This study reports outcomes from the PictouCountyPEP, a single-arm, prospective, 6-month digital behavioral intervention delivered to adults living with chronic health conditions in Pictou County, Nova Scotia, Canada (October 2023 – August 2025). The study was designed as a pragmatic, real-world evaluation of feasibility, engagement, and longitudinal change under community delivery conditions rather than as a controlled efficacy trial; accordingly, observed changes over time cannot be attributed solely to the intervention.

Pictou County is a predominantly rural region with approximately 44,000 residents and limited access to specialty health services [23].

Adults aged 18 years or older residing in Pictou County and living with 1 or more self-reported chronic conditions were eligible to participate. Eligibility was not restricted to a prespecified list of chronic disease diagnoses; the program was intentionally designed for adults living with chronic conditions across diagnostic categories and for those with multimorbidity. Eligibility required the ability to safely engage in low- to moderate-intensity physical activity, assessed using the Physical Activity Readiness Questionnaire for Everyone (PAR-Q+) [24], English proficiency, and access to the internet and email. No specific chronic disease categories were otherwise excluded; however, individuals with a current or recent cancer diagnosis were excluded because the intervention model has been evaluated separately in cancer populations in prior trials [21,22].

Recruitment occurred through community-based outreach, including local media, social media, flyers, and partner organizations [20]. Eligible participants were enrolled consecutively as they presented for participation during the recruitment period.

Participants provided electronic informed consent via the REDCap eConsent framework. No financial compensation for study participation was provided. Data were collected and managed in accordance with Nova Scotia Health privacy policies and the requirements of Canada’s Personal Information Protection and Electronic Documents Act (PIPEDA). No images identifying individual participants are included in the manuscript or supplementary materials. The study was approved by the Nova Scotia Health Research Ethics Board (REB 1029744) prior to participant enrollment and was registered on ClinicalTrials.gov (NCT06064110).

Self-reported surveys were administered using REDCap [25]. Participants completed assessments at baseline, 6 months, and 12 months.

PictouCountyPEP is a home-based, multicomponent digital intervention designed to support self-management and mental well-being among adults living with chronic health conditions. The intervention was adapted from previously evaluated patient empowerment program models and modified to remove all cancer-specific education. The full protocol has been published previously [20].

The program delivered daily automated emails over a 26-week period, linking participants to brief instructional videos hosted by authors GI and RDHR. The videos were approximately 4 to 6 minutes in duration and accessible asynchronously via a public video-hosting platform. The content addressed 6 behavioral domains: aerobic physical activity, strength training or yoga, relaxation and stress management, dietary guidance, habit formation, and relationship and social connection skills. The content was modular, allowing flexible engagement according to individual needs and symptom burden.

Exercise guidance aligned with the national physical activity recommendations, encouraging 150 to 300 minutes per week of moderate-intensity aerobic activity and strength or flexibility training at least twice weekly. The content emphasized pacing, gradual progression, and symptom monitoring, with encouragement to tailor activities in consultation with health care providers. Dietary guidance promoted a predominantly plant-based pattern with reduced intake of ultra-processed foods, consistent with Canada’s Food Guide [26]. The stress management content incorporated mindfulness-based practices, breathing techniques, and cognitive strategies. The relationship modules focused on communication, boundary setting, and social connection, while the habit formation content emphasized small, sustainable behavioral changes.

The intervention was designed for low-bandwidth delivery and mobile accessibility to minimize barriers related to internet speed, device availability, or digital literacy. No specialized applications, wearable devices, or synchronous participation was required. Engagement was encouraged through daily reminders and weekly compliance surveys. Optional components included a peer support option, monthly live group videoconference sessions, and occasional in-person community events. These components were offered to enhance social connection and accountability but were not required or standardized. The intervention was delivered under real-world conditions to maximize external validity and inform future scale-up. Given the single-arm interventional design, this study follows the TREND (Transparent Reporting of Evaluations with Nonrandomized Designs) guidelines for nonrandomized evaluations of behavioral interventions [27]. The daily digital intervention content was delivered during the initial 6 months. During months 7 to 12, participants entered a lower-intensity maintenance period in which monthly group videoconference sessions continued to be offered, but daily intervention emails were no longer delivered. The 12-month assessment therefore reflects outcomes following the initial 6-month daily intervention and the subsequent 6-month lower-intensity maintenance period.

