Ecological Momentary Assessment in Palliative Care: A Qualitative Study of Patient and Clinician Perspectives

Abstract

Introduction:

Symptom management tools based on ecological momentary assessment (EMA) may improve symptom tracking and enhance patient-provider communication in ambulatory palliative care (PC), but their design and implementation must align with the needs of patients and clinicians. This study aimed to explore patient and clinician perspectives on the potential role, design considerations, and implementation challenges of implementing EMA-based tools in ambulatory PC settings.

Method:

We conducted an exploratory-descriptive qualitative study using semi-structured interviews with 10 patients and 9 clinicians recruited from the ambulatory palliative care clinic of an academic medical center. Interviews were transcribed and analyzed using a hybrid inductive-deductive thematic analysis approach. Coding was conducted by a multidisciplinary team, and themes were developed through iterative team discussions.

Results:

Four themes emerged: (1) EMA-based tools can enhance symptom tracking and communication; (2) equity and access must be prioritized to avoid digital exclusion; (3) balancing simplicity and personalization is essential to minimize burden and optimize engagement; and (4) integration into clinical workflows is critical for uptake. Participants emphasized the need for tools that are easy to use, customizable, and embedded in existing systems, with clear protocols for triage and support.

Conclusion:

EMA-based tools have the potential to improve symptom management and communication in ambulatory PC, but must be designed with attention to usability, equity, and workflow integration. Our study findings can inform the co-design of EMA-based tools that are responsive to the needs of patients receiving (and clinicians providing) serious illness care.

Introduction

High symptom burden is the most common reason for palliative care (PC) referral and is universally recognized as a referral trigger in clinical guidelines.^1–4 To that end, 90% of initial PC visits address symptoms.⁵ Yet despite the importance of symptom burden as a driver of serious illness outcomes and the role of PC as experts in symptom management, gaps in symptom assessment limit the overall impact of PC interventions on symptoms.^6,7 In ambulatory PC, where follow-up visits typically occur every several weeks to months,^2,4,5,8 patient reports of symptoms are subject to recall bias,^9–12 which can lead to over and under treatment. Furthermore, PC clinicians underassess psychological symptoms, despite their association with poor serious illness-related outcomes.^13–15 Given meta-analytic findings suggesting only modest effects of PC delivery on symptom burden,⁶ improving the quality of symptom data available to patients and their clinicians is critical to develop better symptom management strategies, especially between visits.

Ecological momentary assessment (EMA) is the repeated, ambulatory sampling of individuals’ experiences in real-time and in the “real world” contexts in which they occur.¹⁶ EMA-based tools could help improve symptom assessment in ambulatory PC. First, by gathering data at frequent intervals, EMA-based tools can limit recall bias and the risk of under- or overtreatment of physical and psychological symptoms based on patient reports during encounters.^17,18 Second, EMA-based applications based on digital, remote data collection and automated data analysis can generate individualized reports that highlight severe or worsening symptoms and may facilitate increased measurement, especially for under-assessed psychological symptoms. Similar tools have been used extensively in research and behavioral health care and have established feasibility among cancer patients^19–21—yet their adaptation to ambulatory PC remains limited. For example, the Symptom Tracking and Reporting app (STAR) is a web-based interface with established feasibility in cancer patients with high symptom burden.²² Further, many existing tools used in oncology settings are not designed to provide feedback to patients or clinicians,^23–25 such as the Remote Self-Reporting of Symptoms by Patients with Palliative Care Needs (RELIEF) app,²³ are proprietary, such as the Tell Us app,²⁶ focus solely on monitoring treatment side effects,^27,28 or are subject to recall bias due to infrequent measurements,^24,27–29 such as the Digital Supportive Care Awareness and Navigation (D-SCAN) app.²⁴

EMA-based tools may help improve patient-clinician communication around symptom management, improve data quality by limiting recall bias, address symptoms commonly underassessed, and support clinical decision-making in outpatient PC. However, given the high symptom burden of PC patients and the high workload of PC providers, developing EMA-based tools that are designed for them is key. We aimed to characterize patient and provider perspectives on EMA-based tools in ambulatory PC, with the overarching goal of informing the development of future applications suited to this setting.

