nature and encourages the participant to use their own words rather than the terminology of the researcher (Yin, 2016). Polit and Beck (2020) explain a closing or clean up question at the end of an interview guide can allow the participant to offer any additional information that may not have been covered. A simple question, such as “Is there anything more you can tell me about the process” can potentially trigger data that was unanticipated (Polit & Beck, 2020).

Document Review

As previously described by Hollnagel et al. (2014), document review is a data collection method that can inform the researchers about the practice setting and assist in preparation of interview questions. Additionally, document review can identify data that provides insight into the context within which research participants operate (Bowen, 2009). When examining a health care process or system there may be numerous documents that are used in everyday practice that may guide or have an influence on how work is carried out. These may include best practice guidelines, policies, procedures, as well as documents that facilitate communication between health care providers. Document review is an advantageous method of data collection because it is unobtrusive and relatively stable, with little concern for the potential for researcher influence when compared to other methods, such as observation and interviews (Bowen, 2009). Alternatively, there are some limitations in using this method: documents may be challenging to retrieve, they may provide little detail, and in the context of a health care organization, the documents are likely to be aligned with the organization’s agenda or principals (Bowen, 2009).

The accuracy of how organizational documents are able to successfully inform and/or guide work in a dynamic health care setting has become the focus of several studies using the FRAM (Clay-Williams et al., 2015; Damen et al., 2021; Damoiseaux-Volman et al., 2021; Furniss et al., 2020; Schreurs et al., 2021; van Dijk et al., 2022). These studies aim to reconcile the gap between ‘work-as-imagined’ and ‘work-as-done’ by examining documents that guide clinical practice and then using the FRAM to examine and analyze how work is accomplished in everyday practice.

Braithwaite et al. (2015) explain documents that are often used in everyday clinical practice are conceived around the premise of ‘‘work-as-imagined’’ rather than ‘‘work-as-done’’ and unfortunately “work-as-imagined always differs from what actually goes on—work-as-done—and the difference increases the further removed people are from the front line” (p. 419). Schrueurs et al. (2021), examined protocols, guidelines, and literature to build a work-as-imagined FRAM model of the process of elastic compression stocking therapy for individuals with chronic venous insufficiency and deep vein thrombosis. After completing the work-as-imagined model, a work-as-done model was built based on interviews with key health care professionals who conduct this work. How practice varied in the process of elastic compression therapy was identified and improvement initiatives for this process were able to be developed.

Observations

Studies employing the FRAM have used observations as part of multi-method data collection (Alm & Woltjer, 2010; Arcuri et al., 2020; Clay-Williams et al., 2020; Furniss et al., 2020; Kaya et al., 2019; Laugaland et al., 2014; MacKinnon et al., 2021; Patriarca et al., 2018; Pickup et al., 2017; Raben et al., 2018; Salehi et al., 2021a; Schutijser et al., 2019; Thude et al., 2021; Tresfon et al., 2022). Weston et al. (2021) found workers perform habitual tasks and activities and may not recall their specific actions or be aware of how “tacit and explicit knowledge” can influence work (p. 105).Observational research can explore these contextual nuances that may be challenging to capture in interviews, focus groups, or document analysis (Weston et al., 2021). Laugaland et al. (2014) used moderate participant observation in their study examining the functions, variability and performance-shaping factors related to hospital discharge of the elderly.

This type of observation “entails that the researcher be present and identifiable, though not an active participant (i.e., does not have a role in the practice setting); the researcher observes and interacts occasionally” (p. 4). The observer used a semi-structured observation guide that was developed using the FRAM approach, to observe the work conducted on the day of an older adults’ discharge. Selectivity regarding observations should be an explicit part of the data collection procedure and should reflect the purpose of the study (Yin, 2016). In their study examining how a protocol for double checking injectable medication administration transfers to practice, Shutijser et al. (2019) conducted work-as-done observations of the daily practice of the double check during the medication rounds. The most important proceedings of the double check protocol were marked by the research team when completed by the nurses. Shutijser et al., (2019), then described how observations were recorded using a standardized observation form.