Because the intervention content was delivered asynchronously and participants could engage flexibly with individual components, standardized individual-level measures of total intervention exposure or module completion were not available for the present analysis; consequently, a formal dose-response analysis was not conducted.

Internal consistency of all multi-item scales was assessed at baseline using Cronbach α with 95% CI to characterize measurement reliability in the study population prior to intervention exposure (Table S1 in Multimedia Appendix 1).

Covariates included age, sex, relationship status, comorbidity burden measured by the Charlson Comorbidity Index, and use of medication for depression or anxiety at baseline. These variables were selected a priori based on theoretical relevance and prior evidence from related patient empowerment program trials identifying their associations with mental health outcomes and intervention response [20-22].

This phase 2 interventional study enrolled 182 participants to evaluate feasibility, engagement, and longitudinal changes across a broad set of health outcomes. The sample size was determined pragmatically based on anticipated recruitment capacity within the study region and time frame rather than a formal a priori power calculation, consistent with the study’s phase 2 implementation-focused objectives.

All participants who consented and completed baseline assessments were included in the analytic cohort. Primary longitudinal analyses incorporated all available observed outcome data from participants who contributed data at the relevant assessment time points.

Longitudinal changes in psychosocial, physical, and behavioral outcomes were examined using generalized estimating equations (GEEs), which estimate population-averaged effects while accounting for within-participant correlation across repeated measurements. Continuous outcomes were modeled using an identity link with Gaussian working variance, and binary outcomes defined using established clinical thresholds (Multimedia Appendix 1) were modeled using a logit link with binomial working variance. An exchangeable working correlation structure was specified for all models, reflecting the limited number of equally spaced follow-up assessments, and robust (sandwich) SEs were used for inference to ensure valid estimation under potential misspecification of the working correlation. GEEs were selected to estimate population-averaged effects aligned with the study objective of evaluating overall program-associated change rather than subject-specific trajectories.

Time was modeled as a categorical variable (baseline, 6 months, and 12 months). All models were adjusted a priori for baseline age, sex, relationship status, comorbidity burden (Charlson Comorbidity Index), and use of medication for depression or anxiety. Model-based estimated marginal means and corresponding SEs were derived at each time point, and pairwise contrasts were used to estimate differences between baseline and follow-up assessments, with 95% CI reported. Primary inference for continuous outcomes was based on time indicator coefficients from these models, representing adjusted population-averaged mean differences from baseline.

Exploratory interaction models were estimated for psychological outcomes only, by including baseline characteristics interacted with time, to examine whether baseline factors modified trajectories of psychological change. These analyses were prespecified as exploratory and interpreted as hypothesis-generating.

Exploratory analyses examined associations between within-participant changes in selected candidate factors and changes in psychological outcomes. These analyses were intended to identify patterns of co-occurring change and generate hypotheses regarding potential pathways rather than to test formal statistical mediation or establish causal mechanisms. Residualized change scores from baseline to follow-up were calculated, and linear regression models were fitted to assess associations between changes in physical, psychosocial, and substance use measures and changes in psychological well-being (WHO-5), perceived stress (PSS-10), life satisfaction (SWLS), and disease-related worry (QSC-R10). Models were adjusted for baseline outcome values and prespecified covariates. Results are interpreted as exploratory associations between longitudinal changes rather than as estimates of mediation or causal effects.

Clinically meaningful improvement was further examined using responder analyses based on established thresholds. Two responder definitions were applied: magnitude-based change, defined as achieving a prespecified minimally important difference regardless of baseline status, and categorical improvement, defined as movement from a clinically worse category at baseline to a more favorable category at follow-up, restricted to participants eligible to improve. Responder proportions at 6 and 12 months were calculated with 95% CI. Exploratory univariate logistic regression models were used to examine associations between baseline characteristics and responder status.

