Development of a Tethered Personal Health Record Framework for Early End-of-Life Discussions
Seuli Bose-Brill, MD; Matthew Kretovics, MPH; Taylor Ballenger, BS; Gabriella Modan, PhD; Albert Lai, PhD; Lindsay Belanger, MPH; Stephen Koesters, MD; Taylor Pressler-Vydra, MS; and Celia Wills, PhD, RN
Advance care planning (ACP) clarifies personal preferences, resulting in written advance directives (ADs) for future medical decisions in the event of health decision-making incapacity.1,2
It helps patients: a) reflect on goals, values, and beliefs; b) consider future treatment preferences; c) appoint a surrogate decision maker; and d) document their wishes regarding future medical treatment.3
ACP is associated with improved patient satisfaction with care, improved quality of life in terminal illness, and better psychological outcomes of family members after patient death.3-5
ACP is also associated with increased use of hospice, reduced intensive care unit use, and reduced costs for unwanted end-of-life care.5
However, rates of completion remain low (seldom >31%),6-8
even for patients with expected survival of less than 4 months.3
For providers in a time-limited encounter in the primary care setting, ACP delivery may not be considered a priority over competing concerns.9
Additional barriers—such as a lack of training and resources, and prognostic uncertainty—have been reported by primary care providers.10,11
These barriers highlight the need for more accessible and time-efficient methods for recording patient ACP preferences in primary care12
that integrate nonphysician clinic providers into the communication process.13
However, a construct for providing efficient, team-based ACP delivery for outpatient practices is needed to help ensure its prioritization.
While validated tools for ACP facilitation exist, they require dedicated time and personnel for administration that are not feasible for many practices. Researchers have recognized the need for electronic tools that empower patients to engage in ACP,7,14-16
because currently, these stand-alone tools do not link documents to the patient’s electronic health record (EHR), aka personal health record (PHR). In order for providers to access documentation, the patient must provide it to them to file in the medical record. Therefore, electronic support tools are a promising approach to translation of validated ACP tools for use in resource-constrained primary care settings, especially if such tools can automatically interface with the medical record.
This innovative systems-level solution can achieve higher rates of ACP and AD completion. Investigation of such a solution remains imperative because of patient satisfaction, disease understanding, and the economic benefits of well-documented ACP.3-5
This study was approved by the Ohio State University Institutional Review Board prior to commencement.
The framework was developed using a mixed methods approach over 2 phases: phase 1—initial framework development using review of literature and best practices in combination with focus group data to inform content; and phase 2—cognitive interviewing of patients to elicit feedback about the framework.
The original framework was established through review of best practices, and patient-desired content was derived from focus-group data. Common themes from delivery methods were summarized for focus-group participants. An interdisciplinary research team used focus-group feedback to inform development of the PHR-delivered ACP framework. The framework was presented during a focus group to all primary care providers practicing at the study site. Feedback about framework content and distribution was incorporated to create an initial draft of the PHR-delivered ACP framework. Additional primary-care patient participants (n = 22) were recruited to complete cognitive interviews. The research team used feedback from cognitive interviews to further revise the language and layout of the framework.
Focus Groups (phase 1)
Four patient-participant focus groups and 1 physician focus group were conducted to elicit preferences for a PHR-delivered ACP framework. These preferences were used to develop content, language, structure, and workflow for the framework. Four to 5 participants were recruited to each patient focus group based on best practice recommendations.17
Purposive sampling was used to ensure robust African American participant involvement. Such feedback was necessary because African American patients are half as likely as Caucasian patients to participate in ACP using existing models.18
All primary care providers at the clinical study site were recruited to participate in the physician focus group.
Phase 1: Patient Participant Focus Groups
Nineteen participants were recruited for 4 different focus groups. Inclusion criteria were: a) current patient at study site, b) aged over 50 years, c) diagnosis of 1 or more chronic diseases, and d) taking 1 or more prescription medications. Inclusion criterion for physician focus group was: primary care provider at study site. Patient participant focus group demographics are summarized in Table 1
. Of the 4 patient focus groups, 2 had Caucasian participants and 2 had African-American participants. Race-specific patient focus groups were conducted to ensure that opinions of African American participants were fully voiced without influence from Caucasian participants.19
Of the 19 subjects recruited, 6 were “no-shows” for their respective focus groups, leaving 13 total participants. Each focus group had between 2 and 4 participants due to “no-shows.” Of the participants, no subjects with less than a high school diploma consented to participate in the study. Six of the 13 participants reported having completed written ADs; no participants with a completed AD were African American.
Setting and content.
