Abstract
Among older adults, depression is a common, morbid, and costly disorder. Older adults with depression are overwhelmingly treated by primary care providers with poor rates of remission and treatment response, despite attempts to improve care delivery through behavioral health integration and care management models. Given one in 10 older adults in primary care settings meet criteria for depression, there is a pressing need to improve the efficacy of depression treatment among affected individuals. Measurement-based care (i.e., the incorporation of systematic measurement of patient outcomes into treatment) for depressed older adults in primary care has had poor uptake, which at least partly underlies the limited efficacy of depression treatments. In this perspective, we discuss the proposal that ecological momentary assessment (EMA) may increase uptake of measurement-based care for depression in primary care, enhance the quality of clinical depression data, and lead to improvements in treatment efficacy without adding to providers’ burden. We describe key issues related to EMA implementation and application in routine settings for depressed older adults, along with potential pitfalls and future research directions.
Introduction
Depression among older adults is common (Park & Unützer, 2011), and it confers an increased risk for morbidity and mortality, especially among those with chronic diseases (Chandrasekar et al., 2019; Fiske et al., 2009). Depressed older adults are commonly treated in primary care (PC) by non-mental health providers, due to a confluence of factors such as mental health workforce shortages and a dearth of mental health providers accepting Medicare (Blazer et al., 2012; Committee on the Mental Health Workforce for Geriatric Populations, Board on Health Care Services, & Institute of Medicine, 2012; Moye et al., 2019; Zhu et al., 2023). While efforts to manage depression in PC have advanced, less than half of all depressed older adults receive effective treatment (Callahan, 2001), including lower than recommended antidepressant doses (Wang et al., 2005). Similarly, treatment outcomes are limited. Anti-depressant medications have modest effects (Kok & Reynolds, 2017), remission is achieved in only one-third of patients (Kok et al., 2012), and 30% of older adults with depression will develop treatment-resistant depression (Subramanian et al., 2023). With 10% of older adults in PC meeting criteria for major depression (Park & Unützer, 2011), there is an imperative need to identify PC-based strategies to improve depression management and increase treatment efficacy.
A related difficulty in depression management in PC is the infrequent assessment of depression symptom trajectories through measurement-based care (MBC), that is, the incorporation of systematic measurement of patient outcomes into treatment. MBC constitutes a standard of care for other chronic conditions such as diabetes or hypertension where serial assessments (e.g., A1C level) are undertaken to determine disease trajectory and adjust treatment accordingly. Paradoxically, MBC for depression remains rare despite an expanding body of evidence from randomized controlled trials showing improved patient outcomes compared to usual care—particularly for those at risk for treatment failure (Fortney et al., 2017). Despite evidence of improved outcomes, MBC for depression remains infrequent even in specialized mental health settings, due to implementation barriers such as low perceived utility, time and resource constraints, and lack of knowledge and self-efficacy in implementing MBC (Keepers et al., 2023). Instead, depression care relies on patient reports during follow-up visits, which are subject to recall bias and produce poor quality data (Horwitz et al., 2023). To improve depression management and outcomes in PC, novel approaches to increase MBC uptake are needed.
We propose that ecological momentary assessment (EMA)—that is, the repeated, ambulatory sampling of individuals’ experiences in real-time, and in the “real world” contexts in which these experiences occur—could help improve depression outcomes for older adults in PC in several ways: (1) Digital, remote EMA data collection and automated analyses of EMA data can generate individualized symptom reports without burdening providers—creating actionable data; (2) thanks to the increasing uptake of mobile technology (Faverio, 2022) and high EMA adherence among older adults (Yao et al., 2023), remote measurement of depression symptoms between PC visits could increase MBC for depression; and (3) by relying on frequent and naturalistic measurements, EMA is less affected by the recall bias of retrospective measures completed during visits. In the following sections, we discuss EMA and its potential to improve depression management for older adults in PC, current applications of EMA in clinical care, barriers to implementation, and outline future directions.
How Could EMA Improve MBC for Depression in Primary Care?