Missing follow-up data were evaluated using several complementary approaches. In unadjusted comparisons, participants with incomplete follow-up had significantly lower baseline well-being (WHO-5; P=.03) and life satisfaction (SWLS; P=.01) than participants with complete follow-up, whereas no statistically significant differences were observed for age, sex, comorbidity burden, relationship status, baseline depression/anxiety medication use, illness perceptions (B-IPQ), perceived stress (PSS-10), or disease-related worry (QSC-R10) (Table S2 in Multimedia Appendix 1). Complete follow-up was defined as completion of assessments at baseline, 6 months, and 12 months; follow-up was nonmonotonic, with some participants missing the 6-month assessment but returning at 12 months. Missingness was further evaluated using logistic regression models for incomplete follow-up, including all prespecified covariates and primary psychological outcomes as predictors. None of the included variables was statistically significantly associated with incomplete follow-up in the multivariable model. Primary analyses used covariate-adjusted GEE models and included all available observed outcome data, regardless of follow-up status. To evaluate the robustness of findings to incomplete follow-up, sensitivity analyses were conducted using both complete case analyses and analyses using stabilized inverse probability weights for attrition. Findings were highly consistent across the primary, complete case, and weighted analyses, indicating that the principal results were not materially altered by the analytic approach to incomplete follow-up.

All analyses were conducted using R software (version 4.5.1; R Foundation for Statistical Computing) [52]. Statistical tests were 2-sided, with a significance level of α=.05. Given the number of outcomes assessed, no formal adjustment for multiple comparisons was applied; results for secondary and exploratory analyses are therefore interpreted with emphasis on effect sizes, CI, and consistency of findings rather than statistical significance alone, and these findings are considered hypothesis-generating rather than confirmatory. Full model–based estimated marginal means and associated SEs for continuous outcomes are provided in Multimedia Appendix 1.

A total of 182 individuals enrolled and completed the baseline assessment. Across outcomes and follow-up assessments, outcome-specific completion ranged from 106 of 182 (58.2%) participants to 119 of 182 (65.4%) participants (Figure 1; Table S4 in Multimedia Appendix 1). Baseline characteristics are shown in Table 1. Participants had a median age of 60 (IQR 51‐68) years; of the 182 participants, 155 (85.2%) were women, and 89 (48.9%) resided in the Municipality of Pictou County. The median number of diagnosed chronic conditions was 3 (IQR 1‐3), and the median age-adjusted Charlson Comorbidity Index was 2 (IQR 1‐3). The baseline symptom burden was substantial: of 182 participants, 52.7% (n=96) screened positive for low well-being (WHO-5<50), 41.2% (n=75) screened positive for high perceived stress (PSS-10≥20), and 48.4% (n=88) screened positive for psychological distress. Additional baseline sociodemographic, clinical, and health care use characteristics are reported in Table S3 in Multimedia Appendix 1. The follow-up status for each outcome is reported in Table S4 Multimedia Appendix 1.

In adjusted population-averaged GEE models, all primary psychological outcomes improved from baseline to 6 months, with effects sustained or larger at 12 months (Table 2; Figure 2). Illness perceptions (B-IPQ) decreased at 6 months (β=−5.1, 95% CI −7.3 to −3.0; P<.001) and 12 months (β=−7.3, 95% CI −9.7 to −4.9; P<.001). Perceived stress (PSS-10) declined at 6 months (β=−2.2, 95% CI −3.2 to −1.2; P<.001) and 12 months (β=−2.4, 95% CI −3.5 to −1.3; P<.001), and disease-related worry (QSC-R10) decreased at 6 months (β=−3.7, 95% CI −5.2 to −2.3; P<.001) and 12 months (β=−6.3, 95% CI −7.8 to −4.7; P<.001). Life satisfaction (SWLS) increased at 6 months (β=1.5, 95% CI 0.5-2.6; P=.004) and 12 months (β=2.5, 95% CI 1.4-3.5; P<.001). Well-being (WHO-5) increased at 6 months (β=10.1, 95% CI 7.2-13.0; P<.001) and 12 months (β=10.4, 95% CI 7.2-13.5; P<.001). Mental health–related quality of life (SF-12 MCS) improved at 6 months (β=3.5, 95% CI 1.8-5.2; P<.001) and 12 months (β=4.3, 95% CI 2.4-6.2; P<.001).

