Abstract
Background: Care pathways are tools that can help family physicians navigate the complexities of the cancer diagnostic process. Our objective was to examine the mental models associated with using care pathways for cancer diagnosis of a group of family physicians in Alberta.
Methods: We conducted a qualitative study using cognitive task analysis, with interviews in the primary care setting between February and March 2021. Family physicians whose practices were not heavily oriented toward patients with cancer and who did not work closely with specialized cancer clinics were recruited with the support of the Alberta Medical Association and leveraging our familiarity with Alberta’s Primary Care Networks. We conducted simulation exercise interviews with 3 pathway examples over Zoom, and we analyzed data using both macrocognition theory and thematic analysis.
Results: Eight family physicians participated. Macrocognitive functions (and subthemes) related to mental models were sense-making and learning (confirmation and validation, guidance and support, and sense-giving to patients), care coordination and diagnostic decision-making (shared understanding). Themes related to the use of the pathways were limited use in diagnosis decisions, use in guiding and supporting referral, only relevant and easy-to-process information, and easily accessible.
Interpretation: Our findings suggest the importance of designing pathways intentionally for streamlined integration into family physicians’ practices, highlighting the need for co-design approaches. Pathways were identified as a tool that, used in combination with other tools, may help gather information and support cancer diagnosis decisions, with the goals of improving patient outcomes and care experience.
In Canada, the process of obtaining a cancer diagnosis after first suspicion of a problem can be fraught with delay,1–5 which has been associated with shorter survival, decreased quality of life posttreatment and sub-optimal patient experience.2,6 Delays may be caused by various factors related to the characteristics of cancer, the patient and the fragmented health care system.5,7,8 There is a set of standard tests and steps required to get to diagnosis, and the coordination of these mostly rests in the hands of family physicians.1,5,9
Care pathways are evidence-based tools outlining a care plan that can help family physicians navigate the diagnostic process.10,11 They are increasingly common12 and mostly promoted by governments and health care authorities, given their potential to improve care coordination11 and patient outcomes, and reduce health care spending.13,14 In the primary care context, pathways are often developed by health care authorities, at times in collaboration with family physicians or specialists.15–17 Whereas some studies in Alberta and Canada suggest that family physicians are interested in and follow pathways in their practice,15 others suggest important challenges related to how pathways might be adopted and used.18 Some authors report low uptake due to pathway characteristics (format, content), availability and accessibility,17,19,20 and family physicians’ practice context.17,20–23 However, there is little discussion about how family physicians think and make decisions when presented with pathways, and how pathways fit into their working strategies. In this study, we intended to contribute to that discussion in the context of cancer diagnosis. The objective was to examine the mental models associated with using pathways for cancer diagnosis of a group of family physicians in Alberta, to inform future approaches to pathway design with the goal of promoting higher uptake and integration into practice. Mental models are an explanation of an individual’s understanding about how something works in the real world.24 They incorporate cognitive functions or mental processes known as “macrocognition” (Table 1) and are explained by macrocognition theory.24–27
Methods
We conducted a qualitative study using cognitive task analysis, an established method used to understand how experienced knowledge workers think and approach their work in real-world settings.24,28 It uses specialized interviews to uncover tacit knowledge, the “subtle, cognitive aspects of task performance,”29 by including specially trained interviewers and keeping the interviewee focused on a specific and recent case, probing and reprobing, for vivid recall of knowledge and skills based on lived experience. In our study, eliciting this knowledge was done using simulation exercises, where participants reviewed hands-on practice and followed a think-aloud process to talk through a particular scenario.24 Think-aloud processes are often used to test health care tools (e.g., pathways), since they help gain insight into participants’ thoughts when using these tools and reveal information needs and usability in their existing workflows.30
Setting and participants