All focus groups were facilitated by the same set of co-facilitators and were given/shown the same materials in order to inform discussion (eAppendix A [eAppendices
are available at www.ajmc.com
]). Educational information about ACP and PHRs were provided to participants at the beginning of the session. Participants were shown a brief video created by National Healthcare Decisions Day about ACP, and were given an informational brochure about MyChart, the institution’s tethered PHR, supported by the Epic EHR system. Educational materials, including a copy of the institution’s AD informational packet, were presented in binders that were distributed to each participant. The institutional AD packet contained an informational sheet about ADs, resources for discussing and developing ADs, and state-specific Health Care Power of Attorney and Living Will forms.
The binder also contained discussion questions and excerpts of ACP language employed in validated delivery systems.2,20,21
Sample questions were selected after review of the CDC’s summary document on ACP resources for the public.22
Discussions were initiated using a semi-structured format, each lasting approximately 60 minutes. One Caucasian participant in the first focus group was noted to be less vocal and participatory than others in the group.
Confidentiality, and the privacy of participants and content, were discussed prior to starting. The study team gave an introduction to ACP as a framework for conversation. Participants were subsequently asked about: a) personal experiences with end-of-life decisions and MyChart use, b) personal preferences on how their ACP plans should be communicated, c) willingness to use MyChart for ACP, and d) perceptions of sample ACP questions used in referenced validated face-to-face interventions (see Appendix A).2,20,21
The same questions were asked during each focus group. Discussion content was used to develop an initial PHR communication framework and workflow for provider feedback.
Phase 1: Physician Focus Group
The physician focus group was recruited from the practicing physicians at the study site. All 4 physicians were approached, and all participated in the focus group. Demographic information about physician participants is summarized below (Table 2
). The physicians at the study site were all younger than 45 years; they had been in practice between 3 and 11 years.
Setting and content.
Physician focus group participants were asked about: a) clinical experiences with ACP, b) clinical experiences using PHR; c) barriers to ACP in practice, d) willingness to engage in ACP using PHR, and e) feedback about draft of framework and proposed workflow. Discussion was initiated using a semi-structured format and lasted approximately 45 minutes.
Phase 1: Data Collection and Analysis Methods (all focus groups)
Data were collected through audio recordings and observational notes taken during the focus groups. Recordings were transcribed using detailed transcription and were transcribed by the same trained transcriptionist. Study researchers decided on the format of detailed transcription prior to initiation because it captures not only verbal content, but also conversational features such as pauses, stuttering, and interruptions. Such factors were taken into consideration during the analysis of the focus groups to better understand the context of different comments. This form of transcription helps capture emotions, such as enthusiasm and discomfort, in addition to content.
Content Analysis Method was selected because it allowed assessment of consensus categories for framework development.23-27
Focused tape review of transcripts and field notes established narrative accuracy of data prior to analysis. Detailed transcripts were coded alongside field notes by 3 analysts. Each coded the transcript separately, and then compared results from the independent coding. To ensure confirmability and credibility of findings, focus group analysis was performed in 5 steps: 1) independent content category development; 2) independent identification of consensus categories; 3) development of a written template defining criteria for categories and subcategories (during a meeting among analysts); 4) initial assessment of inter-rater reliability; and 5) elimination of unreliable categories after discussion among raters.27
Themes found to be common among the coders were summarized and considered for framework development. Additional peer debriefing with 2 members of the research team helped ensure credibility of findings. Content and consensus categories informed initial framework question development (eAppendix A).
Cognitive Interviews (phase 2)
Following development of the framework, primary care patients were approached during clinical visits to participate in cognitive interviews. Cognitive interviews were conducted to receive “real-time” patient participant feedback about the framework, including the content, structure, and manner in which it would be received (ie, over MyChart).28
Patients aged 50 years or older were recruited to participate in 15-minute interviews during clinical sessions over a 6-week period. Demographics of cognitive interview participants are summarized in Table 3
Participants were instructed to “think aloud” in order to provide immediate feedback on the content, structure, and layout of the framework as they read through it for the first time. The research assistants asked probing questions only about comments made by the participants. If a participant specifically requested help in navigating the PHR or the framework, the research assistant: a) provided targeted assistance, and b) documented participant difficulty with the domain for which help was requested. Participant comments were scribed by the research assistant immediately after the interview. Interviews were conducted until data saturation was reached.29
Observations were then compiled and used to revise the initial framework.30
Focus Groups (phase 1)
Patient focus groups were analyzed using the scissor-and-sort method of transcripts and scribe notes.31,32
Only elements that were identified by all 3 analysts were considered in assembling the framework. A summary of these elements was compiled and reviewed during framework development. Patient focus group analysis revealed several common preferences present in each of the respective groups: a preference for clear language in communication tools; endorsement of MyChart as a helpful communication tool; a need to qualify and disqualify preferred decision makers; and a desire to personalize content, often based on previous experiences (Table 4
). These preferences were used to tailor an initial ACP framework for MyChart.