Several secondary outcomes improved over follow-up (Table 2; Figure 2). Fatigue (FACIT-F) improved at 6 months (β=3.6, 95% CI 2.0-5.2; P<.001) and 12 months (β=5.0, 95% CI 3.3-6.8; P<.001), and sleep quality (PSQI) improved at 6 months (β=−1.8, 95% CI −2.4 to −1.2; P<.001) and 12 months (β=−2.2, 95% CI −2.8 to −1.5; P<.001). Diet quality (REAP-S) improved at 6 months (β=3.0, 95% CI 2.4-3.7; P<.001) and 12 months (β=2.6, 95% CI 2.0-3.3; P<.001), and sedentary behavior (SBQ) decreased at 6 months (β=−5.7, 95% CI −10.0 to −1.4; P=.01) and 12 months (β=−8.7, 95% CI −13.2 to −4.2; P<.001). Spiritual well-being (FACIT-SP-12) improved at 6 months (β=2.9, 95% CI 1.8-4.0; P<.001) and 12 months (β=4.6, 95% CI 3.0-6.2; P<.001). Perceived social support (MOS-SSS) increased at 6 months (β=0.2, 95% CI 0.1-0.3; P<.001) but was not clearly different from baseline at 12 months (β=0.1, 95% CI −0.01 to 0.21; P=.08). Physical health–related quality of life (SF-12 PCS) improved at 6 months (β=1.8, 95% CI 0.3-3.3; P=.02) with a larger improvement at 12 months (β=2.9, 95% CI 1.0-4.9; P=.003). Weight and BMI decreased at both follow-ups (weight, 6 months: β=−7.2 lb, 95% CI −10.4 to −4.0, P<.001; 12 months: β=−10.3 lb, 95% CI −15.2 to −5.4, P<.001; BMI, 6 months: β=−1.2, 95% CI −1.8 to −0.7, P<.001; 12 months: β=−1.7, 95% CI −2.5 to −0.9, P<.001). Alcohol frequency decreased at 6 months (Drink score β=−0.5, 95% CI −0.8 to −0.2; P=.002) and 12 months (β=−0.6, 95% CI −0.9 to −0.4; P<.001). Binge drinking frequency showed a small reduction at 6 months (β=−0.21, 95% CI −0.42 to 0.00; P=.047) but was not clearly different from baseline at 12 months (β=−0.03, 95% CI −0.31 to 0.26; P=.90), and cannabis use frequency did not show a clear change at either follow-up (6 months: β=−0.13, 95% CI −0.27 to 0.01; P=.06; 12 months: β=−0.12, 95% CI −0.27 to 0.02; P=.09). Model-estimated marginal means and full contrasts for all continuous outcomes are provided in Tables S5 and S6 in Multimedia Appendix 1.

Relative to baseline, the odds of low well-being (WHO-5<50) were lower at 6 months (odds ratio [OR] 0.17, 95% CI 0.11-0.26; P<.001) and 12 months (OR 0.15, 95% CI 0.10-0.24; P<.001) (Table 3). The odds of high perceived stress (PSS-10≥20) were lower at 6 months (OR 0.31, 95% CI 0.21-0.48; P<.001) and 12 months (OR 0.20, 95% CI 0.13-0.33; P<.001). The odds of clinically elevated disease-related worry (QSC-R10>14) decreased at 6 months (OR 0.39, 95% CI 0.26-0.57; P<.001) and 12 months (OR 0.24, 95% CI 0.15-0.37; P<.001). Poor sleep (PSQI>5) decreased at 6 months (OR 0.45, 95% CI 0.30-0.68; P<.001) and 12 months (OR 0.33, 95% CI 0.21-0.52; P<.001), and high sedentary behavior (≥8 hours/day; ≥56 hours/week) decreased at 6 months (OR 0.59, 95% CI 0.39-0.89; P=.01) and 12 months (OR 0.43, 95% CI 0.28-0.65; P<.001). Obesity prevalence (BMI≥30) did not change at 6 months (OR 0.88, 95% CI 0.72-1.07; P=.20) or 12 months (OR 0.88, 95% CI 0.67-1.15; P=.30).

