Practical thematic analysis: a guide for multidisciplinary health services research teams engaging in qualitative analysisBMJ 2023; 381 doi: https://doi.org/10.1136/bmj-2022-074256 (Published 08 June 2023) Cite this as: BMJ 2023;381:e074256
- Catherine H Saunders, scientist and assistant professor1 2,
- Ailyn Sierpe, research project coordinator2,
- Christian von Plessen, senior physician3,
- Alice M Kennedy, research project manager2 4,
- Laura C Leviton, senior adviser5,
- Steven L Bernstein, chief research officer1,
- Jenaya Goldwag, resident physician1,
- Joel R King, research assistant2,
- Christine M Marx, patient associate6,
- Jacqueline A Pogue, research project manager2,
- Richard K Saunders, staff physician1,
- Aricca Van Citters, senior research scientist2,
- Renata W Yen, doctoral student2,
- Glyn Elwyn, professor2,
- JoAnna K Leyenaar, associate professor1 2
- on behalf of the Coproduction Laboratory
- 1Dartmouth Health, Lebanon, NH, USA
- 2Dartmouth Institute for Health Policy and Clinical Practice, Geisel School of Medicine at Dartmouth College, Lebanon, NH, USA
- 3Center for Primary Care and Public Health (Unisanté), Lausanne, Switzerland
- 4Jönköping Academy for Improvement of Health and Welfare, School of Health and Welfare, Jönköping University, Jönköping, Sweden
- 5Highland Park, NJ, USA
- 6Division of Public Health Sciences, Department of Surgery, Washington University School of Medicine, St Louis, MO, USA
- Correspondence to: C H Saunders
- Accepted 26 April 2023
Through qualitative methods, researchers can provide deep contextual understanding of real world issues, and generate new knowledge to inform hypotheses, theories, research, and clinical care. Approaches to data collection are varied, including interviews, focus groups, structured observation, and analysis of multimedia data, with qualitative research questions aimed at understanding the how and why of human experience.12 Qualitative methods produce unique insights in applied health services research that other approaches cannot deliver. In particular, researchers acknowledge that thematic analysis is a flexible and powerful method of systematically generating robust qualitative research findings by identifying, analysing, and reporting patterns (themes) within data.3456 Although qualitative methods are increasingly valued for answering clinical research questions, many researchers are unsure how to apply them or consider them too time consuming to be useful in responding to practical challenges7 or pressing situations such as public health emergencies.8 Consequently, researchers might hesitate to use them, or use them improperly.91011
Although much has been written about how to perform thematic analysis, practical guidance for non-specialists is sparse.3561213 In the multidisciplinary field of health services research, qualitative data analysis can confound experienced researchers and novices alike, which can stoke concerns about rigor, particularly for those more familiar with quantitative approaches.14 Since qualitative methods are an area of specialisation, support from experts is beneficial. However, because non-specialist perspectives can enhance data interpretation and enrich findings, there is a case for making thematic analysis easier, more rapid, and more efficient,8 particularly for patients, care partners, clinicians, and other stakeholders. A practical guide to thematic analysis might encourage those on the ground to use these methods in their work, unearthing insights that would otherwise remain undiscovered.
Given the need for more accessible qualitative analysis approaches, we present a simple, rigorous, and efficient three step guide for practical thematic analysis. We include new guidance on the mechanics of thematic analysis, including developing codes, constructing meaningful themes, and hosting a thematic analysis session. We also discuss common pitfalls in thematic analysis and how to avoid them.