MBC for depression (usually through time-consuming serial PHQ-9s) remains infrequent even in mental health settings, due to time and resource constraints (Keepers et al., 2023). Instead, depression care typically relies on patients’ verbal reports during follow-up visits, which are subject to recall bias (Henry et al., 1994; Horwitz et al., 2023; Urban et al., 2018) and produce poor quality data (Horwitz et al., 2023). For example, in the case of the PHQ-9, there is a tendency for the most intense and proximal aspects of an experience to disproportionately influence memory for self-reported depressive symptoms (Horwitz et al., 2023). Such inaccuracies might trigger treatment decisions, such as titrating medications up or down. We might imagine a patient whose depression symptoms had been generally improving in the period between visits, who comes in for a follow-up visit on a day when the patient evidenced significantly depressed mood after a night of insomnia due to arguing with their spouse. Due to recall bias, this patient might report worse depression symptoms over the past month than if the argument and insomnia happened a month earlier. Recall bias may be even more prominent in older adults with cognitive impairment, and when follow-up visits are scheduled over lengthy periods (e.g., quarterly visits). Thus, patients’ reports during PC visits may not accurately reflect depression trajectories—leading to risk of both over- and undertreatment.
In contrast, in an EMA-based approach, patient-reported symptoms can be assessed remotely via surveys in between follow up visits. Patients may be asked to provide data on mood, anhedonia, or suicidal ideation in real time via their mobile devices. Such approaches stand in contrast to the global depression measurements obtained during routine clinic visits that rely on recall over lengthy periods and are detached from real-world contexts. We propose that EMA could overcome the limitations of global, retrospective depression reports and improve MBC for depressed older adults in PC. Further, EMA based on digital, remote data collection and automated analyses of patient profiles can generate individualized symptom reports without adding to providers’ workloads—creating actionable data to aid clinical decision making. With the increasing uptake of mobile technology among older adults and meta-analyses showing high EMA compliance rates in this population (Yao et al., 2023), incorporating depression symptom data measured remotely in between visits may increase uptake of MBC for depression in PC. By relying on frequent and naturalistic measurements, it can limit recall bias common in global, retrospective depression measures completed during follow-up visits. Finally, patients’ ability to access their own data may lead to insight into their symptom patterns and triggers as well as increase patient-physician communication around depression treatment. For example, in a qualitative study of adults with diverse psychiatric diagnoses, participants reported that EMA-derived feedback could encourage other patients to continue or begin treatments, enhance awareness around triggers for worsened symptoms, and improve self-management by offering real-time information on symptom response following implementation of new coping strategies or interventions (Bos et al., 2019). Further, access to individual-level symptom patterns could lead to personalized treatment, such as choosing interventions based on personal symptom patterns and trajectories. Despite its potential, to our knowledge no studies have evaluated EMA approaches to depression management for older adults in PC.
Current Applications of EMA in Clinical Care
Multiple calls to action have been made to integrate EMA into routine care in diverse settings, such as to monitor symptoms following serious injury (Mitchell et al., 2022), to deliver just-in-time adaptive interventions for weight management (Spruijt-Metz et al., 2015), and specifically in behavioral health settings to support treatment decisions and tailor clinical approaches (Rodebaugh et al., 2020). Studies in mental health settings are increasingly attempting to design EMA applications to improve care (Bos et al., 2022), highlighting the promise of EMA approaches. For example, in a feasibility study of veterans with post-traumatic stress disorder receiving EMA-derived feedback as an adjunct to usual care, completion of scheduled EMA surveys reached nearly 90% after incorporating patient feedback on acceptability (Smith et al., 2012). Several studies assessing EMA effectiveness as an additional component to treatment or compared to usual care for depression are underway for adolescents and adults under 65 (for a comprehensive review, see Colombo et al., 2019).
