This 12-week pilot study used a single-arm, pretest and posttest, mixed methods design. We collected both quantitative and qualitative data, with the quantitative data analyzed to assess study feasibility, self-monitoring behaviors, and within-subject changes in health outcomes (eg, hemoglobin A_1c [HbA_1c], physical function) from baseline to 12 weeks, representing before-and-after health technology–assisted lifestyle self-monitoring. Qualitative data were collected through semistructured participant interviews to help explain the quantitative findings, particularly patterns of engagement in self-monitoring and observed changes in health outcomes, as well as to explore participants’ experiences using health technologies for lifestyle self-monitoring and identify facilitators and barriers to lifestyle modification for diabetes management.

The inclusion criteria were (1) age 60 years and over, (2) self-reported diagnosis of type 2 diabetes with the confirmation of a health care provider’s report or laboratory report following the American Diabetes Association criteria [27], (3) being overweight or obese (BMI ≥27 kg/m²), (4) having a smartphone, and (5) willingness to participate in the study and provide informed consent.

The sample size calculation was based on prior evidence suggesting that a minimum of 12 participants is a rule of thumb for a pilot study [28]. Assuming a 20% dropout rate based on prior studies in older adults [29], this pilot study proposed enrolling 15 older adults with type 2 diabetes into the study.

Participants were recruited between November 2020 and June 2021 using multiple approaches, including social media platforms (eg, Facebook, Nextdoor), flyers posted at public places (eg, the University of Texas Health Science Center at San Antonio [UTHSCSA]), word of mouth, and mass invites via the UTHSCSA electronic medical record messaging system.

This study was reviewed and approved by the institutional review board (IRB) at UTHSCSA (IRB application number HSC20200531H) with the statement “Your request to conduct this minimal risk research was approved by Expedited Review on August 11, 2020.” Written informed consent was obtained from each participant prior to study enrollment.

Participant privacy and confidentiality were protected by assigning unique study identifiers and storing data on secure, password-protected institutional servers or in a locked drawer in the principal investigator’s office. Wearable and survey data were deidentified prior to analysis. Only authorized study personnel had access to identifiable information.

Participants received US $15 in compensation for completing data collection at baseline and at the 12-week follow-up, respectively, with up to US $30 in compensation for their participation. Participants were allowed to keep the Fitbit device after the study.

Enrolled participants were instructed to track physical activity and food intake using a Fitbit Inspire HR fitness tracker (Fitbit Inc) and its corresponding smartphone app for over 12 weeks. All tracked data were automatically synchronized to UTHSCSA’s connected health platform. Each participant was provided with a Fitbit fitness tracker at no cost to use.

In addition to standard Fitbit instructions, each participant received a simplified step-by-step instruction handbook with a large font size and pictures created by the study team. The handbook included key points for using the device to track physical activity and food intake. The study team helped each participant in setting up all devices and guided them through the lifestyle tracking process step by step. The study team was available via phone during weekdays for any troubleshooting of technological issues or lifestyle logging questions. The study team could see all participants’ tracked lifestyles through the connected health platform. For those who did not log in for 2 consecutive weeks, the study team would call and remind participants to self-monitor. The study was focused on assessing technology-assisted self-monitoring, and no structured lifestyle education intervention was implemented as part of the study.

Following the institution’s COVID-19 research implementation protocol, all participants completed a telephone screening one day prior to each scheduled study appointment to assess COVID-19–related symptoms or potential exposure within the preceding 15 days. Study appointments would be rescheduled if participants reported any symptoms or exposures; however, no participants screened positive during the study period. In-person contact was minimized whenever possible, and mask wearing was encouraged but not required during in-person visits. All equipment used for data collection was sanitized after each participant visit.

Demographic characteristics (eg, age, race or ethnicity, education) were collected at baseline using a survey questionnaire. Health outcome data were collected at baseline and at 12 weeks at UTHSCSA School of Nursing by trained research staff.