Qualitative methods are increasingly valued in applied health services research, but multidisciplinary research teams often lack accessible step-by-step guidance and might struggle to use these approaches
A newly developed approach, practical thematic analysis, uses three simple steps: reading, coding, and theming
Based on Braun and Clarke’s reflexive thematic analysis, our streamlined yet rigorous approach is designed for multidisciplinary health services research teams, including patients, care partners, and clinicians
This article also provides companion materials including a slide presentation for teaching practical thematic analysis to research teams, a sample thematic analysis session agenda, a theme coproduction template for use during the session, and guidance on using standardised reporting criteria for qualitative research
In their seminal work, Braun and Clarke developed a six phase approach to reflexive thematic analysis.412 We built on their method to develop practical thematic analysis (box 1, fig 1), which is a simplified and instructive approach that retains the substantive elements of their six phases. Braun and Clarke’s phase 1 (familiarising yourself with the dataset) is represented in our first step of reading. Phase 2 (coding) remains as our second step of coding. Phases 3 (generating initial themes), 4 (developing and reviewing themes), and 5 (refining, defining, and naming themes) are represented in our third step of theming. Phase 6 (writing up) also occurs during this third step of theming, but after a thematic analysis session.412
Key features and applications of practical thematic analysis
Step 1: Reading
All manuscript authors read the data
All manuscript authors write summary memos
Step 2: Coding
Coders perform both data management and early data analysis
Codes are complete thoughts or sentences, not categories
Step 3: Theming
Researchers host a thematic analysis session and share different perspectives
Themes are complete thoughts or sentences, not categories
For use by practicing clinicians, patients and care partners, students, interdisciplinary teams, and those new to qualitative research
When important insights from healthcare professionals are inaccessible because they do not have qualitative methods training
When time and resources are limited
We present linear steps, but as qualitative research is usually iterative, so too is thematic analysis.15 Qualitative researchers circle back to earlier work to check whether their interpretations still make sense in the light of additional insights, adapting as necessary. While we focus here on the practical application of thematic analysis in health services research, we recognise our approach exists in the context of the broader literature on thematic analysis and the theoretical underpinnings of qualitative methods as a whole. For a more detailed discussion of these theoretical points, as well as other methods widely used in health services research, we recommend reviewing the sources outlined in supplemental material 1. A strong and nuanced understanding of the context and underlying principles of thematic analysis will allow for higher quality research.16
Practical thematic analysis is a highly flexible approach that can draw out valuable findings and generate new hypotheses, including in cases with a lack of previous research to build on. The approach can also be used with a variety of data, such as transcripts from interviews or focus groups, patient encounter transcripts, professional publications, observational field notes, and online activity logs. Importantly, successful practical thematic analysis is predicated on having high quality data collected with rigorous methods. We do not describe qualitative research design or data collection here.1117
In supplemental material 1, we summarise the foundational methods, concepts, and terminology in qualitative research. Along with our guide below, we include a companion slide presentation for teaching practical thematic analysis to research teams in supplemental material 2. We provide a theme coproduction template for teams to use during thematic analysis sessions in supplemental material 3. Our method aligns with the major qualitative reporting frameworks, including the Consolidated Criteria for Reporting Qualitative Research (COREQ).18 We indicate the corresponding step in practical thematic analysis for each COREQ item in supplemental material 4.
Step 1: Reading
Familiarisation and memoing
We encourage all manuscript authors to review the full dataset (eg, interview transcripts) to familiarise themselves with it. This task is most critical for those who will later be engaged in the coding and theming steps. Although time consuming, it is the best way to involve team members in the intellectual work of data interpretation, so that they can contribute to the analysis and contextualise the results. If this task is not feasible given time limitations or large quantities of data, the data can be divided across team members. In this case, each piece of data should be read by at least two individuals who ideally represent different professional roles or perspectives.
We recommend that researchers reflect on the data and independently write memos, defined as brief notes on thoughts and questions that arise during reading, and a summary of their impressions of the dataset.219 Memoing is an opportunity to gain insights from varying perspectives, particularly from patients, care partners, clinicians, and others. It also gives researchers the opportunity to begin to scope which elements of and concepts in the dataset are relevant to the research question.
