The central question of this project concerns the effectiveness of different media platforms in delivering an eHealth intervention for older adults with multiple chronic conditions (MCCs), including pain. Thus far, eHealth interventions for chronic conditions have generally been delivered through computer websites or smartphone apps. It is unknown whether such interventions could be even more effective for older adults if delivered on emerging smart speakers and display platforms.

Smart speakers, such as Amazon Echo, are voice-controlled devices connected to the internet. Users talk to the device and listen to the results rather than typing and reading. Smart displays, such as Google Nest Hub, offer the same voice interactivity but add a screen to allow for viewing of videos, visual navigation of menu options, and touch selection of options. Both smart speakers and smart displays are increasingly designed to move beyond simple voice commands and responses to offer more conversational interactions via chatbots.

Various authors have speculated about the potential of these devices to transform self-management and treatment of chronic diseases such as diabetes [1-3]. For older adults with ongoing pain, limited mobility, limited vision, or hand tremors, interacting via voice may enhance accessibility. In theory, users could ask questions and have answers immediately spoken back to them. They could hear and respond to messages, request audio services such as guided meditation, record health data, and more just by talking or simple taps on the screen. At the same time, the chatbot interactivity could potentially enhance feelings of relationship and support offered by the device and by the specific eHealth intervention on the device. When asked to predict the near future of voice-activated devices in health care, a Delphi panel of 35 physicians, academics, and information technology specialists predicted that such devices would show notable gains in acceptance and use, particularly among older patients, because of their simplicity of use [4].

Indeed, media reports indicate increasing use of speakers among older adults and people with vision loss, tremors, lack of mobility, and dementia [5-9]. Thus far, a handful of studies, primarily conducted with samples of ≤20, suggest that older adults given smart devices generally report liking them, using them mostly for music and simple information requests such as the weather [10-15]. In a study of 12 older adults given a smart display for 16 weeks, the participants reported finding the device easier to use than a computer [12]. However, they also reported forgetting how to wake up the device, control the volume, or turn it off when it was activated unexpectedly, and device log data showed that one-third of attempted interactions were unsuccessful because the device did not understand the request. However, over time, users showed an increasing ability to rephrase unsuccessful requests, and transcriptions of interactions suggested companionable exchanges between the user and the device.

Thus far, only a couple of pilot studies have focused on health-related uses of smart speakers. In one, 58 patients with chronic conditions were given an Amazon Alexa device for 2 months, and a number of them reported using it to remind them to take their medications and to record when they had done so [16]. In a very similar study, 44 adults aged 50 to 90 years with chronic conditions were given Amazon’s Echo Show for 2 months [17]. They too reported using the device for medication and appointment reminders, with additional anecdotes of using it to find recipes tailored to dietary requirements, medication information, exercise videos, guided meditation, or soothing sounds to help with falling asleep. An unspecified subset of users felt that their health and quality of life had improved over the 2 months as a result of the device. Neither study involved a comparison group, reported on actual log data rather than self-reports of use, or systematically assessed health outcomes.

Given these promising small-scale findings, rigorous research seems warranted to examine whether smart displays can facilitate the self-management of chronic conditions. Chronic conditions are prevalent among older adults. Approximately two-thirds of Medicare beneficiaries have ≥3 ongoing conditions such as diabetes or arthritis, and nearly one-fourth have ≥5 [18,19]. The complexity of patients’ comorbidities makes them challenging to serve well in primary care, where time pressures and patient loads necessitate a focus on medication and laboratory results rather than well-being and skills for self-management, treatment adherence, and health tracking [20]. Patients with MCCs are also the most expensive, accounting for 90% of Medicare spending [18]. Most important for the individual patient, having multiple conditions is associated with reductions in physical and psychosocial quality of life [21-23] and increased risk of chronic pain [24].