Sample characteristics and self-monitored lifestyle data were summarized using frequencies and percentages for categorical variables and mean and SD for continuous variables. Baseline and 12-week health outcomes were compared using paired 2-tailed t tests or Wilcoxon signed rank tests for variables that were not normally distributed. Effect sizes were calculated using Cohen d to quantify the magnitude of change from baseline, and 95% CIs were calculated for Cohen d values. The associations of PDWTSs, PDWFLs, and 7-day mean steps with 12-week HbA_1c and 12-week physical function (SPPB) were examined using Pearson correlation analysis or Spearman correlation analysis if not normally distributed. Normality was assessed using the Shapiro-Wilk test. Variables with a P value of <.05 were determined to be not normally distributed, including body weight, chair test, and total SPPB score. Given the sample size of the pilot study, missing baseline SPPB-related variables were not imputed. Analyses for balance, gait speed, chair stand, and total SPPB score were conducted using available data from the 8 participants with complete measurements. In small pilot studies, imputation may produce unstable estimates due to limited data for model estimation; therefore, complete-case analysis is commonly recommended when the primary aim is feasibility assessment and preliminary effect estimation [32,33]. Self-monitored data were analyzed using Excel (Microsoft 365). Other quantitative data were analyzed using R (version 4; R Foundation for Statistical Computing); the significance level was set at .05, 2-sided.

Two researchers with training and experience in qualitative methods independently analyzed the qualitative data using a combination of deductive and inductive thematic analysis. First, the data were independently coded using open coding in NVivo 11 (version 11; QSR International) by the study investigator (YD) and a trained research assistant. Discrepancies between coders were discussed, and consensus was reached to ensure consistency. An initial codebook with operational definitions was developed and refined throughout the coding process. Although formal procedures for data saturation were not prospectively defined, all 15 participants enrolled in the study were interviewed. Given the pilot nature of the study and the relatively focused research questions, the interviews yielded recurring themes, suggesting that the data were sufficient to capture key participant experiences. Analytic credibility was supported through investigator triangulation and systematic documentation of coding decisions.

Second, when reviewing the codes, themes were organized using the social ecological model as an analytic framework to examine how facilitators and barriers to technology use and lifestyle modifications are influenced by factors across multiple levels [34]. The social ecological model consists of five levels of factors: (1) societal or policy (factors concerning large-scale issues, such as social and cultural norms, and guiding rules, such as policies or laws), (2) community (factors limited to a specific geographic area, such as neighborhoods), (3) organizational (factors at institutional settings, such as schools and workplaces), (4) interpersonal (factors involved with direct person-to-person interaction), and (5) individual (personal characteristics, such as health and beliefs) [35,36]. The Centers for Disease Control and Prevention uses a 4-level social-ecological model by combining the community and organizational levels into a single category [37]. We adopted this 4-level approach in our analysis. Relationships among factors within the social ecological model levels were explored,

Finally, the qualitative findings were explicitly integrated with the quantitative results to support mixed methods triangulation. This integration facilitated a comprehensive interpretation of how factors from multiple levels could influence self-monitoring behaviors, lifestyles, and observed clinical outcomes.

The average age of the study sample was 70.5 (SD 4.8) years (Table 1). Forty percent (6/15) of the sample was female, and 73.3% (11/15) of the study participants self-identified as Hispanic. All participants reported some college education or higher. Over half (10/15, 66.7%) of the participants were obese, and the mean baseline HbA_1c level was 7.9% (SD 1.9%) at baseline.

The targeted sample size was successfully enrolled over approximately 6 months (from November 2020 to June 2021) after screening 20 subjects who expressed interest in participation. An additional 6 individuals expressed interest, but they were not screened as the targeted enrollment had been reached. All (n=15, 100%) participants completed both baseline and 12-week follow-up data collection.