Responder analyses showed clinically meaningful improvement across multiple domains (Table S7 in Multimedia Appendix 1). At 6 and 12 months, respectively, 39 of 107 (36.4%, 95% CI 27.3%-45.6%) and 46 of 106 (43.4%, 95% CI 34.0%-52.8%) participants achieved a ≥4-point reduction in perceived stress (PSS-10), while 53 of 108 (49.1%, 95% CI 39.6%-58.5%) and 47 of 108 (43.5%, 95% CI 34.2%-52.9%) achieved a ≥10-point improvement in mental well-being (WHO-5). At 12 months, clinically meaningful improvements were also observed for illness perceptions (B-IPQ; 54/108, 50.0%, 95% CI 40.6%-59.4%), disease-related worry (QSC-R10; 56/111, 50.5%, 95% CI 41.1%-59.8%), fatigue (FACIT-F; 62/109, 56.9%, 95% CI 47.6%-66.2%), physical health–related quality of life (SF-12 PCS; 60/111, 54.1%, 95% CI 44.8%-63.3%), mental health–related quality of life (SF-12 MCS; 53/111, 47.7%, 95% CI 38.5%-57.0%), and ≥3% weight loss (52/115, 45.2%, 95% CI 36.1%-54.3%).

Exploratory adjusted models examined associations between changes in prespecified candidate factors and changes in psychological outcomes (Table 4). Improvements in fatigue were consistently associated with improvements in well-being, life satisfaction, perceived stress, and disease-related worry. From baseline to 6 months, improvement in FACIT-F scores was associated with improvement in WHO-5 (β=0.9, 95% CI 0.6 to 1.2; P<.001), SWLS (β=0.3, 95% CI 0.2-0.4; P<.001), PSS-10 (β=−0.4, 95% CI −0.5 to −0.3; P<.001), and QSC-R10 (β=−0.6, 95% CI −0.8 to −0.5; P<.001) scores.

Improvements in sleep quality were similarly associated with improvements across psychological outcomes. From baseline to 6 months, improvement in PSQI scores was associated with improvement in WHO-5 (β=−2.2, 95% CI −3.1 to −1.3; P<.001), SWLS (β=−0.8, 95% CI −1.2 to −0.5; P<.001), PSS-10 (β=0.8, 95% CI 0.4 to 1.1; P<.001), and QSC-R10 (β=1.4, 95% CI 0.9 to 1.8; P<.001).

More favorable illness perceptions were also associated with improvements across psychological outcomes. From baseline to 6 months, improvement in B-IPQ scores was associated with improvement in WHO-5 (β=−0.6, 95% CI −0.8 to −0.4; P<.001), SWLS (β=−0.2, 95% CI −0.3 to −0.1; P<.001), PSS-10 (β=0.3, 95% CI 0.2-0.3; P<.001), and QSC-R10 (β=0.4, 95% CI 0.3-0.5; P<.001). Similar patterns of association were observed at 12 months (Table 4). Additional exploratory associations between changes in candidate factors and psychological outcomes are reported in Tables S8 and S9 in Multimedia Appendix 1.

Exploratory moderation analyses assessed whether baseline characteristics modified trajectories of psychological change (Table 5). Higher baseline perceived stress was associated with larger reductions in perceived stress at 12 months (time × baseline PSS-10: β=−0.4, 95% CI −0.6 to −0.3; P<.001) and greater improvement in well-being (time × baseline PSS-10 for WHO-5: β=0.6, 95% CI 0.2-1.1; P=.007). Higher baseline well-being was associated with smaller improvement in well-being (time × baseline WHO-5: β=−0.2, 95% CI −0.3 to −0.1; P<.001), and higher baseline life satisfaction was associated with smaller gains in life satisfaction (time × baseline SWLS: β=−0.4, 95% CI −0.5 to −0.2; P<.001). Baseline illness perceptions did not modify trajectories of well-being (time × baseline B-IPQ: β=0.13, 95% CI −0.10 to 0.37; P=.30) or perceived stress (time × baseline B-IPQ for PSS-10: β=−0.06, 95% CI −0.15 to 0.04; P=.20). Full interaction outputs are provided in Table S10 in Multimedia Appendix 1.