Several EMA-based approaches for depression recently developed are of note for their potential to become integrated into routine care, including Therap-I (Riese et al., 2021), PETRA (Bos et al., 2022), MindLogger (Klein et al., 2021), ZELF-i (Bastiaansen et al., 2018), and ESMvis (Bringmann et al., 2021). While there is significant heterogeneity in these technologies, they commonly assess depressive symptoms and other related variables (e.g., positive or negative affect), as well as contextual variables such as daily life events that might be associated with mood (e.g., interpersonal stress). The timing of surveys, frequency, duration, and number of items vary significantly among these tools (e.g., from three daily surveys to ten, and from 3 days to 2 months). EMA-derived feedback is a common addition to these technologies, with variability in the frequency (e.g., weekly, end-of-trial) and mode of feedback delivery (e.g., by a research assistant, via email), and the visualization (e.g., trajectory of symptoms, word clouds). For example, the PETRA application (Bos et al., 2022) is a platform integrated into the electronic health records, and co-designed with clinicians and patients with a wide range of mental health diagnoses. The goal of this tool is to collect personalized EMA data and provide EMA-based feedback to patients and clinicians. Clinicians can select specific symptoms relevant to the patient (e.g., auditory hallucinations for patients with psychotic disorders) and a visual platform provides user-friendly feedback on symptom trajectories (see Bos et al., 2022 for an example of the feedback). Other EMA-based approaches have also been developed with a greater emphasis on the providers, such as Mental Health Telemetry (Schaffer et al., 2013)—a mobile application that monitors patients newly diagnosed with major depression as they are started on a course of psychotropic medication to evaluate treatment response. Importantly, early efficacy results of EMA-based approaches are promising, as shown in a randomized controlled trial of depressed adults on outpatient medication management—with greater symptom reduction in the EMA-derived feedback arm compared to usual care (a 5.5-point score difference on the Hamilton Depression Rating Scale; Kramer et al., 2014). However, despite the increasing number of EMA-based approaches to improve depression treatment, the feasibility and acceptability of these tools in depressed older adults remains understudied. Further, studies have focused on mental health settings—where providers have greater frequency and duration of visits focused solely on mental health concerns—and not in PC, where providers are burdened by short visit times and competing priorities that limit their ability to assess and manage depression in this population (Tai-Seale et al., 2007). Thus, tools that give PC providers actionable data on depression without relying on lengthy questionnaires completed during visits may be of greater use compared to mental health settings.
Implementing EMA in Routine PC with Depressed Older Adults: Challenges and Considerations
Implementation of EMA approaches into routine PC is not without potential pitfalls. Given the growth of mobile health technologies, EMA-based approaches are well poised to improve depression management in PC for older adults. Nevertheless, challenges to implementation of such technologies remain. While other authors have highlighted the potential benefits and challenges of integrating EMA-based approaches into usual care (Greenberg et al., 2023; Kim et al., 2019; Spruijt-Metz et al., 2015), we highlight below some additional issues specific to older adults in PC settings. These include issues of usability, utility/acceptability, measurement burden, adherence to EMA protocols, cognitive impairment, access to technology, and data security which warrant special consideration.
Patient Challenges
While there is great variability in EMA-based approaches in terms of frequency, timing, and duration of measurement bursts, frequent assessments could lead to a high burden of self-reporting data or lack of protocol adherence. However, encouraging evidence from a recent meta-analysis shows high adherence to EMA protocols among older adults when employing best practices (Yao et al., 2023), especially when EMA training is provided at study initiation, and when protocol adjustments are made by incorporating patient feedback. It is also likely that providing EMA-derived reports to patients that offer insight and added value can help with protocol adherence, however this needs further examination. Research is needed into patients’ perceived utility of EMA-based approaches to depression management, and user-centered design will be key to meeting the needs of depressed older adults.