How observations will be recorded in the field is another important consideration for researchers. Field notes are the observer’s efforts to record, synthesize and understand the data. Tresfon et al. (2022) conducted observations over 10 days on a nursing ward to observe the use of restraints to develop a work-as-done FRAM model of the practice. During the observations data was collected using “in field jottings” that the observers later elaborated on in the form of field notes (Tresfon et al.). The observer also kept a research diary reflecting on their role and influence as an observer. Van Djik et al. (2022) argue not using observations in their medication reconciliation study is a limitation due to the potential for participants tailoring their descriptions of their work in interviews. Van Djik et al. (2022) reference Shorrock. (2020) in their description of this phenomenon as “Work-as-Disclosed” rather than “Work-as-Done” and contend interviews may not provide a true description of everyday work and that some activities of work that are performed unknowingly may also be missed.

Focus Groups

Focus groups are a valuable source of data for researchers using the FRAM. A moderator rather than interviewer typically leads the discussion among participants because the aim of a focus group is not to conduct a group interview; the aim should be for the participants to discuss points raised by the moderator among themselves (Clarke & Braun, 2013). The ideal number of focus group participants is not exact, and ranges have been provided in the literature: 3 to 8 (Clarke & Braun, 2013) and 5 to 10 (Moser & Korstjens, 2018; Polit & Beck, 2020).The goal would be for a focus group to be small enough that it allows for the involvement of all participants, but not so small that it limits the richness and diversity of perspectives (Clarke & Braun, 2013) One strength of conducting focus groups is participants can hear one another’s responses and provide additional comments that they may have been reluctant to make individually (Carter et al., 2014)

Alternatively, as with any group setting there may be instances where using a focus group as a data collection method may deter participation from some participants. Oduyale et al. (2020) conducted focus groups with 20 ICU nurses to build a FRAM model representing the process of co-administration of multiple intravenous medications. A limitation identified in the study states “the presence of senior staff members in the focus groups could have prevented some junior nurses from expressing their opinions and co-administration practice freely” (p. 162). Researchers using the FRAM should determine how similar or different focus group participants should be. Heterogenicity can bring different views to the focus group which can generate a more diverse discussion, while homogeneity of the focus group may create a more familiar and comfortable social environment (Liamputtong, 2011).

Clarke and Braun (2013) recommend that similarity within the focus group should be determined in relation to the topic of the research. Researchers using the FRAM will need to determine what information or knowledge they are trying to gain from the focus group to best answer their research question.

Building a FRAM Model – Step 2

Before building the model, researchers employing the FRAM should begin their analysis by taking the time to familiarize themselves with the data. Clarke and Braun (2013) describe this as an “immersion in the data with an aim of becoming intimately familiar with the content of the dataset and to start taking notice of what might be relevant to the research question (p. 216). Familiarization may consist of listening and re-listening to audio recordings, transcribing interviews, and reading and re-reading transcripts, and field notes. The following section will outline the activities to build a FRAM model and present examples of different approaches used by researchers in the FRAM literature.

Identifying and Describing Functions and Aspects

There are several examples of how functions and aspects have been identified in healthcare studies using the FRAM (Damoiseaux-Volman et al., 2022; Laugaland et al., 2014; O’Hara et al., 2020; Schreurs et al., 2021; Sujan et al., 2022; van Dijk et al., 2022; Watson et al., 2022). The way researchers approached the identification of functions and aspects was not overly well explained, with some studies simply describing it as an “iterative process” (Alm & Woltjer, 2010; Furniss et al., 2020; Schreurs et al., 2021; Tresfon et al., 2022). Others were more descriptive and provided readers with further insight.

O’Hara et al. (2020) began their analysis by having two researchers analyze the study data to identify the activities that typify the process of hospitalization and discharge in older adults (O’Hara et al., 2020). The researchers then met several times over the course of a week (approximately 35 hr) to decide how the work activities could be constructed into discrete functions (O’Hara et al., 2020). Laugaland et al. (2014) identified common functions in the discharge process of older adults by having the first and second authors individually review the field note summaries of observations, followed by a review of the same data by a team of four researchers. Laugaland et al. (2014), went on to explain the identified functions were then revised several times until the team reached a final consensus and a detailed description of the functions (including associated aspects—time, control, input, output, resources, and preconditions).