Chronic pain, generally defined as pain lasting ≥3 months, is reciprocally intertwined with a number of variables [25] such as anxiety and depression [26,27], sleep [28], and daily functioning [29]. Chronic pain also intersects with issues of social justice; a recent review noted that those with lower socioeconomic status are more likely to report chronic pain, more severe pain, and more pain-related disability [24]. Various meta-analyses have indicated ongoing racial disparities, such that African Americans and other patients of color tend to be undertreated for pain [30,31]. Increased self-management of chronic pain is a national objective established by the US Department of Health and Human Services in Healthy People 2030 for the next decade [32], and there is evidence that internet-based interventions can improve pain-related outcomes [33-35]. However, no research thus far has examined self-management interventions for chronic pain within the context of MCCs. Furthermore, there is no research on the relative effectiveness of different platforms for delivering such an intervention for this complex patient population.

Large-scale, long-term trials of the effectiveness of delivery platforms may help advance the objective outlined in Healthy People 2030. This paper reports on the study design and methods of a randomized controlled trial of the effectiveness of an evidence-based eHealth intervention for older adults with MCCs, including pain, when delivered on a smart display versus a laptop platform.

The system, ElderTree, was developed by our Agency for Healthcare Research and Quality Center of Excellence in Active Aging to improve quality of life and socioemotional outcomes among older adults and was first tested in a randomized controlled trial involving 390 older adults who were followed for 12 months. In that intention-to-treat trial, patients in the ElderTree group who had had ≥3 primary care visits in the 6 months before baseline showed significantly better results on measures of mental quality of life, social support, and depression compared with control patients who had received a laptop but no ElderTree access [36].

Given that primary care use is relatively high among patients with MCCs, the results suggest that ElderTree may most benefit such patients and may be most effective if integrated into primary care. A subsequent randomized controlled trial, funded by the National Heart, Lung and Blood Institute, is currently testing ElderTree among patients with MCCs, with an added report to primary care clinicians documenting changes in health status. That ongoing trial is testing whether patients assigned to a laptop-based ElderTree versus an attention control will have better quality of life, fewer symptoms, and better condition-specific laboratory scores [37].

The trial described in this protocol is designed to test whether the delivery platform of the intervention affects its effectiveness. Specifically, the study aims to assess whether participants using ElderTree on a smart display will experience decreased pain interference and improved quality of life and related measures in comparison with participants using ElderTree on a laptop and control participants given no device or access to ElderTree.

The trial has a randomized controlled design with 3 groups with 1:1:1 allocation. Participants will be randomized to receive (1) a smart display plus internet access plus ElderTree along with their usual care, (2) a touch screen laptop plus internet access plus ElderTree along with their usual care, or (3) usual care alone. All participants will be followed for 8 months.

A total of 291 older adults will be recruited from the University of Wisconsin–Madison Department of Family Medicine and General Internal Medicine system (UW Health), Access Community Health Centers, and community organizations in the Madison and Beloit, Wisconsin, areas.

Eligible participants will (1) be aged ≥60 years and (2) have chronic pain as indicated by having received a chronic pain diagnosis or reporting pain on some or most days that has lasted for ≥12 weeks (duration); reporting pain in the last 3 months on some, most, or all days (frequency); and reporting a pain intensity of ≥3 on a scale of 0 (no pain) to 10 (worst imaginable pain) in the last 7 days. They will also (3) have at least 3 of the most prevalent chronic conditions among older adults as reported by the Centers for Medicare and Medicaid Services [18]: chronic obstructive pulmonary disease, asthma, diabetes, hyperlipidemia, hypertension, ischemic heart disease, atrial fibrillation, heart failure, stroke, cancer, chronic kidney disease, depression, osteoporosis, and arthritis; we have modified the list by adding obesity (BMI≥30) and dizziness, falls, and loss of vestibular function. In addition, patients (4) will allow reports to their primary care provider about their health tracking and (5) should have no plans to leave during the study period. Patients are not eligible if they require an interpreter or if they have a medical diagnosis of Alzheimer disease, dementia, schizophrenia or other psychotic disorder, autism spectrum disorder, known terminal illness with <6 months to live, or an acute medical problem requiring immediate hospitalization.