Materials and Methods

In this exploratory-descriptive qualitative study,³⁰ we conducted semi-structured interviews with ambulatory PC clinicians and patients regarding EMA-based tools. All study procedures were approved by the Internal Review Board of Weill Cornell Medicine. Study participants provided verbal informed consent for participation.

Recruitment and Participants

Participants were recruited from the ambulatory PC service of Weill Cornell Medicine through purposive sampling to ensure inclusion of patients with a range of serious illnesses and included patients receiving care and PC providers. Participants were compensated with a $50 Visa gift card upon interview completion.

PC providers.

Clinicians (physicians, advanced practice providers, chaplains, or social workers) providing care in the ambulatory PC program were eligible. The study team recruited participants through emails and announcements at clinic-wide meetings.

PC patients.

Patients receiving ambulatory care at Weill Cornell Medicine were eligible to participate if they were: 1) 18 and older; 2) had ≥ 2 ambulatory PC encounters, or one ambulatory PC encounter and inpatient PC services; 3) were able to communicate in English; 4) well enough to participate in a 45-minute semi-structured interview based on a PC provider’s assessment. Patients with a diagnosis of dementia were ineligible. Research assistants (RAs) reviewed patients’ charts to ensure they met the medical eligibility criteria, and called potentially eligible participants to offer participation. Interested patients underwent a verbal informed consent process at the time of the interview.

Our objective in this study was exploratory-descriptive: to elicit stakeholder-informed perspectives that would inform the co-design of an EMA-based symptom assessment application. Accordingly, we planned to conduct 15–20 interviews, balancing dual-stakeholder representation with high information power.³¹ This approach aligns with the goal of the study to deeply understand our local stakeholders’ perspectives as a guide for intervention development, as well as with empirical research demonstrating conceptual saturation often occurs within small samples.³² In this context, our two parallel stakeholder samples (patients, clinicians) each targeted ~10 interviews to ensure coverage across roles and experiences while remaining feasible for a single-site study feeding a rapid, iterative user-centered design cycle. We used purposive, maximum-variation sampling to capture heterogeneity by age, gender, race/ethnicity, primary serious illness (cancer and non-cancer), and (for clinicians) discipline and years in practice. This approach follows established guidance for purposeful sampling in implementation research.³³

Data Collection and Management

One-on-one interviews were conducted virtually over Zoom by a trained RA between December 2024 and February 2025 (average interview 34 minutes, range: 20 – 95 minutes). Audio was recorded via Zoom and transcribed verbatim for analysis, reviewed for accuracy, and de-identified before analysis. De-identified transcripts were stored on institutionally-managed, access-controlled servers with separate linkage files. We did not conduct transcript-return member checking, given participant burden and clinical acuity, but we did present aggregate results to clinicians and to patients involved in intervention co-creation, which will be presented in a future manuscript.

Interview Guide

We utilized a semi-structured interview guide (see Supplement 1) with two subsections: a) recommendations on implementing EMA-based tools in PC and b) barriers and facilitators to integration in PC settings.

Analysis

We used a hybrid inductive-deductive thematic analysis approach.^34,35 Two trained RAs independently coded the transcripts using Atlas.ti (version 9.18.0).³⁶ An initial codebook was developed based on the interview guide and refined iteratively as new codes emerged. We assessed adequacy iteratively during analysis—reviewing codebook stability and the emergence of novel concepts within each stakeholder group to confirm that additional interviews were unlikely to yield materially new themes relevant to user-centered design. Themes were then identified through a framework analysis process,^37,38 incorporating both a priori domains and emergent insights.