We used purposive sampling31 to target family physicians in typical community practices, who represent real-world users and are most influential in uptake of pathways.32 We avoided physicians with practices heavily oriented toward patients with cancer or who worked closely with specialized cancer clinics, as they may have advanced knowledge of the cancer diagnostic process and might not require pathways. Recruitment relied on notices in the provincial newsletters of the Alberta Medical Association (AMA) (Appendix 1, available at https://www.cmajopen.ca/content/11/3/E486/suppl/DC1), and sampling relied on our familiarity with Alberta’s Primary Care Networks and confirmation of alignment with our target population by interested physicians. Our approach was to explore in depth each participant’s input and macrocognition used, and consider commonalities. Based on previous work and experience using cognitive task analysis with physicians, 6 to 8 interviews were expected.33,34 Using the expertise of the research team, and following the criteria for success in cognitive task analysis,29 we stopped recruiting when we had enough valuable insights to describe participants’ mental models and to inform future approaches to pathway design. Individual interviews took place virtually by Zoom, in February and March 2021, after informed consent was granted. No repeat interviews were conducted. Participants were offered a stipend, based on established provincial guidelines.35
Data sources and collection
Data sources were cognitive task analysis interviews from simulation exercises that used 3 similar algorithmic examples of cancer diagnosis pathways developed in Alberta: rectal bleeding, iron deficiency anemia and lymphoma (Appendices 2, 3 and 4, available at https://www.cmajopen.ca/content/11/3/E486/suppl/DC1). Each simulation asked the participant to choose 1 pathway based on a recent case where they could have used it. They were asked to recall that case and think aloud about how they would have cared for that patient using the pathway. We used an interview guide derived from macrocognition theory25–27 to ensure that key information was elicited. We gathered information on how the proposed pathways affected any of participants’ macrocognitive functions (Table 1) involved in the cancer diagnosis process (investigation, referral, diagnosis), the fit between participants’ mental models of diagnostic processes and the proposed pathways, and their use of the pathways (Appendix 5, available at https://www.cmajopen.ca/content/11/3/E486/suppl/DC1). Interview guides were developed by the research team, which included a family physician and 2 allied health professionals, and based on practice expertise and previous work in the area.1,18,33,34 In addition, we pilot-tested the interview guide with an external family physician. Interviews were conducted by an interviewer and a note-taker, both members of the AMA Accelerating Change Transformation Team (AMA-ACTT) trained in cognitive task analysis.36 Transcripts were not returned to participants for comment or correction, as they rarely are in cognitive task analysis; however, participants were notified that they may be contacted with clarifying questions. A demographic questionnaire was also completed for each participant during the interview (Appendix 6, available at https://www.cmajopen.ca/content/11/3/E486/suppl/DC1).
Data analysis
All interviews were audio-recorded, transcribed verbatim, combined with field notes and imported into Excel for analysis. In Excel, transcripts were divided into sections. Each section was coded based on our understanding of the macrocognitive functions presented in cognitive task analysis (Table 1) and explored in our previous cognitive task analysis studies.18,33,34 Additional themes were identified using thematic analysis.37 Coding of each section was completed by 2 members of AMA-ACTT trained in cognitive task analysis. Members were assigned to sections so that the same 2 members did not code together each time. To ensure consistency and trust-worthiness,37 AMA-ACTT and research team members met to review and discuss the coding, resolving any disagreements by consensus. We then met to review all coded data, build a description of each participant’s mental model of their cognitive approach to using clinical diagnostic pathways, and compile similarities and contrasts across participants.
Ethics approval
Ethics approval was received from the Health Research Ethics Board of Alberta Cancer Committee (HREBA.CC-21-0003).
Results
Eight family physicians volunteered, and all volunteers were included (Table 2). Interviews lasted 45–60 minutes. Four physicians chose the rectal bleeding pathway, 3 the iron deficiency anemia pathway and 1 the lymphoma pathway. Two elected to also comment on pathways they did not choose (iron deficiency and lymphoma). The analysis identified the macrocognitive functions most used and physicians’ mental models (with 4 subthemes), and the actual use of the pathways (4 themes). Illustrative quotations are provided in Table 3 and Table 4.
Macrocognitive functions and mental models
The pathways heavily influenced participants’ sense-making and learning (Table 1). Participants used the pathways to gather information, confirm what they already knew or support what they were doing, or as a quick guide for what steps to take when unsure. This was particularly true for the rectal bleeding and iron deficiency anemia pathways. These are health issues regularly seen by physicians, and therefore, physicians had well-developed mental models of the diagnostic process and described using these pathways as a quick confirmation or validation tool (Table 3, subtheme 1). For the lymphoma pathway, a less commonly seen health issue, physicians did not have a well-developed mental model of the diagnostic process, and would use the pathway for guidance and support, and confidence or reinforcement in making decisions about the diagnostic and referral processes (Table 3, subtheme 2).