Cognitive Interviewing (phase 2)
Notes from the interviews were reviewed by 2 independent reviewers; a count of major themes was taken (Table 5
) and categorized into “strengths” and “weaknesses.” Summary lists were compared and discrepancies were noted. Repeat review was conducted until analysts reached agreement in summaries.
Two of the most frequently mentioned “weaknesses” of the framework were that it was difficult to access MyChart—meaning they had difficulty logging in or registering for MyChart on their own (10 of 22)—and to locate the message within the account (14 of 22). These comments did not directly reflect the content of the framework, but they were still important considerations and were subsequently communicated to institutional information technology (IT) developers. Thirteen of the 22 participants mentioned that they would re-word questions. Other common themes were discomfort in answering the questions (5 of 22), need for an introduction to the framework (5 of 22), and desire for follow-up and discussion (5 of 22).
The physician focus group was also analyzed using a scissor-and-sort method.32
Physician comments were used to prepare a specific office workflow for the framework distribution and inform a streamlined ACP framework seeking specific ACP elements. Broader concerns about PHR use and billing that were beyond the scope of ACP framework development and workflow were communicated with participant permission to IT and medical center management.
The ACP process holds several advantages for patients, yet documentation of ACP and ADs remain low. The use of an EHR to deliver and support the ACP process could be advantageous to both care providers and patients, offering a more efficient use of time and resources. Although stand-alone tools aid in the process, these tools do not interface with medical records. Our framework allows for ACP documentation to be accessible by the individual and their medical team when it is most needed. This newly developed framework serves as a clinical tool, yet retains benefits of patient-initiated electronic ACP documentation.
Through focus group testing, we determined that patients desired a clear, concise, and accessible communication tool that would allow them to voice wishes and desires during the ACP process. Provider feedback indicated that the framework should help patients reflect and give a starting point for the conversations, but emphasized that they wanted to know only “vital” information. Providers also voiced their desire for the framework to fit within the EHR and the clinical workflow. While cognitive interviewing using the framework confirmed that patients approved of the content and delivery method, it also highlighted the need for small edits to language and workflow.
The focus groups had a high rate of “no-shows.” In addition, the study results would have been more generalizable had the study taken place across multiple clinics, not just a single clinical population. While minority participants were purposively sampled in focus groups, they were underrepresented during cognitive interviews. The use of only 1 physician focus group should also be considered, as resulting data may be incomplete. Although data from physician focus groups were consistent with those of similar studies, inclusion of a larger number of physicians may have yielded more reliable, complete results. That was not possible in this study, due to the small number of physicians at the study site. Future studies involving multiple practices will allow more robust exploration of perspectives about the framework. Ideally, additional focus groups would have been conducted in order to ensure saturation of opinion within the study population.
This project set out to develop a usable, patient-centered ACP framework to improve ACP documentation. Patient impressions reported during cognitive interviews suggested that patients found the framework accessible. The use of questions and content vetted by patients in the target population was an essential component of the development.
Future investigation should focus on larger, more diverse populations in order to improve the generalizability of this study and the framework. Investigators and providers will also need to consider how to make the electronic framework more accessible to patients who face some barrier to navigating or accessing their EHR.
Qualitative evidence would suggest that the developed framework would meet the needs of both patients and providers as a tool for documentation in the PHR, particularly in primary care. Incorporating the framework should be done with careful discussions with clinic providers in order to tailor workflows to individual practices (eAppendix B).
We would like to thank the following people for their support over the course of the study: Barbara Longo, Patricia Strickland, and Ann Henry of Ohio State University (OSU) Internal Medicine and Pediatrics Grandview for their administrative support of the project; Rose Hallarn, OSU Center for Clinical and Translational Science, for her assistance with participant recruitment strategies; Peter Embi, MD, MS, FACP, FACMI, Vice Chair of the Department of Biomedical Informatics at the Ohio State University for his advice on managing an Epic-based interdisciplinary study; and Lori Blum, Grants Administration for the Department of General Internal Medicine at the Ohio State University, for her assistance in budgetary matters. The project described was supported by Award Number Grant 8UL1TR000090-05 from the National Center for Advancing Translational Sciences. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Advancing Translational Sciences or the National Institutes of Health.