TExT-MED+FANS was a 12-month comparative effectiveness randomized controlled trial that investigated the impact of two exposures: (1) mHealth-augmented social support (FANS) and (2) standard social support (mailed pamphlet) on patient health care use outcomes. Participants were enrolled from June 2017 to December 2019. This study represents a secondary analysis of the original randomized trial data, focused on differences in health care periods and subgroup characteristics not examined in the primary publication. For further information on the methods used in TExT-MED+FANS and the specific mHealth program under study, please refer to the protocol study by Burner et al [17].

The study received approval from the University of Southern California Health Sciences Institutional Review Board (HS-17‐00406). All participants provided written informed consent. Data were stored in an encrypted dataset and were not deidentified to allow linkage to electronic medical record (EMR) information. Consent procedures were conducted in English or Spanish, depending on patient preference. Patients and supporters were provided with compensation ranging from US $20 to US $280, depending on the number of study visits completed [16].

The primary outcome was health care use, assessed through EMR review. Outcome variables that were included are given below:

All calculations underwent scrutiny by 2 reviewers, with any discrepancies resolved through consensus.

The following measurement phases were used to assess changes in health care use over time:

At the time of enrollment, patients self-reported their sex, race or ethnicity, primary language, duration of diabetes, and access to primary care. Age and insurance status were extracted from the EMR. Insurance status was classified as means-tested (eg, Medicaid or other government insurance programs), non–means-tested (eg, Medicare or private insurance), or uninsured (no recorded insurance coverage at the time of the visit). The supporter’s sex was reported by the supporter, and for analysis purposes, supporters identified as cohabitating significant others by the patient were considered spouses. Supporters other than spouses were categorized as nonspouse supporters, encompassing various relationships such as siblings, friends, or children.

The intervention group assignment was otherwise examined with an intention-to-treat approach, regardless of patient or supporter dropping out of the TExT-MED or FANS text messages. Analysis was completed with STATA (version 17.0; StataCorp LLC). Data were examined with histograms and did not violate assumptions of heteroscedasticity. Baseline characteristics were examined descriptively as a whole and by intervention group (Table 1).

In our main effects analysis, we used linear regression models with outcomes of change scores of health care visits from (1) baseline to the end of intervention phases, and (2) from the end of intervention to the end of maintenance phase, examining differences in health care use by the FANS augmented versus traditional mailed pamphlet groups (Table 2). The main independent variable was intervention group assignment. No other independent variables were included in the main effects model. Differences between groups were estimated with the contrast and margins function. We individually examined unscheduled acute care visits (ED and urgent care visits), hospitalizations, and scheduled care (clinic) visits. For subgroup analysis, we used an interaction term for sex and supporter relationship to examine the moderating effect of supporter relationship and sex with the same procedures (Table 3).

In this study, we evaluated the impact of a mHealth-augmented social support intervention (FANS) compared to an active control (mailed pamphlet) on health care use among patients with diabetes, analyzing data from 165 participants to assess changes in acute unscheduled care visits, clinic visits, and hospitalizations. The main findings indicate a significant reduction in acute unscheduled care visits and hospitalizations for both intervention and active control groups, with this effect maintained after intervention. Moreover, an increase in clinic visits was seen during both the intervention and postintervention phases. In addition, we found that the type of supporter, particularly whether or not the supporter was a spouse, played a role in study participants’ use of the mentioned health services.

The reduction in acute unscheduled care visits aligns with existing literature supporting the effectiveness of mHealth interventions and educational pamphlets in improving patient self-management [18]. Previous studies have shown that providing structured support and education to patients with chronic diseases can lead to better health outcomes [19]. Specifically for patients with diabetes, these interventions have been linked to improved diabetes care, glycemic stability, and reduced A_1c levels [20]. Moreover, the integration of mHealth technologies has facilitated more consistent monitoring and timely interventions, enhancing overall disease management. Such approaches not only empower patients but also foster a proactive engagement in their health care journey, ultimately reducing the burden on acute care services [21].