The concept of data saturation (box 2) is a foundation of qualitative research. It is defined as the point in analysis at which new data tend to be redundant of data already collected.21 Qualitative researchers are expected to report their approach to data saturation.18 Because thematic analysis is iterative, the team should discuss saturation throughout the entire process, beginning with data collection and continuing through all steps of the analysis.22 During step 1 (reading), team members might discuss data saturation in the context of summary memos. Conversations about saturation continue during step 2 (coding), with confirmation that saturation has been achieved during step 3 (theming). As a rule of thumb, researchers can often achieve saturation in 9-17 interviews or 4-8 focus groups, but this will vary depending on the specific characteristics of the study.23
Data saturation in context
Braun and Clarke discourage the use of data saturation to determine sample size (eg, number of interviews), because it assumes that there is an objective truth to be captured in the data (sometimes known as a positivist perspective).20 Qualitative researchers often try to avoid positivist approaches, arguing that there is no one true way of seeing the world, and will instead aim to gather multiple perspectives.5 Although this theoretical debate with qualitative methods is important, we recognise that a priori estimates of saturation are often needed, particularly for investigators newer to qualitative research who might want a more pragmatic and applied approach. In addition, saturation based, sample size estimation can be particularly helpful in grant proposals. However, researchers should still follow a priori sample size estimation with a discussion to confirm saturation has been achieved.RETURN TO TEXT
Step 2: Coding
Definition of coding
We describe codes as labels for concepts in the data that are directly relevant to the study objective. Historically, the purpose of coding was to distil the large amount of data collected into conceptually similar buckets so that researchers could review it in aggregate and identify key themes.524 We advocate for a more analytical approach than is typical with thematic analysis. With our method, coding is both the foundation for and the beginning of thematic analysis—that is, early data analysis, management, and reduction occur simultaneously rather than as different steps. This approach moves the team more efficiently towards being able to describe themes.
Building the coding team
Coders are the research team members who directly assign codes to the data, reading all material and systematically labelling relevant data with appropriate codes. Ideally, at least two researchers would code every discrete data document, such as one interview transcript.25 If this task is not possible, individual coders can each code a subset of the data that is carefully selected for key characteristics (sometimes known as purposive selection).26 When using this approach, we recommend that at least 10% of data be coded by two or more coders to ensure consistency in codebook application. We also recommend coding teams of no more than four to five people, for practical reasons concerning maintaining consistency.
Clinicians, patients, and care partners bring unique perspectives to coding and enrich the analytical process.27 Therefore, we recommend choosing coders with a mix of relevant experiences so that they can challenge and contextualise each other’s interpretations based on their own perspectives and opinions (box 3). We recommend including both coders who collected the data and those who are naive to it, if possible, given their different perspectives. We also recommend all coders review the summary memos from the reading step so that key concepts identified by those not involved in coding can be integrated into the analytical process. In practice, this review means coding the memos themselves and discussing them during the code development process. This approach ensures that the team considers a diversity of perspectives.
Coding teams in context
The recommendation to use multiple coders is a departure from Braun and Clarke.2829 When the views, experiences, and training of each coder (sometimes known as positionality)30 are carefully considered, having multiple coders can enhance interpretation and enrich findings. When these perspectives are combined in a team setting, researchers can create shared meaning from the data. Along with the practical consideration of distributing the workload,31 inclusion of these multiple perspectives increases the overall quality of the analysis by mitigating the impact of any one coder’s perspective.30RETURN TO TEXT
Qualitative analysis software facilitates coding and managing large datasets but does not perform the analytical work. The researchers must perform the analysis themselves. Most programs support queries and collaborative coding by multiple users.32 Important factors to consider when choosing software can include accessibility, cost, interoperability, the look and feel of code reports, and the ease of colour coding and merging codes. Coders can also use low tech solutions, including highlighters, word processors, or spreadsheets.