A related concern in older adults is adherence to EMA protocols in those with cognitive impairment. A recent study of older adults with and without mild cognitive impairment achieved excellent adherence to EMA protocols around 95%, with modest differences between patients based on cognitive function in a fully remote protocol (Dowell-Esquivel et al., 2024). However, further research is needed into EMA protocol adherence among patients with greater degrees of cognitive impairment, as well as the feasibility of caregiver-informed EMA strategies for those with severe impairment. Similar concerns have been raised regarding adherence to EMA protocols among patients with depression, and how anhedonia may prevent patients from engaging with burdensome protocols for a sustained period. However, research shows this concern does not necessarily translate to differential adherence rates to EMA in depressed vs non depressed older adults. Even in early EMA research—which was paper-based—compliance rates commonly remained above 80% (Cain et al., 2009), with more recent smartphone-based EMA studies in depressed populations still showing high adherence rates that are similar in those with and without depression (van Genugten et al., 2020).
Concerns have also been raised regarding access to technology and technological literacy among older adults. Smartphone adoption is high in adults ages 65 and above, with 76% using smartphones and 70% having home broadband (Pew Research Center, 2024). Nevertheless, equity concerns remain in terms of the distribution of technological access and literacy. Co-design of EMA-based platforms with end-users, including patients and providers, will be key to addressing these challenges, consistent with models of technology use and adoption centering perceived usefulness and usability as drivers of older adults’ intention to use new technology (Boot et al., 2020). Further, proposed solutions including text-only options for EMA measurement bursts or automated calls can facilitate access.
Provider Challenges
Common concerns about EMA implementation have also been raised regarding providers, especially regarding workload and utility. Evidence (albeit limited) from behavioral health settings show that mental health providers see EMA-derived feedback as less useful than patients do, with some providers reporting that EMA-based feedback does not significantly add to their routine assessments (Frumkin et al., 2021). However—in contrast with behavioral health settings in which providers are trained in the comprehensive assessment and treatment of depression, and visits focus on mental health, PC providers may have limited training in depression management, and deliver care in shorter visits in which many competing health priorities are routinely discussed. Therefore, it is likely that EMA-derived feedback (especially when co-designed with PC providers) can generate data of greater utility for this group of providers compared to providers working in behavioral health settings, in a way that decreases provider burden and fits with the workflow of PC settings. In sum, early consideration of dissemination and implementation issues will be essential to produce actionable data of high utility to PC providers, and maximize uptake in primary care.
Security
Finally, a concern when designing any mobile health application to be used by patients and providers remains the storage and protection of confidential health information to meet the requirements of the Health Insurance Portability and Accountability Act (HIPAA) and other privacy rules. While the long-term goal should include streamlining within electronic health records systems (EHRs), protecting health information in standalone applications in the short- and medium-term as the efficacy and effectiveness of EMA-based approaches are established (likely before EHR integration) will be critical.
There is enormous potential to leverage EMA-based approaches to improve depression management in PC for older adults, yet research into implementation efforts is still emerging and further work is needed. In addition to recommendations for research agendas previously suggested for health care professionals and researchers for the use of EMA-based approaches (e.g., Bartels et al., 2023; Kim et al., 2019), we outline a research agenda in Table 1 based on the existing literature, suggesting a roadmap specifically tailored to the use of EMA-based approaches for depressed older adults in PC settings.
EMA to Improve MBC for Depression in PC for Older Adults—A Research Agenda.
Conclusion
EMA constitutes a highly promising approach to improving MBC for depression in older adults, specifically one with the potential to enhance care already being provided in PC settings. Despite its potential, no studies have evaluated if EMA approaches for depression management among older adults in PC are: (1) feasible and acceptable to patients and providers, and (2) can lead to improved mental health outcomes. Ironically, although EMA might seem technically complex, if pipelines can be developed and maintained that deliver data directly to PC providers (e.g., by integrating the data directly into the electronic health record), it might prove simpler to implement EMA into PC than questionnaire paper forms, in that little would need to happen within the PC setting itself. Challenges remain regarding implementation barriers that need consideration and exploration. As a growing number of older adults have access to technology allowing for remote data collection of symptom patterns, understanding how we can harness EMA to improve depression outcomes in older adults receiving primary care is critical.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by a T32 grant from the National Institute on Aging [grant number T32 AG049666].