For UW Health recruitment, the university’s Clinical and Health Informatics Institute will use clinic records to identify patients meeting the aforementioned eligibility criteria and send their name, address, birthdate, age, UW Health clinic location, and primary care physician to the university’s Office of Clinical Trials via REDCap (Research Electronic Data Capture; Vanderbilt University). Potential participants will receive an opt-in letter describing the study and consent form from the Office of Clinical Trials plus a stamped return letter inviting contact from the study team.

We will supplement recruitment at UW Health with community efforts to increase the racial diversity of our patient population. The selection of and engagement with community organizations will be led by a senior advisor on our team with extensive experience in local health campaigns and deep roots in the local African American health care and patient communities. Through collaborations with the African American Health Network, Rebalanced-Life Wellness Association, African American Opioid Coalition, and the Community-Academic Aging Research Network, we will reach out to Black churches, community centers, fraternity and sorority health service programs, and other organizations as selected by our advisor in the Madison and Beloit areas of Wisconsin. We will distribute a recruitment flyer and consent form with a return card or opt-in and conduct video chat and in-person community sessions when possible to introduce the study and invite participation.

Through Access Community Health Centers, we expect to reach a larger number of African Americans and other participants of color as well as expand our reach to underserved patients. We will work directly with Access Centers staff to disseminate the recruitment flyer, consent form, and return card or opt-in to potential patients.

Regardless of the recruitment method, when a return card or opt-in is received, study staff will call and assess eligibility; provide a study overview that includes potential benefits and risks, study procedures, and compensation; thoroughly walk through informed consent; and address questions. The baseline survey will be mailed. Patients will be given adequate time to decide whether to participate and, in some cases, a follow-up phone call will be scheduled after the patient has had time to review the survey and consent form.

Once a patient verbally confirms that they want to participate, a home visit will be scheduled to collect the baseline survey, randomize, set up the equipment, and train. All study staff who conduct home visits will be fully vaccinated against COVID-19, wear face masks, and socially distance as much as possible while in the patient’s home. If no visit is desired, the equipment will be shipped once the study staff receives the completed baseline survey via mail, and technology setup and training will be conducted by phone. The baseline survey should take 20 to 30 minutes. The measures will be the same in all 3 arms to avoid differential dropout. We will document those who choose not to participate and why following the CONSORT (Consolidated Standards of Reporting Trials) guidelines.

Once informed consent and the baseline survey have been obtained from patients, the project manager will use a computer-generated allocation sequence to randomize on a 1:1:1 ratio to the ET-SD, ET-LT, or control group stratified by sex (male or female), site (UW Health, Access Centers, or the community), and number of chronic conditions (≤5 or ≥6). A research staff member will then conduct equipment setup and training for patients based on group assignment, which is provided in a sealed opaque envelope. Obviously, once an assignment is made and a device has been set up, participants cannot be blinded to their condition. To set up participants on their assigned system (ET-SD or ET-LT), the researcher conducting the training cannot be blinded to the condition after assignment. Figure 1 shows the flow of participants during the trial.

Textbox 1 shows the timeline by year of the study, with year 1 beginning in August 2020 and year 5 ending in July 2025.

Our primary outcome used to power the study is reduction in pain interference. On the basis of PROMIS validation studies with chronic pain samples, a difference of ≥3 points between study arms was considered to indicate a clinically meaningful difference [80]. The PROMIS website also provides the mean and SD of the T-score metric (mean 50, SD 10). We powered the analysis to be able to detect a 3-point difference (an effect size of d=0.30) between the control and ET-LT groups, and then the same magnitude of effect between the ET-LT and ET-SD groups (an effect size of d=0.60 for control vs ET-SD). Across 10,000 linear mixed model simulations, a postattrition N=255, we would have power of >80% to detect the study arm × time interaction. In a previous ElderTree trial, 353 (90.5%) of 390 participants completed the 6-month survey, and 310 (79.5%) of 390 completed the 12-month survey [36]. In the completed surveys, data were missing for approximately 2% of the core items. We expect similar rates in this study, or approximately 87% survey completion at 8 months. Thus, we increased the total sample size after attrition to 291 to increase the likelihood of detecting effects on the outcomes with 2 moderators: sex and number of chronic conditions.