Coding

Several studies used the data analysis strategy of coding to identify and describe functions (Damoiseaux-Volman et al., 2021; Kaya et al., 2019; Oduyale et al., 2020; Salehi et al., 2021a; Thude et al., 2021). Coding aims to “capture (potentially) relevant meanings in data, in relation to a research question” (Clarke & Braun, 2021, p. 297). Linneberg and Korsgaard (2019) describe the coding of qualitative data as an approach that enables a deep immersion in the data as well and ensures structure and transparency in the development and presentation of findings. Selective coding is a type of coding that is particularly applicable to the FRAM. This type of coding requires pre-existing theoretical knowledge that provides the researcher with the ability to identify the analytic concepts they are looking for (Clarke & Braun, 2013). A researcher’s pre-existing knowledge of the FRAM can guide the analysis and be used to identify functions and describe aspects of functions.

Coding can be done by hand or by using coding software, such as NVivo 14 (Lumivero, 2020). Coding software also assists in the storage, organization, and management of large qualitative data sets that can be accessible to multiple members of a research team (Linneberg & Korsgaard, 2019). Damoiseaux-Volman et al. (2021), took a different approach to coding by developing a code tree based on a work-as-imagined model for the prevention of falls and delirium in older adults admitted to hospital. The code tree was a list of functions identified in a work-as-imagined model of the process.

To build a work-as-done model of the process, researchers used the code tree to analyze semi-structured interviews with front-line staff and identify the differences between work-as-imagined and work-as-done (Damoiseax-Volman et al., 2021). Three studies used the qualitative data analysis method of Thematic Analysis to analyze data and build a FRAM model (Oduyale et al., 2020; Salehi et al., 2021a; van Dijk et al., 2021). Thematic Analysis is a theoretically flexible approach to developing patterns of meaning (themes) across a dataset that addresses a research question (Braun & Clarke, n.d.). Patterns/themes are generated by the researcher using a six-phase process, see Table 3 (Braun & Clarke, 2021; Clarke & Braun, 2021). In this approach researchers using the FRAM code the data to identify and describe functions in terms of their aspects by treating functions as themes.

Table 3.

Definition and Steps of Thematic Analysis (Braun & Clarke, 2021; Clarke & Braun, 2013, 2021).

Definition: Thematic analysis	Steps of Thematic Analysis
A widely used qualitative data analysis method that uses a six-phase process to identify patterns of meaning (themes) across a dataset related to a research question, themes are developed through systematic processes of data coding (Clarke & Braun, 2013).	1. Data familiarization and writing familiarization notes
	1. Systematic data coding
	1. Generating initial themes from coded and collated data
	1. Developing and reviewing themes
	1. Defining, refining, and naming themes
	1. Writing the report

There are several approaches to data analysis that can be taken by researchers to identify functions and describe them in terms of their aspects. Data analysis using coding strategies appears to be an approach that can lend structure and transparency to the identification and description of functions. In the spirit of expanding the reach of the FRAM in the healthcare domain, future studies should ensure a detailed explanation is provided of the approach taken to identify and describe functions and build a FRAM model.

Considerations for Researchers

Time and Human Resources

A potential limitation of the FRAM is the resource and time intensive nature of the method (Damen & de Vos, 2021; Kaya et al., 2019; Laugaland et al., 2014; McNab et al., 2018; O'Hara et al., 2020). Damen et al. (2021) tracked the time required to build a work-as-done FRAM model, estimated to be 15 hr. The time needed to conduct a full FRAM analysis was tracked in four studies with recorded times of 35 to 60 hr (Damen et al., 2021; Furniss et al., 2020; O’Hara et al., 2020; Shutijser et al., 2019). Damen and de Vos (2021) estimate the full workload of a FRAM analysis to be approximately 47 hr and suggest the time decreases as researchers become more proficient. Future healthcare studies using the FRAM should consider recording the time allotted for each step of the process. Damen et al. (2021) present a table with a breakdown of the time devoted to each step using the FRAM. The data is provided to demonstrate the usability of the FRAM. The authors found the time required to conduct a FRAM analysis to be comparable to more traditional methods of analysis, such as Root Cause Analysis. Having this understanding can assist in determining the human resources required to build a FRAM model. Damen et al. (2021) conducted 20 hr of interviews with three interviewers at two sites in their study examining the process of preoperative anticoagulation management. To conduct observations, Laugaland et al. (2014) conducted 90 hr of day-of-discharge observations over several weeks and suggested the importance of having more than one researcher conducting observations to avoid potential observer bias.