Drafting effective codes
To draft effective codes, we recommend that the coders review each document line by line.33 As they progress, they can assign codes to segments of data representing passages of interest.34 Coders can also assign multiple codes to the same passage. Consensus among coders on what constitutes a minimum or maximum amount of text for assigning a code is helpful. As a general rule, meaningful segments of text for coding are shorter than one paragraph, but longer than a few words. Coders should keep the study objective in mind when determining which data are relevant (box 4).
Code types in context
Similar to Braun and Clarke’s approach, practical thematic analysis does not specify whether codes are based on what is evident from the data (sometimes known as semantic) or whether they are based on what can be inferred at a deeper level from the data (sometimes known as latent).41235 It also does not specify whether they are derived from the data (sometimes known as inductive) or determined ahead of time (sometimes known as deductive).1135 Instead, it should be noted that health services researchers conducting qualitative studies often adopt all these approaches to coding (sometimes known as hybrid analysis).3RETURN TO TEXT
In practical thematic analysis, codes should be more descriptive than general categorical labels that simply group data with shared characteristics. At a minimum, codes should form a complete (or full) thought. An easy way to conceptualise full thought codes is as complete sentences with subjects and verbs (table 1), although full sentence coding is not always necessary. With full thought codes, researchers think about the data more deeply and capture this insight in the codes. This coding facilitates the entire analytical process and is especially valuable when moving from codes to broader themes. Experienced qualitative researchers often intuitively use full thought or sentence codes, but this practice has not been explicitly articulated as a path to higher quality coding elsewhere in the literature.6
Depending on the nature of the data, codes might either fall into flat categories or be arranged hierarchically. Flat categories are most common when the data deal with topics on the same conceptual level. In other words, one topic is not a subset of another topic. By contrast, hierarchical codes are more appropriate for concepts that naturally fall above or below each other. Hierarchical coding can also be a useful form of data management and might be necessary when working with a large or complex dataset.5 Codes grouped into these categories can also make it easier to naturally transition into generating themes from the initial codes.5 These decisions between flat versus hierarchical coding are part of the work of the coding team. In both cases, coders should ensure that their code structures are guided by their research questions.
Developing the codebook
A codebook is a shared document that lists code labels and comprehensive descriptions for each code, as well as examples observed within the data. Good code descriptions are precise and specific so that coders can consistently assign the same codes to relevant data or articulate why another coder would do so. Codebook development is iterative and involves input from the entire coding team. However, as those closest to the data, coders must resist undue influence, real or perceived, from other team members with conflicting opinions—it is important to mitigate the risk that more senior researchers, like principal investigators, exert undue influence on the coders’ perspectives.
In practical thematic analysis, coders begin codebook development by independently coding a small portion of the data, such as two to three transcripts or other units of analysis. Coders then individually produce their initial codebooks. This task will require them to reflect on, organise, and clarify codes. The coders then meet to reconcile the draft codebooks, which can often be difficult, as some coders tend to lump several concepts together while others will split them into more specific codes. Discussing disagreements and negotiating consensus are necessary parts of early data analysis. Once the codebook is relatively stable, we recommend soliciting input on the codes from all manuscript authors. Yet, coders must ultimately be empowered to finalise the details so that they are comfortable working with the codebook across a large quantity of data.
Assigning codes to the data
After developing the codebook, coders will use it to assign codes to the remaining data. While the codebook’s overall structure should remain constant, coders might continue to add codes corresponding to any new concepts observed in the data. If new codes are added, coders should review the data they have already coded and determine whether the new codes apply. Qualitative data analysis software can be useful for editing or merging codes.