There is a notable gap in the literature regarding health care use following mHealth interventions, particularly concerning ED visits, hospitalizations, and clinic visits for patients with diabetes. While extensive data exists on the impact of such interventions on other chronic conditions, like chronic obstructive pulmonary disease (COPD), similar comprehensive studies on diabetes-focused mHealth interventions are lacking. Yang et al [22] demonstrated that an mHealth curriculum for patients with COPD significantly reduced the risk of hospitalizations, although it did not yield a substantial difference in the average length of stay. In this study focusing on diabetes, a marked reduction in the frequency of hospitalizations was observed both during and after the intervention; however, we did not examine the length of stay. While both conditions benefit from mHealth interventions, the nature and progression of diabetes and COPD may present unique challenges and opportunities for reducing healthcare utilization, emphasizing the need for more tailored research in diabetes-specific programs.

The subgroup analysis revealed that while acute unscheduled care visits decreased significantly across most supporter groups, females with spouse supporters in the FANS group showed no significant change. The type of supporter may influence the intervention’s effectiveness, aligning with prior work indicating that spousal support by sex can vary widely in its impact on health behaviors [23]. These patterns underscore the complex role of social support dynamics, as highlighted in previous studies which have shown that the quality and type of support can differentially impact health outcomes [24,25]. Specifically, research has emphasized the effectiveness of spousal support in chronic disease management, including diabetes, due to the consistent and close nature of the support provided [26]. Clinic visits increased significantly for females with spouse supporters during the intervention phase in both groups, which aligns with studies indicating that females are generally more proactive in seeking preventive care [27-29]. In this study, this increase was only sustained in the FANS group, suggesting the potential for digital health tools to maintain engagement with health care services. This study’s findings align with these insights, suggesting that while mHealth interventions can be broadly beneficial, their impact may be moderated by the type of support relationship involved.

The findings of this study have substantial implications for public health and policy. The reduction in acute unscheduled care visits among patients with diabetes through mHealth-augmented social support interventions highlights the potential of such technologies in enhancing chronic disease management. Policy makers should consider integrating mHealth solutions into standard care protocols, particularly for populations with limited access to care, to improve health care accessibility and outcomes [20,30,31]. In addition, investing in mHealth platforms could alleviate the financial burden on health care systems by reducing the frequency of costly acute unscheduled care visits and hospitalizations [32,33]. Public health initiatives should also focus on promoting digital literacy and access to mHealth technologies, ensuring that communities that have been historically marginalized can fully benefit from these advancements. By incorporating mHealth interventions into broader health care strategies, a more sustainable and equitable health care system can be created that better addresses the needs of patients with chronic conditions like diabetes. Future research and policy initiatives should also consider incorporating structured family or supporter-based engagement as a mechanism to enhance intervention effectiveness and long-term adherence.

This study has several limitations that must be considered when interpreting the findings. First, the small sample size and the secondary analysis nature of the study may limit the internal validity of the results. Using secondary data introduces inherent limitations, including potential bias related to how the original study was designed and executed. Since this analysis was not prespecified in the original study protocol, there may be risks of selective reporting or post hoc interpretation. In addition, certain variables of interest may not have been collected or controlled for in the original trial design, potentially limiting the scope of inferences. Future research with larger, more diverse populations is needed to confirm these findings. Second, acute unscheduled care visits in this study included both urgent care and ED visits, due to the structure of the Los Angeles County Department of Health Service Electronic Health Record System. While urgent care visits are less costly than ED visits, they still are inferior to scheduled clinic care and are not the desired health care use pattern [34,35]. Third, the sample was predominantly composed of individuals from Hispanic or Latino and low-socioeconomic backgrounds who used means-based insurance. These findings may not be generalizable to populations with different demographic and socioeconomic characteristics. Despite these limitations, the study provides valuable insights into the potential benefits of mHealth interventions. Addressing limitations and exploring tailored approaches can enhance the efficacy and reach of such interventions.

This study contributes to the growing body of evidence examining the use of mHealth tools to support chronic disease management, reduce the burden on health care systems, and support the effectiveness of structured support systems and educational interventions in enhancing patient self-management and reducing health care use. mHealth-augmented social support interventions, such as the FANS program, may be able to reduce acute unscheduled care visits and hospitalizations among patients with diabetes. Though more work in this area is needed to identify who will most benefit, we saw indications that these interventions, particularly when involving spouse supporters, can lead to meaningful improvements in health care outcomes and engagement with preventive care services. The significant reductions in unscheduled care visits suggest that mHealth-augmented social support can effectively improve diabetes management, particularly for disproportionately affected groups. These findings underscore the importance of incorporating digital health solutions into chronic disease management social support strategies to improve health care outcomes and reduce acute care reliance.