Data analysis has also been a stage where more than one researcher has conducted the work. In several studies there were 2 to 4 researchers analyzing data and then coming together to review their findings and reach a consensus on constructed FRAM models (Damoiseaux-Volman et al., 2021; Laugaland et al., 2014; O’Hara et al., 2020; Salehi et al., 2021a; Sujan et al., 2022). When designing a research study using the FRAM, an understanding of the resources (human and time) required is an important consideration. Recording and sharing the time and human resources allotted to the different steps in a FRAM analysis would be an important contribution for future studies.

Participant Validation of FRAM Models (Member Checking)

After constructing an initial FRAM model, researchers often take further steps to ensure the model is an accurate representation of the process or system under examination. Several studies have described their approach to ‘FRAM model validation’ (Buikstra et al., 2020; Salehi et al., 2021a; Schreurs et al., 2022; Sujan et al., 2022; Thude et al., 2021; van Dijk et al., 2022). In qualitative research this practice is known as participant validation or member checking, which has been described as a vital strategy in establishing credibility (Lincoln & Guba, 1985). Member checking is described as a means of assessing whether an analysis faithfully or fairly represents the experiences of study participants (Braun & Clarke, 2021). Focus groups, meetings, or workshops are approaches researchers have employed where FRAM models have been returned to study participants to assess accuracy. Schreurs et al. (2021) presented their FRAM model for participant validation to stakeholders at a meeting, then adjusted the model based on feedback, and subsequently presented the revised model to stakeholders prior to model finalization.

Salehi et al., 2021a used data from focus groups to improve a constructed FRAM model of the hospital to home transition process. The focus group data added more functions and new couplings, highlighting the importance of validation prior to finalization of a model. When considering the goal of a FRAM analysis is often to provide recommendations for improving quality, efficiency, and/or safety, the model informing the analysis needs to be an accurate reflection of everyday work. FRAM model participant validation (member checking) is a key step researchers should include in their study design and can also enhance the credibility and trustworthiness of their study findings.

Building Understandable FRAM Models

Researchers need to ensure FRAM models truly capture the process of interest and the constructed model is understandable yet not overly simplified for end users, such as clinicians and administrators (McGill et al., 2022). Damen and de Vos (2021) report clinicians easily grasp the relevance, background, and design of the FRAM. Clay-Williams et al. (2015) conveys similar findings in their study reporting clinicians easily understood the visual representation of functions and additionally found the model to be a useful tool to initiate discussions. Bos et al. (2022) conducted a study examining work-as-imagined and work-as-done in pediatric follow up. The authors found the inclusion of the FRAM models in their reflection sessions with staff to be challenging as they required some effort to explain. Clay-Williams et al. (2020) suggests that FRAM models should not be overly crowded to remain a useful tool. Patriarca et al. (2018) conducted a case study examination of iatrogenic injury in the neurosurgery perioperative patient pathway using the FRAM and found the FRAM to be overwhelmingly complex with sixty-eight identified functions.

They go on to explain, a larger number of functions could potentially result in an analyst having to oversimplify the work domain or narrow the scope of an analysis to deal with the visualization and management of variability. Rather than reducing the number of functions in a process to manage the visualization of variability, Salehi et al. (2021a) organized thirty-eight functions into five colour coded functional categories of admission; assessment; synthesis; decision-making; and readmission. The organization and colour coding of functions provides some orderliness to the model and names the functional categories with terms clinicians would be familiar with in their everyday work.