We recommend that coders periodically compare their code occurrences (box 5), with more frequent check-ins if substantial disagreements occur. In the event of large discrepancies in the codes assigned, coders should revise the codebook to ensure that code descriptions are sufficiently clear and comprehensive to support coding alignment going forward. Because coding is an iterative process, the team can adjust the codebook as needed.52829
Quantitative coding in context
Researchers should generally avoid reporting code counts in thematic analysis. However, counts can be a useful proxy in maintaining alignment between coders on key concepts.26 In practice, therefore, researchers should make sure that all coders working on the same piece of data assign the same codes with a similar pattern and that their memoing and overall assessment of the data are aligned.37 However, the frequency of a code alone is not an indicator of its importance. It is more important that coders agree on the most salient points in the data; reviewing and discussing summary memos can be helpful here.5
Researchers might disagree on whether or not to calculate and report inter-rater reliability. We note that quantitative tests for agreement, such as kappa statistics or intraclass correlation coefficients, can be distracting and might not provide meaningful results in qualitative analyses. Similarly, Braun and Clarke argue that expecting perfect alignment on coding is inconsistent with the goal of co-constructing meaning.2829 Overall consensus on codes’ salience and contributions to themes is the most important factor.RETURN TO TEXT
Step 3: Theming
Definition of themes
Themes are meta-constructs that rise above codes and unite the dataset (box 6, fig 2). They should be clearly evident, repeated throughout the dataset, and relevant to the research questions.38 While codes are often explicit descriptions of the content in the dataset, themes are usually more conceptual and knit the codes together.39 Some researchers hypothesise that theme development is loosely described in the literature because qualitative researchers simply intuit themes during the analytical process.39 In practical thematic analysis, we offer a concrete process that should make developing meaningful themes straightforward.
Themes in context
According to Braun and Clarke, a theme “captures something important about the data in relation to the research question and represents some level of patterned response or meaning within the data set.”4 Similarly, Braun and Clarke advise against themes as domain summaries. While different approaches can draw out themes from codes, the process begins by identifying patterns.2835 Like Braun and Clarke and others, we recommend that researchers consider the salience of certain themes, their prevalence in the dataset, and their keyness (ie, how relevant the themes are to the overarching research questions).41234RETURN TO TEXT
Constructing meaningful themes
After coding all the data, each coder should independently reflect on the team’s summary memos (step 1), the codebook (step 2), and the coded data itself to develop draft themes (step 3). It can be illuminating for coders to review all excerpts associated with each code, so that they derive themes directly from the data. Researchers should remain focused on the research question during this step, so that themes have a clear relation with the overall project aim. Use of qualitative analysis software will make it easy to view each segment of data tagged with each code. Themes might neatly correspond to groups of codes. Or—more likely—they will unite codes and data in unexpected ways. A whiteboard or presentation slides might be helpful to organise, craft, and revise themes. We also provide a template for coproducing themes (supplemental material 3). As with codebook justification, team members will ideally produce individual drafts of the themes that they have identified in the data. They can then discuss these with the group and reach alignment or consensus on the final themes.
The team should ensure that all themes are salient, meaning that they are: supported by the data, relevant to the study objectives, and important. Similar to codes, themes are framed as complete thoughts or sentences, not categories. While codes and themes might appear to be similar to each other, the key distinction is that the themes represent a broader concept. Table 2 shows examples of codes and their corresponding themes from a previously published project that used practical thematic analysis.36 Identifying three to four key themes that comprise a broader overarching theme is a useful approach. Themes can also have subthemes, if appropriate.4041424344
Thematic analysis session
After each coder has independently produced draft themes, a carefully selected subset of the manuscript team meets for a thematic analysis session (table 3). The purpose of this session is to discuss and reach alignment or consensus on the final themes. We recommend a session of three to five hours, either in-person or virtually.
The composition of the thematic analysis session team is important, as each person’s perspectives will shape the results. This group is usually a small subset of the broader research team, with three to seven individuals. We recommend that primary and senior authors work together to include people with diverse experiences related to the research topic. They should aim for a range of personalities and professional identities, particularly those of clinicians, trainees, patients, and care partners. At a minimum, all coders and primary and senior authors should participate in the thematic analysis session.
