The current study is a secondary analysis of the pilot RCT that examined the potential efficacy of the CARE program on HbA_1c levels. A total of 60 Chinese American participants with T2D were enrolled and randomly assigned in a 1:1 ratio to either the CARE intervention group or a waitlist control group. This study follows the CONSORT (Consolidated Standards of Reporting Trials) guidelines (Multimedia Appendix 2) [25]. Detailed descriptions of the pilot RCT have been published elsewhere [20,24].

Participants were recruited from March 2021 to March 2022 from various New York City–based health care facilities, including the Charles B Wang Community Health Center, private primary care providers, and NYU Langone Health and its affiliates. Recruitment strategies included (1) distributing study flyers, (2) direct referrals from health care providers, and (3) NYU Langone Health electronic medical records. Study staff members screened interested individuals by telephone to confirm eligibility.

Participants were eligible if they met the following criteria: (1) self-identified as Chinese immigrants or Chinese Americans; (2) were between the ages of 18 and 70 years; (3) were proficient in Mandarin, one of the most widely spoken Chinese languages; (4) had a diagnosis of T2D with a most recent HbA_1c level of 7% or higher at baseline; (5) used WeChat, a free social media app widely used among Chinese Americans; (6) owned a smartphone or tablet or agreed to use a study-provided device; and (7) were interested in receiving T2D-related videos via WeChat.

Exclusion criteria included significant uncorrected vision or hearing impairments, reluctance to comply with random assignments, pregnancy or plans to become pregnant during the study, breastfeeding, and residing in environments impeding T2D self-management, such as nursing homes.

The blinding of participants and community health workers who delivered the intervention was not feasible due to the nature of the intervention. However, the data management and statistical analyses were conducted by analysts who were blinded to the allocation.

This study is a pilot RCT primarily aimed at establishing the feasibility, acceptability, and potential efficacy of the CARE program on HbA_1c levels. The sample size of this pilot RCT was determined based on HbA_1c levels, which is the most commonly evaluated primary outcome when testing DSMES programs. According to Whitehead et al [30], in order to detect a small group difference of 0.2% in HbA_1c levels with 80% power and 5% type I error in a future full-scale RCT, we would need to recruit 50 participants in a pilot RCT. To account for a 20-25% attrition rate observed in previous studies with Chinese Americans [31-34], we aimed to enroll 60 participants, with 30 in each of the intervention and control groups.

We followed the intention-to-treat principle in data analysis. We used chi-square tests and independent 2-sample t tests to compare variables between the intervention and control groups. We applied piecewise linear mixed models to compare changes in outcomes over 0-6 months between the intervention and control groups. In these models, the 2 time periods (0-3 months and 3-6 months), the intervention group, and the intervention × time period interactions were modeled as fixed effects. We considered the participant as a random effect and included variables that showed statistically significant differences as covariates for adjustment. We applied the Tukey method to adjust P values for differences across time points within each group and the Bonferroni method to adjust P values for differences in changes over time between the 2 groups. We used R software (version 4.4.2; R Foundation for Statistical Computing) for statistical analyses and set the significance level at .05 for 2-tailed tests.

In addition to statistical significance, this study used the minimal clinically important difference (MCID) to assess the clinical significance of between-group differences. In this study, an MCID of 0.25 was considered clinically significant for the Diabetes Distress Scale [35]. As no established MCIDs are available for the Stanford 8-item Self-Efficacy for Diabetes Scale, the Summary of Diabetes Self-Care Activities, or the Beliefs in Diabetes Self-Care Activities, a threshold of 6% to 10% of the total score was used for these scales, as recommended by previous research [36]. Therefore, an MCID of 4.8-8.0 points on the Self-Efficacy for Diabetes Scale (total score: 80), 0.42-0.7 on the Summary of Diabetes Self-Care Activities (total score: 7), and 0.3-0.5 on the Beliefs in Diabetes Self-Care Activities (total score: 5) were used to evaluate the clinical significance of these outcomes. Clinical significance was determined by comparing the 95% CI of the observed differences to the MCID threshold for each scale: differences were considered “clinically significant” if the lower bound of the 95% CI exceeded the threshold, “may be significant” if the 95% CI included the threshold, and “not clinically significant” if the upper bound of the 95% CI was below the threshold [37].

This study was conducted in accordance with the principles of the Declaration of Helsinki. The study was reviewed and approved by the NYU Grossman School of Medicine Institutional Review Board (IRB# s18-00609). All participants took part in the study voluntarily. Informed consent was obtained from all participants prior to their enrollment in the study. Participants’ identifiable information was deidentified before data analysis to ensure their privacy and confidentiality. Participants received US $90 for completing the baseline, 3-month, and 6-month follow-ups. Participants from both groups were also compensated US $5 per week (up to US $60) to offset mobile data costs related to viewing intervention videos. The compensation was provided via a ClinCard.

A CONSORT flow diagram is presented in Figure 1. The mean age of participants was 54.3 (SD 11.5) years, with 62% (37/60) self-identifying as male, 78% (47/60) being married, 58% (35/60) being currently employed, 70% (42/60) having a high school education or lower, 75% (45/60) reporting an annual household income under $55,000, and 88% (53/60) indicating limited English proficiency. At baseline, the mean HbA_1c level was 8.2% (SD 1.2%) and the mean BMI was 26.5 (SD 5.1) kg/m² (Asian individuals with a BMI of 23 kg/m² or greater are considered to be overweight). There were no significant differences in these demographic characteristics between the intervention and control groups. See Table 1 for details.

Table 2 presents the changes in participants’ outcomes at baseline, 3 months, and 6 months. At baseline, the intervention group had a self-efficacy score of 53.3 (SD 14.2), while the control group scored 56.9 (SD 14.5). This difference was not statistically significant (P=.35).