Conclusion

Gaining an in-depth understanding of how everyday work is conducted in a complex healthcare process or system has been a difficult undertaking for researchers and decision-makers to date. Without a comprehensive understanding of how care is delivered and received, healthcare system improvements and sustainability efforts will continue to be difficult. The FRAM can provide this enhanced understanding by combining process modelling and qualitative inquiry. This paper drew from the FRAM literature and practical experience to examine and suggest how researchers can operationalize qualitative data collection and analysis to inform the building of a FRAM model. Considerations for researchers were also presented that highlight the need for FRAM models to be understandable to end users and for researchers to account for the time and human resources required to build a FRAM model. Future papers could expand methodology further in the healthcare domain by offering guidance on approaches to identifying and describing variability and the aggregation of variability in complex healthcare processes and systems.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research funding was provided to the first author by the Ocean Frontier Institute through an award from the Canada First Research Excellence Fund. The funders have not had any role in content development.

ORCID iDs

Alexis McGill

Vahid Salehi

References

Alm

Woltjer

(2010). Patient safety investigation through the lens of FRAM. Human factors: A system view of human, technology and organisation (pp. 153–165). Shaker Publishing.

Arcuri

Bulhões

Jatobá

Bellas

H. C.

Koster

d’Avila

A. L.

Vidal

M. C. R.

Burns

C. M.

Carvalho

P. V. R.

(2020). Gatekeeper family doctors operating a decentralized referral prioritization system: Uncovering improvements in system resilience through a grounded-based approach. Safety Science, 121, 177–190. https://doi.org/10.1016/j.ssci.2019.08.023

Bos

Roorda

de Sonnaville

van Boven

Oosterlaan

van Goudoever

Klazinga

Kringos

(2022). Implementing structured follow-up of neonatal and paediatric patients: An evaluation of three university hospital case studies using the functional resonance analysis method. BMC Health Services Research, 22(1), 191. https://doi.org/10.1186/s12913-022-07537-x

Bowen

G. A.

(2009). Document analysis as a qualitative research method. Qualitative Research Journal, 9(2), 27–40. https://doi.org/10.3316/QRJ0902027

Braithwaite

Wears

R. L.

Hollnagel

(2015). Resilient health care: Turning patient safety on its head. International Journal for Quality in Health Care, 27(5), 418–420. https://doi.org/10.1093/intqhc/mzv063

Braun

Clarke

(2021). One size fits all? What counts as quality practice in (reflexive) thematic analysis? Qualitative Research in Psychology, 18(3), 328–352. https://doi.org/10.1080/14780887.2020.1769238

Braun

Clarke

(n.d.). Thematic analysis. Retrieved 18, 08, 2023 from. https://www.thematicanalysis.net/

Buikstra

Clay-Williams

Strivens

(2021). Modelling a typical patient journey through the geriatric evaluation and management ward to better understand discharge planning processes. In Resilient health care (pp. 81–100). CRC Press. https://doi.org/10.4324/9781003095224-12

Buikstra

Strivens

Clay-Williams

(2020). Understanding variability in discharge planning processes for the older person. Safety Science, 121, 137–146. https://doi.org/10.1016/j.ssci.2019.08.026

10.

Carter

Bryant-Lukosius

DiCenso

Blythe

Neville

A. J.

(2014). The use of triangulation in qualitative research. Oncology Nursing Forum, 41(5), 545–547. https://doi.org/10.1188/14.Onf.545-547

11.

Clarke

Braun

(2013). Successful qualitative research: A practical guide for beginners (pp. 1–400). Sage Publications.

12.

Clarke

Braun

(2021). Thematic analysis: A practical guide. Sage Publications.

13.

Clay-Williams

Austin

Braithwaite

Hollnagel

(2020). Qualitative assessment to improve everyday activities: Work-as-imagined and work-as-done. In Transforming healthcare with qualitative research (pp. 71–82). Routledge. https://doi.org/10.4324/9780429299797-11

14.

Clay-Williams

Hounsgaard

Hollnagel

(2015). Where the rubber meets the road: Using FRAM to align work-as-imagined with work-as-done when implementing clinical guidelines. Implementation Science, 10(1), 125–128. https://doi.org/10.1186/s13012-015-0317-y

15.

Damen

N. L.

de Vos

M. S.

(2021). Experiences with FRAM in Du##tch hospitals: Muddling through with models. In Resilient health care (pp. 71–80). CRC Press. https://doi.org/10.4324/9781003095224-11

16.