This secondary analysis of PictouCountyPEP found sustained improvements across multiple psychological, behavioral, and physical health domains over 12 months in rural adults living with chronic conditions. Improvements were observed across all primary psychological outcomes and were accompanied by favorable changes in fatigue, sleep, diet quality, sedentary behavior, spiritual well-being, and mental health–related quality of life. Exploratory analyses further suggested that psychological improvements co-occurred with changes in fatigue, sleep, illness perceptions, social support, and spiritual well-being, pointing to plausible multidomain pathways through which digital self-management interventions may influence well-being over time.

Exploratory analyses further showed that improvements in several candidate factors, notably fatigue, sleep quality, illness perceptions, social support, and spiritual well-being, were associated with improvements in psychological outcomes. These patterns of co-occurring improvement may help identify candidate multidomain pathways for future investigation but should not be interpreted as evidence of statistical mediation or causal mechanisms. Establishing whether these factors mediate intervention-related psychological outcomes will require an appropriate temporal and causal analytic framework, ideally within a controlled study designed prospectively to test these pathways.

Social support and binge drinking frequency were less clearly improved at 12 months, and improvements in diet quality were somewhat attenuated, but the overall pattern remained favorable. Binary threshold and responder analyses were broadly concordant with the continuous outcome models. Obesity status was the main exception: although the odds of meeting the binary obesity threshold did not change, continuous BMI and weight outcomes improved, as did responder analyses based on clinically meaningful weight loss. Program engagement, adherence, and acceptability have been reported elsewhere and are therefore not repeated here [20].

These findings contribute to the growing literature on digital health interventions for chronic disease self-management in community settings [53]. Importantly, this study evaluated the program under real-world delivery conditions rather than within a tightly controlled clinical trial environment. Participants accessed the intervention remotely using their own devices, engagement occurred asynchronously, and optional components were used flexibly. These characteristics provide context for interpreting the observed outcomes and for future evaluation of feasibility, accessibility, and acceptability in routine community settings [53]. Such implementation-oriented evidence is particularly relevant for digital health interventions intended to reach geographically dispersed populations [20,53].

The sustained psychological improvements observed at 12 months contrast with several recent systematic reviews and meta-analyses of digital interventions, which commonly report attenuation of mental health effects over time. For example, digital behavioral activation interventions have shown reductions in depressive symptoms at shorter follow-up, but no significant pooled benefit at 12 months; pooled anxiety outcomes were not significantly improved at the assessed follow-up points [54]. Digital peer support interventions have similarly demonstrated weaker mental health effects with longer follow-up [55], and digital exercise interventions have shown improvements at the end of the intervention that are not consistently maintained at later follow-up [56]. By contrast, psychological improvements in PictouCountyPEP remained evident at 12 months, during the lower-intensity maintenance period. Although direct comparisons across interventions and follow-up designs should be made cautiously, these findings suggest that comprehensive, multicomponent approaches warrant further evaluation for their potential to support longer-term psychological benefits. These findings complement evidence that interventions targeting multiple health behaviors can support chronic disease management [11] and indicate a need to evaluate their longer-term psychological outcomes.

The durability of effects may also reflect intervention design choices intended to reduce common barriers to digital engagement in rural settings, including low-bandwidth delivery, brief daily content, and minimal onboarding requirements [57]. Although engagement mechanisms were not directly tested in the present study, this microdose format may support habit formation and continued participation among individuals managing fluctuating symptoms and competing life demands [58]. These findings are also relevant to the well-described mismatch between health care systems historically oriented toward acute, reactive care and the proactive, ongoing support required for effective chronic disease self-management [59,60]. Many determinants most closely linked to symptom burden and progression unfold between clinical encounters, including sleep disruption, fatigue, stress, dietary patterns, social connection, and illness perceptions [60]. In this context, PictouCountyPEP may function as a low-burden adjunct that supports daily self-regulation and patient capability in routine living, complementing rather than replacing biomedical care.

Exploratory moderation analyses addressed an important question for digital self-management programs: for whom do these interventions appear to work best? Participants with higher baseline perceived stress experienced larger reductions in stress and greater improvements in well-being over 12 months, whereas those with higher baseline well-being or life satisfaction showed smaller gains in those same domains. This pattern suggests that the program may confer greater benefit among individuals with greater baseline psychological need. We found no consistent evidence that demographic characteristics substantially altered psychological trajectories, although these exploratory analyses may have been underpowered to detect smaller subgroup differences.