We will promote retention by providing ready access to support for patients’ use of the technologies and by actively following up with patients to encourage them to return surveys. If a survey is not returned within 2 weeks, a research team member will call to check that the survey was received and encourage the patient to complete and return it in the addressed stamped envelope. The date and time of the phone call will be recorded in REDCap along with whether the researcher talked to the participant directly or left a message and any information gathered during the phone conversation. If we cannot reach the participant, another copy of the survey will be sent with a personal note asking them to complete it or call us on our toll-free number if they have questions or are no longer interested. In our study of older adults with MCCs, for which we have recently completed data collection, survey response rates were 94.7% at 6 months, 93.6% at 12 months, and 92.3% at 18 months [37].

To mitigate the risk of breaches of patient confidentiality, all participants are assigned a unique code number. All contact information and survey data are housed electronically in REDCap. Survey data are double-entered by 2 different individuals to ensure accuracy. Paper-based files are stored in a locked room in locked file cabinets and can be accessed only by authorized personnel. The database administrator provides access to study data at appropriate levels for various members of the research team. Members of the research team are able to view deidentified individual and clinic-level aggregations of the variables.

This study protocol received ethical approval from the University of Wisconsin Health Sciences and Minimal Risk Research Institutional Review Board (reference 2020-0868). All amendments to the protocol have been submitted to the institutional review board and approved. This study complies with the Declaration of Helsinki and its later amendments.

To our knowledge, this is the first study to examine with a large sample and a long time frame whether a smart device can, in reality, perform as well as or better than a laptop in implementing a health intervention. To assess this, we will examine group differences in pain and quality of life measures among patients in high need of resources to help manage their chronic conditions. A 2021 systematic review of research on the use of “voice-based conversational agents for the prevention and management of chronic conditions” concluded that the field is “in its infancy” [83].

Our center is well-positioned to pursue this question as we have an evidence-based laptop system available for adaptation and comparison. The results of our first trial of ElderTree indicated that older patients coping with multiple comorbidities were most likely to benefit from such an intervention [36], and our subsequent study is testing that hypothesis [37]. The study described in this protocol, targeting similar patients, is designed to test and compare how delivery platform affects pain and psychosocial variables that have shown improvement with the laptop version [36]. Comparing platforms with this highly challenged population may highlight more starkly the relative advantages and pitfalls of voice-controlled platforms for eHealth delivery.

Applications for voice-controlled devices are in the early stages of development. Some known limitations, such as problems with dictation and word recognition, are certain to improve. Others, including the lack of ability to review longer lists of results or read larger amounts of text, may be endemic. Still other limitations, as yet unknown and coming to light as developers stumble upon them, may or may not be resolved. Whether the advantages of the technology outweigh the disadvantages and how its unique capabilities may be optimized, if at all, are empirical questions to be explored in this and future studies.

A specific question is whether smart devices for eHealth may be more effective depending on the health context. For example, this protocol describes an intervention targeting chronic pain. We are also in the system development stage of a trial of smart devices funded by the National Heart, Lung and Blood Institute in which the primary outcome is functional health, a variable that includes multiple physical and psychoemotional factors.

The goal of ElderTree, regardless of delivery platform, is to improve patient self-management and provide convenient, ongoing support as a means to improve health and quality of life. At the same time, and as a result, interventions such as ElderTree may help relieve some of the burden on the primary health care system. As older patients with MCCs are such a large and growing cohort, the implications for access and cost as well as patient well-being are significant. Making the best use of current and developing technologies is a critical part of this effort.