All 15 enrolled participants used a Fitbit fitness tracker and its associated mobile phone app to self-monitor their physical activity and dietary intake, although adherence varied. Multimedia Appendices 2-4 show the Fitbit fitness-tracked data. Over the 12-week study period, most participants tracked their steps daily, with 11 out of 15 participants tracking their steps for more than 90% of the study days. Seven out of 15 participants had an average of 5000 daily steps. Nine out of 15 participants tracked over 60% of days with food logs, and only 4 out of 15 participants tracked their food for over 90% of days. Additional feasibility-related findings are reported in the qualitative findings section.

Figure 1 illustrates HbA_1c values at baseline, 12 weeks, and individual-level changes over time for each participant. All participants with a baseline HbA_1c greater than 7 had a decrease in HbA_1c at the 12-week follow-up (HbA_1c changes ranging from −0.1 to −3.7). Those with a baseline HbA_1c above 9 showed the greatest reductions (HbA_1c changes ranging from −1.2 to −3.7). Participants with baseline HbA_1c levels below 7 maintained levels below 7 at the 12-week follow-up.

Table 2 displays the changes in weight, BMI, HbA_1c, and physical function from baseline to 12 weeks. The mean HbA_1c decreased from 7.93 (SD 1.93) to 7.32 (SD 1.08) at 12 weeks, with a statistically significant reduction (mean change −0.61, 95% CI −1.18 to −0.05; effect size −0.49; P=.04). Although weight and BMI also showed improvements, these changes were not statistically significant. However, the effect sizes suggest potentially clinically relevant trends. Specifically, body weight had a large effect size of −0.83; BMI also decreased, with a moderate effect size of −0.61. No significant SPPB improvements were found over the study period.

Table 3 demonstrates the correlations of Fitbit-tracked data with average daily steps and health outcomes (BMI, HbA_1c, and SPPB) at 12 weeks. PDWTSs, PDWFLs, and average daily steps showed expected trends in their associations with BMI, HbA_1c, and SPPB, but not all of the correlations were statistically significant. PDWTSs and PDWFLs showed a high correlation with each other (r=0.71; P=.002). PDWFL was significantly correlated with lower HbA_1c levels (r=−0.53; P=.04). In addition, a negative relationship between BMI and SPPB was noted (r=−0.57; P=.03), indicating that higher BMI was related to lower physical function.

The qualitative findings demonstrated participants’ experiences with technology-assisted lifestyle monitoring and provided insights into factors influencing the feasibility, engagement, and sustainability of diabetes self-management behaviors. Barriers and facilitators to lifestyle behaviors for diabetes management were identified across multiple ecological levels, including individual, interpersonal, organizational/community, and societal factors (Figure 2).

The factors at different levels influence both the feasibility of technology use and healthy lifestyle behaviors. For instance, an individual’s self-monitoring practices and motivation may be strengthened by interpersonal support from family and friends, while access to community programs and health education further reinforces these behaviors. Additionally, societal-level influences, such as health care policies and cultural norms, can either enable or hinder diabetes management by shaping access to resources and behavioral expectations. We described the factors at each level below. The participant identification number and gender are provided following each example quote.

Several factors were found at the individual level, such as knowledge, perceived benefits, goals, self-monitoring, and health status. Participants mentioned that their self-discipline, motivation, self-efficacy, and goal setting influenced their ability to engage in physical activity and maintain a healthy diet. However, challenges such as fatigue, pain, and time constraints hindered adherence to recommended behaviors. Examples of participant quotes are provided below.

In addition, at the individual level, older adults frequently report challenges with using Fitbit and its paired smartphone app to track food intake, primarily due to limited technological literacy. Some participants found it difficult to navigate the app or use it consistently in real time. As a result, many resorted to writing down their meals on paper or a calendar first and then manually transferring the information to the app later.

At the interpersonal level, diabetes management behaviors could be influenced by others, such as family members and friends. For instance, challenges involved managing family dynamics around dietary preferences. While family and friends provided emotional support and encouragement, conflicting food choices and social eating norms sometimes made it difficult to sustain healthy eating habits. Peer support and accountability played a role in promoting physical activity, with some participants relying on social networks to stay engaged.