Damen

N. L.

de Vos

M. S.

Moesker

M. J.

Braithwaite

de Lind van Wijngaarden

R. A. d. L.

Kaplan

Hamming

J. F.

Clay-Williams

(2021). Preoperative anticoagulation management in everyday clinical practice: An international comparative analysis of work-as-done using the functional resonance analysis method. Journal of Patient Safety, 17(3), 157–165. https://doi.org/10.1097/PTS.0000000000000515

17.

Damoiseaux-Volman

B. A.

Medlock

van der Eijk

M. D.

Romijn

J. A.

Abu-Hanna

van der Velde

(2021). Falls and delirium in older inpatients: Work-as-imagined, work-as-done and preferences for clinical decision support systems. Safety Science, 142, 105355. https://doi.org/10.1016/j.ssci.2021.105355.

18.

DeJonckheere

Vaughn

L. M.

(2019). Semistructured interviewing in primary care research: A balance of relationship and rigour. Family medicine and community health, 7(2), e000057. https://doi.org/10.1136/fmch-2018-000057

19.

Furniss

Nelson

Habli

White

Elliott

Reynolds

Sujan

(2020). Using FRAM to explore sources of performance variability in intravenous infusion administration in ICU: A non-normative approach to systems contradictions. Applied Ergonomics, 86, 103113. https://doi.org/10.1016/j.apergo.2020.103113

20.

Greenhalgh

Papoutsi

(2018). Studying complexity in health services research: Desperately seeking an overdue paradigm shift. In (Vol. 16, pp. 1–6): Springer. https://doi.org/10.1186/s12916-018-1089-4

21.

Gustafson

Vollam

Morgan

Watkinson

(2021). A human factors analysis of missed mobilisation after discharge from intensive care: A competition for care? Physiotherapy, 113, 131–137. https://doi.org/10.1016/j.physio.2021.03.013

22.

Hedqvist

A.-T.

Praetorius

Ekstedt

(2022). Exploring interdependencies, vulnerabilities, gaps and bridges in care transitions of patients with complex care needs using the Functional Resonance Analysis Method.

23.

Hollnagel

(2012). FRAM, the functional resonance analysis method: Modelling complex socio-technical systems. Ashgate Publishing, Ltd.

24.

Hollnagel

Hill

(2020). FRAM model visualiser instructions. https://functionalresonance.com/onewebmedia/FMV_instructions_2.1.pdf

25.

Hollnagel

Hounsgaard

Colligan

(2014). FRAM-the Functional resonance analysis method: A handbook for the practical use of the method. Centre for Quality, Region of Southern Denmark.

26.

Hollnagel

Slater

(2022). FRAMSYNT A FRAM handbook. https://www.researchgate.net/profile/David-Slater/publication/364959115_A_FRAM_HANDBOOK/links/63611d2f6e0d367d91e7b7e2/A-FRAM-HANDBOOK.pdf

27.

Jatobá

Bellas

Viana

de Castro Nunes

Leal

Bulhões

Arcuri

de Carvalho

P. V. R.

(2022). Unveiling conflicting strategies in the Brazilian response to COVID-19: A cross-sectional study using the functional resonance analysis method. Dialogues in Health, 1, 100056. https://doi.org/10.1016/j.dialog.2022.100056

28.

Kaya

G. K.

Ovali

H. F.

Ozturk

(2019). Using the functional resonance analysis method on the drug administration process to assess performance variability. Safety Science, 118, 835–840. https://doi.org/10.1016/j.ssci.2019.06.020

29.

Laugaland

Aase

Waring

(2014). Hospital discharge of the elderly-an observational case study of functions, variability and performance-shaping factors. BMC Health Services Research, 14(1), 365–415. https://doi.org/10.1186/1472-6963-14-365

30.

Liamputtong

(2011). Focus group methodology: Principle and practice. Sage publications.

31.

Lincoln

Y. S.

Guba

E. G.

(1985). Naturalistic inquiry. Newberry Park.

32.

Linneberg

M. S.

Korsgaard

(2019). Coding qualitative data: A synthesis guiding the novice. Qualitative Research Journal, 19(3), 259–270.

33.

Lumivero . (2020). NVivo. In www.lumivero.com

34.