Techniques to enhance trustworthiness:

We enhanced credibility and dependability through analyst triangulation: two trained coders independently coded all transcripts, resolving discrepancies by consensus with input from the broader team; themes were refined using a framework-oriented approach. We strengthened reflexivity via a brief positionality statement and regular team debriefs throughout analysis. We maintained an audit trail (versioned codebooks, decision logs, analytic memos) and explicitly sought disconfirming cases when developing and revising themes. We adhered to the Standards for Reporting Qualitative Research (SRQR) and provide the completed checklist in Supplement 2.³⁹

Reflexivity Statement:

Interviews were conducted by trained RAs under the supervision of two principal investigators (PIs): a PC physician with qualitative methods training and a clinical health psychologist with EMA training. While one PI was a member of the PC clinical team, they had no prior therapeutic relationships with study participants. The interview guide was informed by input from a programming team experienced in EMA-based tool development. To enhance reflexivity and reduce bias, the team engaged in collaborative coding, iterative discussions, and critical reflection on assumptions throughout analysis.

Results

We interviewed ten patients and nine PC clinicians (see Table 1 for demographic characteristics). Four overarching themes emerged from the data, reflecting cross-cutting perspectives on the potential role of EMA-based tools in PC, key design considerations, and perceived barriers to adoption. Additional representative quotes are presented in Table 2.

Table 1:

Demographic Characteristics of Study Participants (N = 19)

Characteristic	% (N)
Clinicians (N=9)
Age
≤ 30	11 (1)
31–40	44 (4)
41–50	33 (3)
≥ 51	11 (1)
Female	44 (4)
Race/Ethnicity1
Black or African American	11 (1)
Hispanic/Latino	11 (1)
White	33 (3)
Other	44 (4)
Role
Physician	67 (6)
Social Worker	22 (2)
Chaplain	11 (1)
Years of experience, M (SD)	9.4 (12.1)
Patients (N=10)
Age
31–40	10 (1)
41–50	40 (4)
≥ 51	50 (5)
Female	50 (5)
Race/Ethnicity1
Black or African American	10 (1)
Asian	10 (1)
White	40 (4)
Multiracial	30 (3)
Marital status
Single	50 (5)
Married	40 (4)
Divorced	10 (1)
Primary Serious illness2
Cancer	50 (5)
Heart disease	20 (2)
End stage renal disease	10 (1)
Cirrhosis	20 (2)
Interstitial lung disease	10 (1)
Number of PC visits, M (SD)	6.2 (4.6)

¹One patient did not identify racial or ethnic background.

²One patient identified two primary serious illnesses (heart disease and end stage renal disease).

Table 2:

Theme 1 Illustrative Quotes

Theme 1: Remote symptom assessment can augment current approaches to symptom tracking and assessment, enhancing both patient self-efficacy and patient-clinician communication	Subtheme 1.1: Technology-facilitated EMA can improve tracking of symptom trajectories over time	C03: […] I see a lot of benefits. […] I think one is, I guess we really don’t know how patients are doing unless they call us or there's an adverse event that leads to a hospitalization and then we see that in the EMR. […] this would let us know how our patients are doing in between visits […]. I think it could also honestly be helpful with helping to figure out follow-up appointments when there are wait times for palliative care and a big demand with limited supply. P13: I would put down what I'm feeling […]and whatever pain that I'm having. And they would be able to tell me if I'm doing right […]. Because when I used to send my list of things that my symptoms and what I did and what medicines I took and so forth, they always told me that I was always good enough to write it where they understand what I was doing, and they told me I had done right about sending it that way.
	Subtheme 1.2: Technology-facilitated EMA can improve recognition of underreported symptoms	C04: What happens a lot is that they focus a lot more on their physical symptoms more than their psychological symptoms. That’s very tangible for them…but what you're also seeing is like deep depression or you're seeing anxiety…[…] People are just much more comfortable saying what physically is happening with them. C03: And so it can help if patients are doing really well, maybe they don’t need follow-up as quickly, but if patients aren’t doing well, based on that reported outcomes, then the follow-up can be created or can be pushed up earlier.
	Subtheme 1.3: Technology-facilitated EMA can improve patient-clinician communication regarding symptoms	P09: I think it would help personalize the visit to focus more on me and how I'm feeling versus talking about generic symptoms related to my multiple diseases. P10: I think it would help me when I'm in one of my feelings. And that’s how I'm feeling, and we can both see it. Which is kind of intriguing.
	Subtheme 1.4: Technology-facilitated EMA provides an opportunity for early intervention on emerging or worsening symptoms	P09: I do think it would help me to keep track of all the small things that come up. Because I focus on the severe pain, and this could let me focus on the daily flare ups. C07: […] It might be a good way for us to like to have an intervention earlier. […] It would be nice that if the system flags someone rating a really high scale consistently, it would […] shoot us a message […] that, hey, you should probably call this person.
Theme 2: Equity and Access are Key in Technology-Facilitated Remote Symptom Assessment	Subtheme 2.1: Barriers such as language, cognitive, and physical limitations must be considered to advance equity	C07: [EMA would be] a digital diary for their symptoms, which requires work, and which requires people to be comfortable with technology, which many of our patients do not have that ability to do… And those are the same people that don’t call us. Either because they're not comfortable with technology or there are language barriers for them. P05: I don’t like to text…my hands they will try and pull when I start doing anything […] because my hands, they cramp…I don’t even wanna hold the phone in my hand.
	Subtheme 2.2: Individuals with structural vulnerabilities may be inadvertently excluded from technology-facilitated palliative care	C02: Are we going to get the same type of patients who are health literate, technologically savvy, have access to smartphones and computers? What about the patients who don’t? Who would benefit from this, but because of poverty or lack of resources, just don’t have the access to technology.
	Subtheme 2.3: Inclusion of caregivers as proxy-reports may improve access	C08: Patients may be joined by family members or caregivers […] who are able to also provide a history or a trend of what the day-to-day activities have been and what the patients report. Sometimes many caregivers are actually in charge of the medication management.
Theme 3: Balancing minimizing burden and optimizing engagement is an inherent tension in technology-facilitated symptom assessment	Subtheme 3.1: Both clinicians and patients desire personalization, but not at the expense of simplicity	P11: The app should have symptoms not only related to your condition but an overall list of symptoms. Something that would let me see what else is out there so I can see if I experience that. Because this disease is so rare, and nobody knows what to do with it. P01: Something that’s really easy to you, that has basic generic questions that you could just click through and not have to spend a lot of time filling out.
	Subtheme 3.2: Clinicians and patients worry about technology fatigue and overload	P09: It depends on how complicated it all is. If I have to spend time just learning how to use it then, it would be too much to do. C02: […] when we do telehealth visits with people, we have a questionnaire that they have to fill out. And part of that is also a mood assessment. People balk at it. We have so many patients who see so many different medical providers and are being asked to fill out so many questionnaires. […]we would have to be able to demonstrate how it would make their care better.
	Subtheme 3.3: Notifications and alerts can be useful for patients and clinicians, but only if used sparingly and strategically	P08: You could text …when somebody texts [to] give me my appointment they text it and I open it and I see… C07: I don’t know if I would want to get a notification every time someone rated their pain like seven or eight out of ten… The trend and consistency of the rating I think is more important. So if somebody is at that level for a couple of times …I would want to know about it. But for a single measurement I wouldn’t necessarily need to know about it.
Theme 4: Integration with clinical workflows is critical for uptake	Subtheme 4.1: Clinicians worry that technology-facilitated RSA will create unsustainable increases to workload	C07: Figuring out how to do this in a way that minimizes the amount of time is going to be super important because if it ends up taking too much time, I'm not going to do it if … for people who have a lot of problems, there's no way … I'll be able to add on an extra 15 or 20 minutes to do this. If my follow-up visit is 30 minutes and I've got back-to-back visits.
	Subtheme 4.2: Both clinicians and patients believe that integration into existing EHR systems will improve uptake	P13: I would do it on the computer because I’m so used to sending messages to the doctor on the computer [on the EMR app]. Because […] you can send it to whatever doctor you have to send it to, so I’m used to doing that.. C07: I would probably find that fairly annoying. having to look at multiple different things. It would be nice if there was a way within our EMR system for our patients to, let’s say when they log on to their [EMR app] either on a browser or their phone… If there was a symptom assessment thing where they could like log how their symptoms are doing so that let’s say when I had a visit with them I can open that up and see it.
	Subtheme 4.3: Training for clinicians and patients is a workforce need that clinicians cannot fill	C01: Additional staff … would be really helpful to help onboard the patient to using an EMA platform. The way endocrinology has support from diabetic educators when it comes to like continuous glucose monitors, I think similar workflows…would be really helpful… additional clinical support people such as nurses or like LPNs or even volunteers.