The session begins with each coder presenting their draft themes with supporting quotes from the data.5 Through respectful and collaborative deliberation, the group will develop a shared set of final themes.
One team member facilitates the session. A firm, confident, and consistent facilitation style with good listening skills is critical. For practical reasons, this person is not usually one of the primary coders. Hierarchies in teams cannot be entirely flattened, but acknowledging them and appointing an external facilitator can reduce their impact. The facilitator can ensure that all voices are heard. For example, they might ask for perspectives from patient partners or more junior researchers, and follow up on comments from senior researchers to say, “We have heard your perspective and it is important; we want to make sure all perspectives in the room are equally considered.” Or, “I hear [senior person] is offering [x] idea, I’d like to hear other perspectives in the room.” The role of the facilitator is critical in the thematic analysis session. The facilitator might also privately discuss with more senior researchers, such as principal investigators and senior authors, the importance of being aware of their influence over others and respecting and eliciting the perspectives of more junior researchers, such as patients, care partners, and students.
To our knowledge, this discrete thematic analysis session is a novel contribution of practical thematic analysis. It helps efficiently incorporate diverse perspectives using the session agenda and theme coproduction template (supplemental material 3) and makes the process of constructing themes transparent to the entire research team.
Writing the report
We recommend beginning the results narrative with a summary of all relevant themes emerging from the analysis, followed by a subheading for each theme. Each subsection begins with a brief description of the theme and is illustrated with relevant quotes, which are contextualised and explained. The write-up should not simply be a list, but should contain meaningful analysis and insight from the researchers, including descriptions of how different stakeholders might have experienced a particular situation differently or unexpectedly.
In addition to weaving quotes into the results narrative, quotes can be presented in a table. This strategy is a particularly helpful when submitting to clinical journals with tight word count limitations. Quote tables might also be effective in illustrating areas of agreement and disagreement across stakeholder groups, with columns representing different groups and rows representing each theme or subtheme. Quotes should include an anonymous label for each participant and any relevant characteristics, such as role or gender. The aim is to produce rich descriptions.5 We recommend against repeating quotations across multiple themes in the report, so as to avoid confusion. The template for coproducing themes (supplemental material 3) allows documentation of quotes supporting each theme, which might also be useful during report writing.
Visual illustrations such as a thematic map or figure of the findings can help communicate themes efficiently.4364244 If a figure is not possible, a simple list can suffice.36 Both must clearly present the main themes with subthemes. Thematic figures can facilitate confirmation that the researchers’ interpretations reflect the study populations’ perspectives (sometimes known as member checking), because authors can invite discussions about the figure and descriptions of findings and supporting quotes.46 This process can enhance the validity of the results.46
In supplemental material 4, we provide additional guidance on reporting thematic analysis consistent with COREQ.18 Commonly used in health services research, COREQ outlines a standardised list of items to be included in qualitative research reports (box 7).
Reporting in context
We note that use of COREQ or any other reporting guidelines does not in itself produce high quality work and should not be used as a substitute for general methodological rigor. Rather, researchers must consider rigor throughout the entire research process. As the issue of how to conceptualise and achieve rigorous qualitative research continues to be debated,4748 we encourage researchers to explicitly discuss how they have looked at methodological rigor in their reports. Specifically, we point researchers to Braun and Clarke’s 2021 tool for evaluating thematic analysis manuscripts for publication (“Twenty questions to guide assessment of TA [thematic analysis] research quality”).16RETURN TO TEXT
Avoiding common pitfalls
Awareness of common mistakes can help researchers avoid improper use of qualitative methods. Improper use can, for example, prevent researchers from developing meaningful themes and can risk drawing inappropriate conclusions from the data. Braun and Clarke also warn of poor quality in qualitative research, noting that “coherence and integrity of published research does not always hold.”16
An important distinction between high and low quality themes is that high quality themes are descriptive and complete thoughts. As such, they often contain subjects and verbs, and can be expressed as full sentences (table 2). Themes that are simply descriptive categories or topics could fail to impart meaningful knowledge beyond categorisation.164950
Researchers will often move from coding directly to writing up themes, without performing the work of theming or hosting a thematic analysis session. Skipping concerted theming often results in themes that look more like categories than unifying threads across the data.