The exploratory mechanistic analyses point to several plausible pathways through which psychological improvement may occur. One pathway may involve modification of upstream cognitive processes, including illness perceptions that shape coping and behavioral responses, consistent with Leventhal’s common-sense model of self-regulation [61]. Another may reflect increased engagement in meaningful, reinforced daily activities supported by social connection and sustained routines, aligning with behavioral activation frameworks [62]. More broadly, the observed pattern is consistent with interdependent changes across energy, sleep, cognition, and social resources that collectively support psychological recovery [61-63]. These findings extend earlier work on Prostate Cancer Patient Empowerment Program (PC-PEP), in which self-efficacy and illness perceptions explained improvements in psychological distress [64]. However, these analyses are exploratory and should not be interpreted as evidence of causal mediation. Future studies should formally test these candidate pathways using longitudinal mediation models or experimental designs.

Although this study was not designed to assess disease incidence, the improvements observed across stress, sleep, diet quality, sedentary behavior, and social support are relevant to secondary prevention in chronic disease. These domains represent potentially modifiable targets relevant to chronic disease self-management and health [60,65]. Future controlled studies with longer follow-up should examine whether these changes translate into downstream clinical outcomes, multimorbidity accumulation, or health service use.

Beyond the observed health improvements, PictouCountyPEP builds on a digital delivery model already implemented in PC-PEP across Canada and internationally. The program delivers core content through automated email prompts and brief asynchronous video modules, limiting the need for clinicians to deliver individual sessions. An international phase 4 implementation trial of PC-PEP reported similar follow-up completion across rural and urban participants, high program acceptability, and sustained improvements in mental health [66]. This implementation experience supports the scalability of the delivery model and provides a foundation for extending it to rural chronic disease self-management. Broader implementation of PictouCountyPEP should build on this experience while accounting for local digital access, staffing arrangements, and integration with existing services [57,67].

From a health system perspective, scalable self-management programs that can be delivered at low cost and without reliance on specialized personnel are of increasing interest, particularly in rural and resource-constrained settings [66,68]. An economic evaluation within the PC-PEP randomized controlled trial estimated program delivery costs of CAD $200.07 (CAD $1=US $0.73 as of August 29, 2025) per participant and found that early delivery improved outcomes while lowering overall health care costs [68]. These findings provide an economic rationale for broader implementation of the delivery model. A formal cost-effectiveness analysis of PictouCountyPEP from the health care system perspective is currently underway to assess downstream resource use and longer-term value. These findings also have implications for how digital self-management programs could be integrated into chronic disease care pathways. In many health systems, patients with multimorbidity receive episodic clinical care but limited ongoing behavioral or psychosocial support between visits [59,60]. Digital empowerment programs may complement traditional care by providing continuous, low-intensity behavioral guidance that patients can access independently. For primary care teams serving rural populations, such programs could function as an adjunct resource that reinforces lifestyle recommendations, supports stress regulation, and promotes patient activation without substantially increasing clinical workload. Work is currently underway in collaboration with Nova Scotia Health to explore potential implementation pathways for the program within existing care structures. Together with the ongoing cost-effectiveness analysis, these efforts aim to determine whether digitally delivered self-management programs can offer a sustainable and cost-efficient complement to traditional chronic disease care models.

Although PictouCountyPEP was not designed as a nurse-delivered intervention, its asynchronous, home-based structure has potential implications for nursing and primary care practice in rural settings. Nurse-led remote digital approaches have been used to support adults with chronic conditions through health education, remote support, monitoring, and other forms of digitally enabled care [69]. More broadly, digital interventions are increasingly being evaluated as adjuncts to chronic disease self-management, although their effectiveness varies across conditions, intervention components, and self-care domains [65]. Rather than requiring nurses to deliver the intervention content directly, a program of this type could provide a standardized adjunct to health education and self-management support, with nurses and other primary care professionals potentially identifying appropriate patients, reinforcing individualized self-management goals, recognizing emerging needs during routine encounters, and directing patients to additional clinical or community services when required. This model may be particularly relevant in settings where clinical time and access to specialized chronic disease services are constrained because the core educational and behavioral content is delivered outside the clinical encounter and does not require continuous clinician involvement [69].