Theme 1: EMA-based tools can augment current approaches to symptom tracking and assessment, enhancing both patient self-efficacy and patient-clinician communication

Participants highlighted the potential for EMA-based tools to enhance symptom tracking and management in ambulatory PC. Patients and clinicians described how EMA-based tools could improve understanding of symptom trajectories, identify underreported symptoms, and enhance clinical conversations.

Subtheme 1.1: Technology-facilitated EMA can improve tracking of symptom trajectories over time

Patients and clinicians emphasized the value of capturing fluctuations in symptoms that might otherwise be missed between visits. “A lot of the symptoms we think about in PC are dynamic and sometimes people have really bad days or sometimes people have better days and I think having objective data to record they’re like averaging symptom experience over the time since the last visit would be really helpful” (Clinician (C) 01). Patients similarly perceived benefits to using this approach: “I do think it would help me to keep track of all of the small things that come up” (Patient (P)09), suggesting that more continuous tracking could support their own awareness and reporting. Clinicians reported that they already often encourage patients to log symptoms using analog methods, but that a technology-facilitated process may be preferable: “I often tell [my patients], particularly when I want to know the trend, and things aren’t… going well to have a little notebook or keep a little diary where they log how much they’re able to do physically or how much medicine they’re taking for how much pain they’re in. Not everybody’s consistent with that. But if it’s something like an app that they can just open up on their phone… then I imagine the barrier to them using it might not be particularly high.” (C07)

Subtheme 1.2: Technology-facilitated EMA can improve recognition of underreported symptoms

Clinicians noted that certain symptoms—particularly mood, anxiety, and appetite—were often underreported in clinic visits. Structured daily questions as done with EMA could normalize reporting and encourage earlier intervention. One clinician remarked, “I think mood and anxiety tend to be underreported. I think sometimes patients…feel like that’s not necessarily something to bring up.” (C03).

Subtheme 1.3: Technology-facilitated EMA can improve patient-clinician communication regarding symptoms

Participants saw EMA as a tool to help patients and clinicians align their priorities. “I think it would help personalize the visit to focus more on me and how I’m feeling versus talking about generic symptoms related to my multiple diseases” (P09). Clinicians described how objective data from EMA could help guide conversations and decisions, particularly around adjusting medications or treatment plans: “It’s important that we match the symptom with the amount of pain medications we’re giving… [if] a patient is saying that their pain is extremely uncontrolled and we want to make sure that it’s…persistent” (C03).

Subtheme 1.4: Technology-facilitated EMA provides an opportunity for early intervention on emerging or worsening symptoms

Both patients and clinicians saw promise in EMA-based tools for early symptom intervention. Clinicians noted that patients sometimes delay contacting the team, leading to missed opportunities for proactive management. “It might be a good way for us to implement an intervention earlier,” one clinician explained, particularly for patients reluctant or unsure when to reach out (C07).

Theme 2: Equity and access are key in digital EMA-based tools

Participants emphasized that considerations of equity and access must inform the development of digital EMA-based tools in PC. Participants identified barriers—ranging from language, literacy, cognitive and physical limitations, to broader structural inequities—that could lead to the exclusion of the patients who might benefit most.

Subtheme 2.1: Barriers such as language, cognitive, and physical limitations must be considered to advance equity

Clinicians stressed the need to address technological barriers. “Some people are just not good with technology. Some people might not be able to understand the steps that you’re asking them to do, or it might not be available in a language that […] they read or speak. And those are the people that I tend to worry about the most” (C07). Patients likewise pointed to potential physical limitations impacting usability, such as difficulty using smartphones: “If they have a really small button, I’d have a hard time using those. It’s hard for me to click on things” (P11).

Subtheme 2.2: Individuals with structural vulnerabilities may be inadvertently excluded from using EMA-based tools

Several participants highlighted socioeconomic barriers that could limit access to EMA-based tools. “We see people from all different socioeconomic backgrounds. Not everyone has access to technology” (C02). Another clinician reflected, “You have a subset of patients who would love it, but those are the patients who usually show up for their appointments […] the issue is really those patients who really need it, but don’t have […] the means” (C04).