At the community and organizational levels, participants mentioned access to physical activity programs and healthy food options as important facilitators. However, logistical barriers, such as the limited availability of facilities, transportation challenges, and financial constraints, were reported. Access to health education and relevant community resources was also noted as a crucial factor influencing lifestyle behaviors.

At the societal level, the role of health providers as advocates and sources of guidance was emphasized. Participants expressed that provider involvement and encouragement significantly impacted their motivation to adhere to diabetes management behaviors. Broader societal factors, such as holidays, cultural norms, insurance policies, and unexpected societal disruptions such as the COVID-19 pandemic, also shaped lifestyle behaviors by affecting access to care, dietary patterns, and social routines.

Participants who demonstrated consistent step tracking and dietary logging described strong motivation, clear goal-setting strategies, and reinforcement from interpersonal support, which aligned with observed associations between food logging frequency and lower HbA_1c levels (eg, participant 3). Conversely, qualitative barriers related to limited digital literacy, pain, fatigue, and competing social demands provided explanatory context for inconsistent dietary logging and the modest glycemic outcomes. Multilevel influences, such as interpersonal encouragement and access to community resources, emerged as key factors that supported engagement with self-monitoring despite technological challenges, highlighting conditions under which technology-assisted lifestyle self-monitoring may be more feasible and acceptable among older adults with type 2 diabetes.

This mixed methods study explored the feasibility and potential health benefits of using wearable and mobile devices for older adults to manage type 2 diabetes and identified potential strategies to support lifestyle modification for diabetes management. The study proved to be feasible even during the challenging period when COVID-19 was still actively circulating in the community. The quantitative results highlight the potential benefits of digital devices in supporting self-monitoring and improving glycemic control among older adults with type 2 diabetes. A significant reduction in HbA_1c, as well as the positive association between HbA_1c and the percentage of days with food logs, suggests that consistent tracking of food intake plays a critical role in managing blood glucose levels. These findings align with previous research emphasizing the importance of health technologies and lifestyle modifications in diabetes management. The qualitative findings underscore the multilevel challenges and facilitators of achieving sustained behavioral changes. The findings suggest that multilevel interventions, which address individual motivation and self-efficacy, family and peer support, community resources, and health care involvement, are necessary for the comprehensive management of diabetes.

This pilot study demonstrated the feasibility of using technology-assisted lifestyle monitoring for older adults with type 2 diabetes. The strong enrollment rate and 100% (n=15) retention at 12 weeks indicate high engagement and acceptability, key indicators of feasibility in pilot research [38]. All participants used a Fitbit fitness tracker and its companion app for 12 weeks to monitor physical activity and diet with varied adherence. Prior studies have shown that older adults are capable of using wearable devices for physical activity tracking when provided with proper training and support [39,40]. However, dietary logging was found to be less consistent, with only 2 participants having logged food intake on over 90% (81/90) of days. This is consistent with the literature showing that food tracking can be cognitively demanding and burdensome, particularly for older adults [41]. This suggests the need to explore approaches to ease the food logging tasks and provide tailored support strategies. Additionally, strategies that offer ongoing, convenient, and hands-on support may help mitigate the challenges of low digital literacy among older adults.

The observed reduction in HbA_1c levels in this pilot study should be interpreted cautiously, given the small sample size and short duration. However, the findings are consistent with prior research suggesting that self-monitoring may support diabetes self-management [42]. Although we did not provide any additional interventions beyond instructing participants on how to use the Fitbit fitness tracker and its paired smartphone to track physical activity and food intake, our findings are consistent with previous studies suggesting that self-monitoring is associated with improved physical activity and dietary adherence, leading to improved glycemic outcomes in individuals with type 2 diabetes [43,44]. Although the average daily step counts were not significantly associated with BMI, glycemic control, or physical function, trends were in the expected directions. Additionally, emerging evidence also indicates that the timing of physical activity may influence glucose regulation, with postmeal or evening activity potentially offering advantages in glycemic management [43-45]. These findings highlight areas for further investigation, particularly in larger and more rigorously designed studies.