Because data collection for qualitative research is often semi-structured (eg, interviews, focus groups), not all data will be directly relevant to the research question at hand. To avoid unfocused analysis and a correspondingly unfocused manuscript, we recommend that all team members keep the research objective in front of them at every stage, from reading to coding to theming. During the thematic analysis session, we recommend that the research question be written on a whiteboard so that all team members can refer back to it, and so that the facilitator can ensure that conversations about themes occur in the context of this question. Consistently focusing on the research question can help to ensure that the final report directly answers it, as opposed to the many other interesting insights that might emerge during the qualitative research process. Such insights can be picked up in a secondary analysis if desired.
Presenting findings quantitatively (eg, “We found 18 instances of participants mentioning safety concerns about the vaccines”) is generally undesirable in practical thematic analysis reporting.51 Descriptive terms are more appropriate (eg, “participants had substantial concerns about the vaccines,” or “several participants were concerned about this”). This descriptive presentation is critical because qualitative data might not be consistently elicited across participants, meaning that some individuals might share certain information while others do not, simply based on how conversations evolve. Additionally, qualitative research does not aim to draw inferences outside its specific sample. Emphasising numbers in thematic analysis can lead to readers incorrectly generalising the findings. Although peer reviewers unfamiliar with thematic analysis often request this type of quantification, practitioners of practical thematic analysis can confidently defend their decision to avoid it. If quantification is methodologically important, we recommend simultaneously conducting a survey or incorporating standardised interview techniques into the interview guide.11
Neglecting group dynamics
Researchers should concertedly consider group dynamics in the research team. Particular attention should be paid to power relations and the personality of team members, which can include aspects such as who most often speaks, who defines concepts, and who resolves disagreements that might arise within the group.52
The perspectives of patient and care partners are particularly important to cultivate. Ideally, patient partners are meaningfully embedded in studies from start to finish, not just for practical thematic analysis.53 Meaningful engagement can build trust, which makes it easier for patient partners to ask questions, request clarification, and share their perspectives. Professional team members should actively encourage patient partners by emphasising that their expertise is critically important and valued. Noting when a patient partner might be best positioned to offer their perspective can be particularly powerful.
Insufficient time allocation
Researchers must allocate enough time to complete thematic analysis. Working with qualitative data takes time, especially because it is often not a linear process. As the strength of thematic analysis lies in its ability to make use of the rich details and complexities of the data, we recommend careful planning for the time required to read and code each document.
Estimating the necessary time can be challenging. For step 1 (reading), researchers can roughly calculate the time required based on the time needed to read and reflect on one piece of data. For step 2 (coding), the total amount of time needed can be extrapolated from the time needed to code one document during codebook development. We also recommend three to five hours for the thematic analysis session itself, although coders will need to independently develop their draft themes beforehand. Although the time required for practical thematic analysis is variable, teams should be able to estimate their own required effort with these guidelines.
Practical thematic analysis builds on the foundational work of Braun and Clarke.416 We have reframed their six phase process into three condensed steps of reading, coding, and theming. While we have maintained important elements of Braun and Clarke’s reflexive thematic analysis, we believe that practical thematic analysis is conceptually simpler and easier to teach to less experienced researchers and non-researcher stakeholders. For teams with different levels of familiarity with qualitative methods, this approach presents a clear roadmap to the reading, coding, and theming of qualitative data. Our practical thematic analysis approach promotes efficient learning by doing—experiential learning.1229 Practical thematic analysis avoids the risk of relying on complex descriptions of methods and theory and places more emphasis on obtaining meaningful insights from those close to real world clinical environments. Although practical thematic analysis can be used to perform intensive theory based analyses, it lends itself more readily to accelerated, pragmatic approaches.