In some cases, participants stated they would use the pathways to help patients make sense of their situation, either to show them that a process exists and where they are within that process, or to provide information in patient handouts (Table 3, subtheme 3). Some participants noted that it would be useful to have an easy-to-find, printable patient version of the pathway, with information on procedures and adverse effects (Table 3, subtheme 3.1).
Participants suggested that the 3 pathways could help care coordination and decision-making. Participants perceived a lack of agreement among family physicians and specialists (and among specialists themselves) about what is considered a “high-risk” scenario, which determines whether referrals are “semiurgent” or “urgent.” Pathways were intended to inform or validate decision-making regarding referral priority; however, with uncertainty about definitions of cancer risk, and without clarity about urgency of testing and referral, most participants stated they would simply “pick up the phone and call a specialist” to avoid a potential miss of cancer diagnosis (Table 3, subtheme 4).
Use of the pathways
Participants noted that the pathways presented in this study are not tools they would necessarily use for all patients. They explained that for common health issues, they had already developed an approach and relied on existing tools to inform care decisions, such as generally recommended clinical guidelines (e.g., Towards Optimized Practice guidelines [https://actt.albertadoctors.org/CPGs], addressing particular care gaps), or more detailed and condition-specific clinical support tools (e.g., UpToDate [https://www.wolterskluwer.com/en/solutions/uptodate]). Participants reported that if they were to use the pathways presented in the study, they would use them only to complement tools they typically use (Table 4, theme 1).
Participants indicated that the most valuable use of the pathways was to support and guide the referral process. First, the pathways could act as a platform to build a shared understanding between family and specialist physicians of what “high risk” means and when to send a referral for “semiurgent” or “urgent” treatment. Second, the pathways could provide steps to guide the referral process, including criteria, requirements, and when and to whom to refer. Some participants proposed the pathways could be part of the referral process itself by including a clickable referral form that could be submitted. Participants indicated this may help improve time to diagnosis, and communication and care coordination with specialists (Table 4, theme 2).
Although the pathways were presented to participants as algorithms, participants demonstrated that they would not use them algorithmically in clinical practice. In the time-pressured primary care setting, participants emphasized their need to access and process information quickly. They walked us through how they would rapidly review the pathways to identify recognizable patterns or the minimum information necessary to make decisions, confirm knowledge, guide what steps to take when unsure, or build new patterns to drive a satisfactory decision. The participants emphasized they typically would only access the first page, which should present the most valuable information in a concise and user-friendly way (Table 4, theme 3). Lastly, most participants stated that the pathways needed to be located on the same webpage and easy to find. A few participants noted that having access to the pathways through their electronic medical records would be ideal (Table 4, theme 4).
Interpretation
In this cognitive task analysis, the pathways presented to participating family physicians had little effect on their diagnostic process for the 3 conditions (rectal bleeding, iron deficiency anemia and lymphoma) beyond sense-making and learning, some diagnostic decision-making, and potentially care coordination. The pathways did not conflict with participants’ mental models, meaning that they could fit within their diagnostic and referral processes. Findings indicated that participants would use the pathways to gather information, confirm or validate what they already know, locate steps to take when unsure, and complement tools they already use.
Our findings align well with and help explain previous studies reporting low uptake of pathways in primary care, even though family physicians may see them as highly relevant.38 Evidence on successful pathway implementation and uptake is not abundant,21 and mostly refers to barriers and facilitators related to the contextual factors linked to patients (e.g., epidemiologic, socioeconomic and ethical aspects), health care providers (knowledge, attitudes and behaviour) and work environment (e.g., inadequate staffing and time pressure).23 Our results elaborate on factors related to family physicians and suggest that uptake will be less successful than anticipated if pathways conflict with physicians’ cognitive work strategies. The pathways in the study were presented as algorithms, but participants used pattern recognition and satisficing approaches39 to quickly find the minimum information necessary to make a satisfactory decision, or to confirm that their decision was appropriate. It is crucial that future design of primary care pathways takes into account the cognitive implications of the primary care time-pressured context, in which family physicians, in order to quickly work through the issue at hand, are heavily dependent on fast, automatic and instinctive thinking strategies.40 Future pathway design considerations might include simple checklists or providing key information upfront and very succinctly.
