We conducted a patient randomized feasibility study, using a posttest design to examine the effects of PATTERN. This paper largely presents data from the intervention arm pertaining to the feasibility and acceptability of PATTERN.

We first sought approval, via email, from physicians affiliated with any primary care clinic (including geriatrics) within the participating academic health center in Chicago, IL, to contact their patients for recruitment into PATTERN. Patients were eligible to participate if they were aged 65 years or older, spoke English as their primary language, self-reported contending with at least two chronic health conditions, were prescribed eight or more medications they were primarily responsible for taking, and had a visit scheduled within the next 1 week with the physician who had provided contact approval. For intervention purposes, individuals also needed to have access to the internet, an active email address, and a patient portal account. Exclusion criteria included having participated in our prior, formative research or having any severe, uncorrectable visual, hearing or cognitive impairments that would have precluded study consent or participation.

Potentially eligible patient participants were identified using an EHR query, phoned by a trained research coordinator, informed about the study, and screened for eligibility. Those who met all eligibility criteria and expressed an interest in participating then provided electronic informed consent prior to enrollment. The research coordinator scheduled remote posttest interviews with enrolled participants; these occurred approximately 2 weeks after their scheduled clinic visit.

Participants were randomized one-to-one to usual care or the PATTERN intervention using random block sizes of two and four, stratified by clinic type (geriatrics or other primary care clinic). The allocation table was created using SAS PROC PLAN and implemented using REDCap software [12]. Randomization took place immediately following enrollment (Multimedia Appendix 1). The purpose was to be able to compare differences between those randomized to PATTERN and those in usual care.

Participants who were randomized to the PATTERN intervention were sent an email 2 to 3 days ahead of their clinic visit. The email instructed them to log into their patient portal accounts to complete a brief medication adherence assessment prior to their upcoming primary care visit. Those who logged in were instructed to begin the two-tiered assessment adapted from a prior study [10]. The first three questions identified if the participant was experiencing any challenges taking their medications as prescribed. Specifically, these asked: (1) During the past month, did you decide to stop taking any of your medicines without telling your doctor? (2) During the past month, did you decide to take more or less of any of your medicines than you were supposed to without telling your doctor? (3) Are you having any problems with your medicines? Those who endorsed at least one adherence challenge were asked to respond to a subsequent set of 18 questions. These questions, informed by the model of medication self-management, were designed to “phenotype” or identify the distinct type of medication adherence challenge a participant might have been experiencing. Challenges could be classified as cognitive (eg, forgetfulness), psychological (eg, health literacy, depression, or motivation), medical (eg, acute illness), regimen-related (eg, side effects or complex dosing schedules), social (eg, transportation or support), and economic (eg, cost) [5].

Any identified and “phenotyped” challenges were designed to be automatically and securely forwarded, via the EHR, to the participant’s primary care physician with whom they had an upcoming clinic visit. The intent was to ensure physicians received relevant and actionable data from patients regarding any medication challenges they might have been experiencing. This information could subsequently inform the scheduled visit or referral to available support services such as social workers or clinical pharmacists, depending on clinic-specific resources.

Data were collected between March and August 2024. Remote, structured study interviews were held with individual participants approximately 2 weeks after each participant’s regularly scheduled clinic visit. The research coordinator phoned participants at previously agreed-upon times and administered a brief questionnaire. Data were captured using the REDCap software [12]. On average, it took participants 20 minutes to complete the questionnaire, and all participants were compensated US $50 for their time.

Descriptive statistics were used to examine participant characteristics, overall and by study arm. All feasibility and acceptability data were also assessed using descriptive statistics in the intervention arm only. Analyses were conducted using SAS 9.4 software.

All study procedures were approved by the Northwestern University’s Institutional Review Board (STU00217555). All participants provided electronic informed consent prior to study participation and were compensated US $50 for their time. Data presented herein have all been deidentified.

Participant characteristics are presented in Table 1. In brief, a total of 64 patient participants were enrolled. The average (SD) age was 75 (6.8) years. Most indicated they were female (66%, 42/64), non-Hispanic (94%, 60/64), White (58%, 37/64), and in good or better health (64%, 41/64). A total of 16% (10/64) were classified as having mild or moderate depression, and 38% (24/64) as having low patient activation. A large majority of participants (80%, 51/54) reported they used the patient portal more than 12 times a year, and that when they access the portal, they view laboratory results (98%, 63/64), confirm appointments (89%, 57/64), and send messages to their care teams (91%, 58/64). Actual use, assessed via EHR data, revealed that most patients (72%, 46/64) had previously (within the past year) completed any type of assessment in the patient portal. No significant differences in participant characteristics were observed between study arms.

Our assessment of intervention feasibility revealed that 47% (15/32) of participants in the intervention arm completed the adherence assessments that were sent in their patient portal accounts. Of those, only 1 indicated they had any adherence challenges, according to EHR data, though this participant self-reported they did not remember having any challenges. Similarly, when asked by the research coordinator, 75% (24/32) of intervention arm participants self-reported they were unaware they had received an email or portal message asking them to complete the PATTERN medication adherence assessment, and therefore did not complete it. When probed for an explanation, most indicated they had accessed their email but did not notice the request (71%, 17/24). Of those who self-reported receiving the request (25%, 8/32), all but 1 found the adherence assessment to be somewhat or very easy to access, and most (63%, 5/8) reported they completed it. Of the 5 who said they completed the survey, 3 (60%) were very satisfied with the experience, 1 (20%) was a little satisfied, and 1 (20%) did not reply. Moreover, although none of these participants reported having any challenges with their medications, 2 revealed they had discussed the medication assessment and medication taking strategies with their clinical care teams and were satisfied with that conversation. Participants who accessed but did not complete the assessment (37%, 3/8), revealed it was not because they encountered challenges accessing or responding to it; rather, they indicated they did not want to complete it at that time.