Subtheme 2.3: Inclusion of caregivers as proxy-reports may improve access

To address some barriers, participants suggested designing EMA-based tools to accommodate caregiver input. “It would be great […] if they have a caregiver that can do it for them” (C08). Sometimes many caregivers are actually in charge of the medication management […] so they’ll be able to provide that additional input” (C08).

Theme 3: Balancing minimizing burden and optimizing engagement is an inherent tension with EMA-based tools

Participants noted that successful use of EMA-based tools in PC would require balancing flexibility and personalization with simplicity and ease of use. Both patients and clinicians expressed concerns about technology fatigue and emphasized that prompts and alerts should be used judiciously.

Subtheme 3.1: Both clinicians and patients desire personalization, but not at the expense of simplicity

Participants emphasized the need for EMA-based tools to be simple and intuitive, particularly for older adults. “It has to be simple and easy to get into. It’s not easy to get into a lot of these apps” (P03). Another patient suggested, “Something that’s really easy to use… basic generic questions that you could just click through and not have to spend a lot of time filling out” (P01). Clinicians stressed that overly complex tools could reduce patient engagement. Conversely, both clinicians and patients felt that personalization in symptoms tracked, and frequency of tracking, would be helpful. For instance, one patient wished “to be able to add new symptoms so I can talk about them” (P09), while another suggested “Maybe a small place to put your notes down…Why some days might have been worse” (P01).

Subtheme 3.2: Clinicians and patients worry about technology fatigue and overload

Both groups expressed concerns that EMA-based tools might contribute to technology burden. “It’s[…] just more to do on the phone,” one patient remarked (P12). Another reflected, “It’s as long as things don’t become… more cumbersome than they need to be… People are already stressed out” (P04). Clinicians echoed these concerns: “I imagine I would probably find that fairly annoying… having to like look at multiple different things” (C07).

Subtheme 3.3: Notifications and alerts can be useful for patients and clinicians, but only if used sparingly and strategically

Participants valued timely reminders but cautioned that excessive notifications could be overwhelming: “If I’m always getting flooded by it, it wouldn’t help” (P05). Clinicians similarly suggested prompts should be balanced and designed to support engagement without contributing to alert fatigue: “They have to get a prompt on their cell phones to say check-in… there’s just too many things happening in a lot of these patients’ lives for that to be something they remember” (C04).

Theme 4: Integration with clinical workflows is critical for uptake

Clinicians consistently emphasized that successful adoption of EMA-based tools would depend on their fit within clinical workflows. Participants raised concerns about workload implications, the need for seamless integration with the electronic health record (EHR), and the importance of providing training and support to both patients and clinicians.

Subtheme 4.1: Clinicians worry that digital EMA-based tools will create unsustainable increases to workload

Clinicians described concerns that EMA-based tools could add to already high clinical workloads. “A worry that I have is that it would… significantly escalate the amount of work I have to do… in a way that’s not useful for patient care…No one’s going to use it if… it means that I’m going to stay at work an extra two hours” (C07).

Subtheme 4.2: Both clinicians and patients believe that integration into existing EHR systems will improve uptake

Participants emphasized that EMA-based tools should be embedded in the EHR to facilitate use and avoid fragmented data. “It would be helpful if it’s already integrated… into the EHR. It’s less helpful if it’s another link that you have to click on to access that data.” (C08). Another clinician noted that a “tab in their chart” would be preferable to “a completely separate dashboard” (C04). Similarly, patients valued integration into their existing patient portals. [what would make a symptom tracking system easy] I think is it’s connected with the portal. To make it easier” (P05). Additionally, patients highlighted how this would fit into their existing workflows of communicating with providers. “I would do it on the computer because I’m so used to sending messages to the doctor on the computer, on [on the patient portal] (P13).