In Alberta, pathways have been identified by the Strategic Clinical Networks (networks of clinicians and patients with knowledge about a specific health area) at Alberta Health Services as a preferred strategy to improve the quality of care provided to Albertans.16,41 Findings from our study provide 2 relevant considerations related to this. First, they suggest that pathways should be co-designed with the intended physician end-users, and include explicit examination of their cognitive work patterns to fulfill their information needs, thereby increasing the likelihood of incorporation into their workflow. Postimplementation solutions to promote pathway uptake and use, such as dissemination strategies,42 training activities43–45 or additional materials like explanatory handbooks,46 may not be enough to support successful implementation. Second, findings suggest that pathways may need to be complemented by other tools. Pathways in cancer care are identified as key to support enhanced cancer diagnosis,47 yet participants reported that they would refer their patients to a specialist when there was any — even minimal — chance of cancer, and mentioned the lack of shared understanding and communication with specialists. In alignment with current discussions on suboptimal care continuity and inconsistent collaboration between family physicians or specialists,5,9,48,49 our findings emphasize the importance of pathways, while also suggesting the need for additional improved supports for family physicians. A previous study by this team1 reported that family physicians and cancer specialists supported the implementation of a centralized service where primary and specialist physicians converge in their roles. Pathways, low-risk guides and other initiatives, such as rapid access clinics50,51 and specialty teleconsultation systems,15 could be considered.
Limitations
Interviewers and participants might have had previous interaction through other work, and this may have led to social desirability and response bias by participants. Experts often provide learned theoretical responses to questions, or may consider their reputations when answering, which can lead to response bias. The cognitive task analysis method of using specialized interviews to ground experts in an actual case in a real-world setting was deliberately designed to address these biases. Not all groups of potential participants were represented in the study. For example, just 2 physicians practising in northern Alberta participated, and as such, the findings may not reflect the experiences of physicians in northern rural and remote communities of the province, who, given the characteristics of these populations and lower access to specialist care, may use pathways differently.
Conclusion
We found that family physicians might use pathways developed for primary care, but not necessarily in the manner intended. Although pathways are designed to algorithmically guide the care process, our findings suggest that pathways might be used to quickly gather information, confirm decisions or provide information to patients. Our findings highlight the need to use co-design approaches to develop pathways, ensuring information needs, cognitive strategies and workflows of family physicians are accounted for. Our findings also underline the need to think about cancer diagnosis pathways not as “the tool” to guarantee improved cancer diagnosis, but as “one tool” that may be used in combination with others to help enhance cancer diagnosis. Future studies should explore and rigorously assess existing and innovative approaches to develop pathways and additional supports that can be easily integrated into family physicians’ practices.
Acknowledgements
The authors acknowledge the Alberta Medical Association for its assistance with recruiting participants, and the Alberta Medical Association Accelerating Change Transformation Team for its help with data collection and analysis. The authors also acknowledge study participants for their time and valuable insights.
Footnotes
Competing interests: Anna Pujadas Botey and Paula Robson report in-kind support from the Cancer Strategic Clinical Network, Alberta Health Services, through employment with Alberta Health Services. Tanya Barber reports indirect funding of the research project via salary support (employment with University of Alberta). No other competing interests were declared.
This article has been peer reviewed.
Contributors: Anna Pujadas Botey, Tanya Barber, Barbara O’Neill and Lee Green conceptualized the study. Tanya Barber and Lee Green designed the study with supports from Anna Pujadas Botey, Paula Robson and Barbara O’Neill. Tanya Barber managed the project, did the ethics submission, recruited participants and analyzed data. Anna Pujadas Botey prepared the first draft of the manuscript with important input from Tanya Barber. Paula Robson, Barbara O’Neill and Lee Green made substantial contributions to the interpretation of data and revised the manuscript critically for important intellectual content. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work.
Funding: The authors received no specific funding for this work.
Data sharing: All relevant data are within the manuscript. Further data excerpts are available on request from the corresponding author.
Supplemental information: For reviewer comments and the original submission of this manuscript, please see www.cmajopen.ca/content/11/3/E486/suppl/DC1.
This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/by-nc-nd/4.0/
References
- © 2023 CMA Impact Inc. or its licensors