These implementation questions are particularly relevant in the present rural context: at baseline, 19 of 182 (10.4%) participants reported receiving primary care from another provider such as a nurse practitioner, while 46 of 182 (25.3%) participants reported having no regular primary care provider (Table S3 in Multimedia Appendix 1). Importantly, however, these potential nursing and primary care roles were not evaluated in the present study. Future implementation research should examine how referral to digital self-management programs can be incorporated into rural primary care and nursing workflows, including which patients are most appropriate for referral, whether some patients require additional adherence or digital navigation support, and whether integration can extend self-management support without increasing clinical workload.

This study has several limitations. First, the study was conducted in a single, predominantly rural region, and the sample was predominantly female (85.2%) and White (96.2%), limiting generalizability to men and more ethnoculturally diverse populations. Eligibility also required English proficiency and access to the internet and email; consequently, the findings cannot be assumed to generalize to non–English-speaking populations or to individuals experiencing greater barriers to digital access or with lower digital health literacy. Although the intervention was intentionally designed for low-bandwidth delivery and did not require specialized applications, wearable devices, or synchronous participation, these design features do not eliminate potential digital access or digital literacy barriers. Generalizability to other rural regions may also depend on local sociodemographic characteristics, health care access, digital infrastructure, and community resources [57,67]. Similarly, the findings should not be assumed to apply directly to urban populations, where health service availability, referral pathways, and opportunities for in-person chronic disease support may differ. Evaluation in more diverse populations and geographic settings is therefore needed. Second, the single-arm design precludes causal attribution of the observed longitudinal changes to PictouCountyPEP. Without a concurrent control group, changes associated with natural disease trajectories, secular or time effects, regression to the mean, unmeasured co-interventions, and residual confounding cannot be separated from changes occurring during program participation. Although covariate adjustment and sensitivity analyses strengthen the assessment of longitudinal associations, they do not overcome this design limitation. Future randomized controlled trials should evaluate the efficacy of PictouCountyPEP relative to an appropriate comparator and determine whether the observed multidomain changes are attributable to the intervention. Third, all outcomes were self-reported and may therefore be subject to recall, reporting, and social desirability biases. The study also did not assess longer-term clinical or health service outcomes, such as hospitalization or mortality; consequently, the observed changes in patient-reported outcomes should not be interpreted as evidence of effects on these outcomes. Fourth, participants with incomplete follow-up had lower baseline well-being and life satisfaction, raising the possibility of attrition-related bias. However, none of the included variables was significantly associated with incomplete follow-up in the adjusted model, and the primary findings were consistent in complete case and weighted sensitivity analyses. Fifth, because multiple outcomes and exploratory analyses were examined without formal adjustment for multiple comparisons, some statistically significant findings may reflect chance. Accordingly, secondary and exploratory findings should be interpreted with caution, with emphasis on consistency and effect size rather than statistical significance alone. Finally, responder analyses were only possible for measures with available thresholds for meaningful change; for some outcomes, such thresholds remain poorly defined in general chronic disease populations. Future research should prioritize the development and validation of clinically meaningful benchmarks across relevant outcomes in chronic disease settings. Notwithstanding these limitations, strengths of the study include its comprehensive multidomain scope, use of validated instruments, prospective longitudinal design, and analytic approach incorporating all available observed outcome data, supported by complete case and inverse probability–weighted sensitivity analyses.

PictouCountyPEP is a digitally delivered self-management program associated with sustained improvements across multiple psychological, behavioral, and physical health domains in rural adults living with chronic conditions. These findings suggest that low-burden, multicomponent digital interventions may support sustained improvements in mental well-being alongside favorable changes in related behavioral and psychosocial outcomes. The program’s low-bandwidth, asynchronous delivery model highlights its potential scalability and relevance for rural and underserved populations, where access to ongoing behavioral and psychosocial support is limited. By addressing shared, upstream determinants of chronic disease across conditions, digitally delivered self-management programs may complement traditional care by extending support into patients’ daily lives without increasing clinical workload. Future research should evaluate long-term clinical and health system outcomes, including health service use and cost-effectiveness, and prospectively test the candidate pathways suggested by these exploratory analyses to inform intervention optimization and implementation at scale.