MacKinnon

R. J.

Pukk-Härenstam

Kennedy

Hollnagel

Slater

(2021). A novel approach to explore Safety-I and Safety-II perspectives in in situ simulations—the structured what if functional resonance analysis methodology. Advances in Simulation, 6(1), 21. https://doi.org/10.1186/s41077-021-00166-0

35.

McGill

Smith

McCloskey

Morris

Goudreau

Veitch

(2022). The functional resonance analysis method as a health care research methodology: A scoping review. JBI Evidence Synthesis, 20(4), 1074–1097. https://doi.org/10.11124/JBIES-21-00099

36.

McNab

Freestone

Black

Carson-Stevens

Bowie

(2018). Participatory design of an improvement intervention for the primary care management of possible sepsis using the functional resonance analysis method. BMC Medicine, 16(1), 174–220. https://doi.org/10.1186/s12916-018-1164-x

37.

Moser

Korstjens

(2017). Series: Practical guidance to qualitative research. Part 1: Introduction. The European Journal of General Practice, 23(1), 271–273. https://doi.org/10.1080/13814788.2017.1375093

38.

Moser

Korstjens

(2018). Series: Practical guidance to qualitative research. Part 3: Sampling, data collection and analysis. The European Journal of General Practice, 24(1), 9–18. https://doi.org/10.1080/13814788.2017.1375091

39.

Oduyale

M. S.

Patel

Borthwick

Claus

(2020). Co‐administration of multiple intravenous medicines: Intensive care nurses’ views and perspectives. Nursing in Critical Care, 25(3), 156–164. https://doi.org/10.1111/nicc.12497

40.

O’Hara

J. K.

Baxter

Hardicre

(2020). ‘Handing over to the patient’: A FRAM analysis of transitional care combining multiple stakeholder perspectives. Applied Ergonomics, 85, 103060. https://doi.org/10.1016/j.apergo.2020.103060

41.

Patriarca

Di Gravio

Woltjer

Costantino

Praetorius

Ferreira

Hollnagel

(2020). Framing the FRAM: A literature review on the functional resonance analysis method. Safety Science, 129, 104827. https://doi.org/10.1016/j.ssci.2020.104827

42.

Patriarca

Falegnami

Costantino

Bilotta

(2018). Resilience engineering for socio-technical risk analysis: Application in neuro-surgery. Reliability Engineering and System Safety, 180, 321–335. https://doi.org/10.1016/j.ress.2018.08.001

43.

Pickup

Atkinson

Hollnagel

Bowie

Gray

Rawlinson

Forrester

(2017). Blood sampling - two sides to the story. Applied Ergonomics, 59(Pt A), 234–242. https://doi.org/10.1016/j.apergo.2016.08.027

44.

Polit

Beck

(2020). Essentials of nursing research: Appraising evidence for nursing practice. Lippincott Williams and Wilkins.

45.

Raben

D. C.

Viskum

Mikkelsen

K. L.

Hounsgaard

Bogh

S. B.

Hollnagel

(2018). Application of a non-linear model to understand healthcare processes: Using the functional resonance analysis method on a case study of the early detection of sepsis. Reliability Engineering and System Safety, 177, 1–11. https://doi.org/10.1016/j.ress.2018.04.023

46.

Ross

Sherriff

Kidd

Gnich

Anderson

Deas

Macpherson

(2018). A systems approach using the functional resonance analysis method to support fluoride varnish application for children attending general dental practice. Applied Ergonomics, 68, 294–303. https://doi.org/10.1016/j.apergo.2017.12.005

47.

Salehi

Hanson

Smith

McCloskey

Jarrett

Veitch

(2021a). Modeling and analyzing hospital to home transition processes of frail older adults using the functional resonance analysis method (FRAM). Applied Ergonomics, 93, 103392. https://doi.org/10.1016/j.apergo.2021.103392

48.

Salehi

Veitch

Smith

(2021b). Modeling complex socio‐technical systems using the FRAM: A literature review. Human factors and ergonomics in manufacturing & service industries, 31(1), 118–142. https://doi.org/10.1002/hfm.20874

49.