Strengths and limitations
Our approach is designed to smooth the qualitative analysis process and yield high quality themes. Yet, researchers should note that poorly performed analyses will still produce low quality results. Practical thematic analysis is a qualitative analytical approach; it does not look at study design, data collection, or other important elements of qualitative research. It also might not be the right choice for every qualitative research project. We recommend it for applied health services research questions, where diverse perspectives and simplicity might be valuable.
We also urge researchers to improve internal validity through triangulation methods, such as member checking (supplemental material 1).46 Member checking could include soliciting input on high level themes, theme definitions, and quotations from participants. This approach might increase rigor.
We hope that by providing clear and simple instructions for practical thematic analysis, a broader range of researchers will be more inclined to use these methods. Increased transparency and familiarity with qualitative approaches can enhance researchers’ ability to both interpret qualitative studies and offer up new findings themselves. In addition, it can have usefulness in training and reporting. A major strength of this approach is to facilitate meaningful inclusion of patient and care partner perspectives, because their lived experiences can be particularly valuable in data interpretation and the resulting findings.1130 As clinicians are especially pressed for time, they might also appreciate a practical set of instructions that can be immediately used to leverage their insights and access to patients and clinical settings, and increase the impact of qualitative research through timely results.8
Practical thematic analysis is a simplified approach to performing thematic analysis in health services research, a field where the experiences of patients, care partners, and clinicians are of inherent interest. We hope that it will be accessible to those individuals new to qualitative methods, including patients, care partners, clinicians, and other health services researchers. We intend to empower multidisciplinary research teams to explore unanswered questions and make new, important, and rigorous contributions to our understanding of important clinical and health systems research.
All members of the Coproduction Laboratory provided input that shaped this manuscript during laboratory meetings. We acknowledge advice from Elizabeth Carpenter-Song, an expert in qualitative methods.
Coproduction Laboratory group contributors: Stephanie C Acquilano (http://orcid.org/0000-0002-1215-5531), Julie Doherty (http://orcid.org/0000-0002-5279-6536), Rachel C Forcino (http://orcid.org/0000-0001-9938-4830), Tina Foster (http://orcid.org/0000-0001-6239-4031), Megan Holthoff, Christopher R Jacobs (http://orcid.org/0000-0001-5324-8657), Lisa C Johnson (http://orcid.org/0000-0001-7448-4931), Elaine T Kiriakopoulos, Kathryn Kirkland (http://orcid.org/0000-0002-9851-926X), Meredith A MacMartin (http://orcid.org/0000-0002-6614-6091), Emily A Morgan, Eugene Nelson, Elizabeth O’Donnell, Brant Oliver (http://orcid.org/0000-0002-7399-622X), Danielle Schubbe (http://orcid.org/0000-0002-9858-1805), Gabrielle Stevens (http://orcid.org/0000-0001-9001-178X), Rachael P Thomeer (http://orcid.org/0000-0002-5974-3840).
Contributors: Practical thematic analysis, an approach designed for multidisciplinary health services teams new to qualitative research, was based on CHS’s experiences teaching thematic analysis to clinical teams and students. We have drawn heavily from qualitative methods literature. CHS is the guarantor of the article. CHS, AS, CvP, AMK, JRK, and JAP contributed to drafting the manuscript. AS, JG, CMM, JAP, and RWY provided feedback on their experiences using practical thematic analysis. CvP, LCL, SLB, AVC, GE, and JKL advised on qualitative methods in health services research, given extensive experience. All authors meaningfully edited the manuscript content, including AVC and RKS. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding: This manuscript did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interests: All authors have completed the ICMJE uniform disclosure form at https://www.icmje.org/disclosure-of-interest/ and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Provenance and peer review: Not commissioned; externally peer reviewed.