The qualitative findings helped contextualize variability in self-monitoring adherence, lifestyles, and health outcome trends observed in the quantitative data. Qualitative data reinforce the view that self-monitoring and lifestyle behaviors are shaped by a complex interplay of individual, interpersonal, organizational or community, and societal factors, echoing the social ecological model [34]. At the individual level, factors such as motivation, goal setting, and self-monitoring emerged as crucial determinants, yet physical limitations, pain, and fatigue were notable barriers, similar to what has been reported among older adults with chronic disease [45]. Social and familial contexts further influenced dietary behaviors and physical activity, with both supportive and conflicting dynamics observed. Previous studies have also shown that family engagement can enhance or impede diabetes self-care depending on the alignment of goals and routines [46]. Interventions for diabetes management, involving support from family members, peers, or friends, could be promising.

At the community level, access to programs and transportation challenges mirrored structural barriers commonly faced by older adults. Organizational support, such as restarting community exercise classes, was valued and perceived as helpful. These insights support broader calls for age-friendly and community-based programs, such as exercise classes and diabetes education for older adults [47]. On the societal level, participants highlighted the role of health care providers as motivators, though some noted insufficient attention to lifestyle counseling, suggesting an opportunity to better integrate behavioral support into health care settings [48] or to bridge health care and community services [49,50]. External societal events, such as the COVID-19 pandemic, were cited as disruptive to routine care and physical activity access, underscoring the need for adaptable and resilient health promotion strategies in the face of such events.

This study has several limitations that should be acknowledged. The study had a small sample size (n=15) using a 1-arm design, so the statistical power to detect significant changes was limited. In addition, SPPB data were not collected at baseline for the first 7 participants due to COVID-19 restrictions, which further constrained the interpretation of physical function outcomes. As a result, several observed trends may not have reached statistical significance. The lack of a control group limits the ability to attribute observed changes directly to the intervention. Future studies should include randomized control groups to strengthen causal inferences. The 12-week intervention period may have been insufficient to observe meaningful changes in physical function and other long-term health outcomes. Longer follow-up periods are needed to assess the sustainability of the observed improvements. The study population consisted predominantly of Hispanic participants, which may limit the generalizability of the findings to other ethnic groups. Additionally, participants were required to own smartphones, which might have introduced selection bias toward individuals who were more technologically engaged and health motivated. The study participants were of relatively moderate-to-high socioeconomic status, which may further limit generalizability. While wearable devices provide objective data, their health benefits may also depend on participants’ compliance and engagement with self-monitoring activities. Variability in participants’ adherence may have influenced the outcomes. The study defined PDWFLs and PDWTSs using the “any nonzero value” as a valid entry, which may not accurately capture the true level of engagement in physical activity and dietary intake. Finally, although all study participants were interviewed, the extent to which true data saturation was achieved may be limited, which may in turn constrain the robustness of the qualitative findings.

Using technology-assisted self-monitoring of lifestyle behaviors was feasible among a select group of older adults with type 2 diabetes who were smartphone users and relatively well educated. Although favorable within-subject signals were observed for glycemic control and related outcomes, these findings should be interpreted cautiously, given the study’s single-arm design and potential selection bias. Importantly, the results highlight that the benefits of technology-assisted self-monitoring are likely contingent on broader contextual factors, including digital and health literacy, access to devices, and the availability of interpersonal, organizational, and social support. To inform equitable and scalable lifestyle–based diabetes management interventions, future studies should investigate more diverse study populations and include older adults with lower digital literacy or limited access to technology. Additionally, qualitative findings revealed that multilevel lifestyle interventions facilitated by health technologies to enhance lifestyle-based diabetes self-management hold promise. However, the use of technology presents both opportunities and challenges. Further research is warranted to examine how health technologies can be integrated with multilevel strategies to support long-term, sustainable lifestyle-based diabetes management among older adults with type 2 diabetes.