Saunders

Sim

Kingstone

Baker

Waterfield

Bartlam

Burroughs

Jinks

(2018). Saturation in qualitative research: Exploring its conceptualization and operationalization. Quality and Quantity, 52, 1893–1907. https://doi.org/10.1007/s11135-017-0574-8

50.

Schreurs

R. H. P.

Joore

M. A.

Ten Cate

Ten Cate-Hoek

A. J.

(2021). Using the functional resonance analysis method to explore how elastic compression therapy is organised and could be improved from a multistakeholder perspective. BMJ Open, 11(10), e048331. https://doi.org/10.1136/bmjopen-2020-048331

51.

Schutijser

B. C. F. M.

Jongerden

I. P.

Klopotowska

J. E.

Portegijs

de Bruijne

M. C.

Wagner

(2019). Double checking injectable medication administration: Does the protocol fit clinical practice? Safety Science, 118, 853–860. https://doi.org/10.1016/j.ssci.2019.06.026

52.

Shorrock

(2020). Proxies for work-as-done: 3. Work-as-Disclosed. https://humanisticsystems.com/2020/11/01/proxies-for-work-as-done-3-work-as-disclosed/

53.

Slater

(2022). A systems analysis of the UK COVID 19 pandemic response: Part 2-Work as imagined vs work as done. Safety Science, 146, 105526. https://doi.org/10.1016/j.ssci.2021.105526

54.

Slater

Hollnagel

MacKinnon

Sujan

Carson-Stevens

Ross

Bowie

(2022). A systems analysis of the COVID-19 pandemic response in the United Kingdom–Part 1–The overall context. Safety Science, 146, 105525. https://doi.org/10.1016/j.ssci.2021.105525

55.

Sujan

Bilbro

Ross

Earl

Ibrahim

Bond-Smith

Ghaferi

Pickup

McCulloch

(2022). Failure to rescue following emergency surgery: A FRAM analysis of the management of the deteriorating patient. Applied Ergonomics, 98, 103608. https://doi.org/10.1016/j.apergo.2021.103608

56.

Sujan

M. A.

(2021). Muddling through in the intensive care unit: A FRAM analysis of intravenous infusion management. In Resilient health care (pp. 101–107). CRC Press. https://doi.org/10.4324/9781003095224-13

57.

Thude

B. R.

Brink

Hansen

M. S.

Morsø

(2021). How to ensure referral and uptake for COPD rehabilitation—Part 1: Disentangling factors in the cross-sectorial workflow of patients with COPD to understand why most patients are not referred to rehabilitation. International Journal of Integrated Care, 21(1), 1–9. https://doi.org/10.5334/ijic.5502

58.

Tresfon

Brunsveld-Reinders

A. H.

van Valkenburg

Langeveld

Hamming

(2022). Aligning work-as-imagined and work-as-done using FRAM on a hospital ward: A roadmap. BMJ Open Quality, 11(4), e001992. https://doi.org/10.1136/bmjoq-2022-001992

59.

van Dijk

L. M.

Meulman

M. D.

van Eikenhorst

Merten

Schutijser

B. C.

Wagner

(2021). Can using the functional resonance analysis method, as an intervention, improve patient safety in hospitals? A stepped wedge design protocol. BMC Health Services Research, 21(1), 1228–1310. https://doi.org/10.1186/s12913-021-07244-z

60.

van Dijk

L. M.

van Eikenhorst

Wagner

(2022). Daily practice performance (Work-as-Done) compared to guidelines (Work-as-Imagined) of medication reconciliation at discharge: Outcomes of a FRAM study. Safety Science, 155, 105871. https://doi.org/10.1016/j.ssci.2022.105871

61.

Watson

Rahul

Dominic

Higgs

Williamson

Turner

(2022). A human factors approach to quality improvement in oxygen prescribing. Clinical Medicine, 22(2), 153–159. https://doi.org/10.7861/clinmed.2021-0164

62.

Weston

L. E.

Krein

Harrod

(2021). Using observation to better understand the healthcare context. Qualitative Research in Medicine and Healthcare, 5(3), 9821. https://doi.org/10.4081/qrmh.2021.9821

63.

Yin

R. K.

(2016). Qualitative research from start to finish. Guilford publications.