Subtheme 4.3: Training for clinicians and patients is a workforce need that clinicians cannot fill

Clinicians noted that patients might require support to use EMA-based tools effectively, which clinicians could not provide. “It would be really helpful to have… additional clinical support people such as nurses or LPNs or even volunteers” (C01). Another emphasized, “Who’s going to be troubleshooting the software? It’s not going to be the physician… It’s going to be the IT person… I guess it would be helpful to have… also a user guide for patients” (C08). Clinicians themselves reported a need for training “From the provider side, we would probably need like some sort of orientation on how to like use this platform” (C03), though also suggested this would be less critical than patient training.

Discussion

In this qualitative study, ambulatory PC patients and clinicians perceived the potential of EMA-based tools to enhance symptom tracking, improve patient-clinician communication, and facilitate early symptom intervention. However, they also emphasized the need to balance personalization and simplicity, ensure seamless integration into clinical workflows, and address structural access barriers.

Most EMA research emerged from behavioral health settings, where it is increasingly seen as a meaningful clinical tool.^40–42 Interest in adapting EMA for serious illness care—particularly oncology—has grown, with evidence suggesting potential benefits for symptom control.⁴³ This has been driven partly by the growing recognition that use of Patient Reported Outcome Measures (PROMs) in cancer patients improve outcomes.⁴⁴ Yet despite PC’s focus on symptom management, EMA-based tools have seen limited uptake in PC settings. This may be in part because existing platforms are proprietary, disease-specific, or administered too infrequently to address the dynamic nature of symptom fluctuation in serious illnesses.

Participants in our study envisioned EMA-based tools as a therapeutic adjunct to reinforce trust, enhance visit efficiency, and support shared decision-making. Importantly, clinicians and patients alike emphasized the potential of EMA-based tools to operationalize the routine assessment of underrecognized mental health symptoms that are often minimized or overlooked.^45,46 Their insights align with national efforts to improve mental health care delivery in serious illness and suggest EMA could fill an important gap in outpatient PC practice.^47,48

Nonetheless, digital exclusion and clinical burden concerns were reported. Individuals with cognitive, linguistic, social, or financial barriers to technology use are at risk of “digital exclusion” as health technology is integrated into standard of care.⁴⁹ PC patients, who are often older, have cognitive sequelae, and are at risk of financial toxicity, are vulnerable to exclusion. Clinicians further cautioned that unless integrated into existing systems, EMA-based tools risk increasing workload and alert fatigue.

These findings offer clear directions for future EMA-based tool development in PC. Tools should prioritize usability, customization, and equitable access, including options for caregiver proxies. Integration into clinical workflows (e.g., EHR compatibility, team-based triage) will be essential for uptake. EMA-based tools must not only generate meaningful data but also support patient-clinician relationships, facilitate timely interventions, and minimize burden. Thus, future research should focus on co-design and real-world testing of EMA-based tools.

This study has several limitations. First, it was conducted at a single, urban academic ambulatory palliative care clinic. Because Aim 1 was to identify how EMA-based symptom assessment could fit local workflows and to inform a prototype intervention, transferability, rather than generalizability, is the lens we used. Several elements are likely portable across ambulatory palliative care clinics (e.g., symptom review cadence, triage/escalation logic, and roles responsible for responding to alerts), whereas others may be site-specific (e.g., staffing patterns, EHR/patient-portal adoption, and oncology-embedded vs stand-alone service structures). Our English-speaking, interview-capable sample may under-represent patients with limited English proficiency, low digital literacy, unreliable device or internet access, and those with the highest symptom burden. Additionally, caregivers play a crucial role in symptom monitoring and should be included proactively in subsequent phases. Future work will extend recruitment across multiple sites with diverse populations and care models, intentionally oversample groups at risk for digital exclusion, and evaluate scalability and implementation outcomes.

Conclusions

EMA-based tools can potentially improve communication, symptom management, and quality of life in PC, but only if developed and implemented with attention to usability, equity, and integration with existing workflows. Future directions should focus on co-design of EMA-based tools that center PC patient and clinician perspectives, particularly those at risk for exclusion. Pilot testing in real-world clinical workflows will be essential to assess feasibility, clinical impact, and equity.

Data availability:

Data for the study cannot be shared as consent was not sought from participants to deposit their data in a public repository.