Compared to the control group, participants in the intervention group demonstrated a statistically significantly greater increase in self-efficacy for diabetes management at 3 months (estimated difference in change: 8.47; 95% CI 2.44 to 14.5; adjusted P=.02). The observed between-group difference in self-efficacy at the 3-month follow-up may indicate a clinically meaningful improvement. However, at 6 months, the increase was not statistically or clinically significant (estimated difference in change: 5.87; 95% CI –0.23 to 11.98; adjusted P=.18).

At baseline, participants in the intervention group and control group did not differ in self-care activities (all P>.05). Specifically, participants in the intervention group followed their healthful eating plan for an average of 3.60 (SD 2.67) days per week, compared to 4.22 (SD 2.51) days per week in the control group, with no statistically significant difference (P=.36). For the specific diet, the intervention group followed it for 2.62 (SD 1.84) days per week, while the control group followed it for 2.77 (SD 1.63) days per week, which was not statistically significant (P=.74). Regarding exercise, the intervention group exercised for 2.53 (SD 2.45) days per week, compared to 3.23 (SD 2.45) days per week in the control group, with no significant difference (P=.27). Foot care was performed by the intervention group for 1.64 (SD 2.00) days per week, compared to 0.967 (SD 1.56) days per week in the control group, with no significant difference (P=.16). Finally, both groups adhered to their medication regimens similarly, with the intervention group adhering for 6.79 (SD 0.82) days per week and the control group for 6.61 (SD 0.99) days per week, with no statistically significant difference (P=.44).

Participants in the intervention group were more adherent to a healthy diet at 3 months (estimated difference in change: 1.61; 95% CI 0.46 to 2.75; adjusted P=.02) and 6 months (estimated difference in change: 1.64; 95% CI 0.48 to 2.81; adjusted P=.02) compared to the control group. The observed between-group differences in healthy diet at both the 3- and 6- month follow-ups may indicate a clinically meaningful improvement. There was no statistical or clinical significance in the between-group differences in other self-care activities at 3 months or 6 months, including exercise, blood glucose testing, foot care, and medication adherence (all P>.05).

At baseline, both the intervention group (mean 4.57, SD 0.58) and the control group (mean 4.54, SD 0.55) recognized the importance of self-management activities for their diabetes, including following a healthy diet, exercising, self-monitoring blood glucose, taking medication, and checking their feet. The difference was not statistically significant (P=.85). There were no statistically or clinically significant differences in beliefs about self-management at 3 months (adjusted P=.70) or 6 months (adjusted P>.99).

At baseline, both the intervention group (mean 2.09, SD 1.01) and the control group (mean 1.40, SD 0.43) reported minimal levels of diabetes distress. However, participants in the intervention group reported significantly higher diabetes distress compared to those in the control group (P=.002).

Compared to the control group, participants in the intervention group reported a significant improvement in diabetes distress at 6 months (estimated difference in change: –0.43; 95% CI: –0.71 to –0.15; adjusted P=.009). The observed between-group differences in diabetes distress at the 6-month follow-up may indicate a clinically meaningful improvement. No statistically or clinically significant differences in diabetes-related distress between the intervention and control groups were observed at 3 months (adjusted P=.23).

This study found that, compared to usual care, our 12-week CARE intervention, which included weekly diabetes education videos and biweekly support calls from community health workers, demonstrated statistically significant improvements in self-efficacy at 3 months, adherence to a healthy diet at both 3 and 6 months, and diabetes distress at 6 months. These between-group improvements may be clinically significant. These findings highlight the potential benefits of the CARE program, a culturally and linguistically tailored, social media–based intervention, in improving self-efficacy and diabetes self-care activities among Chinese Americans with T2D. The results suggest that culturally and linguistically tailored interventions may be a promising strategy for addressing the unique challenges faced by this population in managing T2D and improving diabetes-related health outcomes.

The intervention group demonstrated a significant increase in self-efficacy in managing diabetes compared to the control group, and this increase may have reached the threshold for clinical significance. Our findings contribute to the growing body of literature on the effectiveness of mobile health–based interventions aimed at diabetes education and self-management in enhancing self-efficacy, as evidenced in a recent systematic review [38]. Notably, none of the US-based interventions included in the review were specifically tailored for Chinese immigrants, an underserved population with a high burden of T2D and unique barriers to accessing these evidence-based programs [1-4,15-19]. By tailoring the intervention to the cultural and linguistic needs of the Chinese immigrants, the CARE program was able to engage participants in a more meaningful and effective way, which likely increased their confidence in adhering to self-care practices, a key determinant of change in other self-management behaviors [23].

The significant improvement in adherence to a healthy diet in the intervention group compared to the control group further supports the potential of the CARE intervention, particularly in addressing dietary changes, one of the most challenging aspects of T2D management for Chinese immigrants [15]. This improvement may represent a clinically significant change. This group typically follows a traditional Chinese diet [39,40], which is often not reflected in Western dietary guidelines commonly recommended in existing DSMES programs [16]. To address these challenges, the CARE program provided information on culturally appropriate food alternatives for Chinese Americans with T2D. It included video tutorials on reading food labels in local ethnic grocery stores. These strategies could facilitate the adoption of healthier dietary practices in this group. These findings underscore the importance of culturally tailored interventions in diabetes management for this immigrant population.

The intervention group demonstrated a significant improvement in managing diabetes distress compared to the control group. This improvement may be clinically significant. This finding aligns with previous interventional studies, which have reported that diabetes education interventions effectively reduce diabetes distress and offer emotional support to Chinese Americans [31,34]. The improvement in diabetes distress may be attributed to the multifaceted components of the CARE intervention, which integrates diabetes-related knowledge, behavioral techniques, and biweekly phone calls from community health workers to address SDOH concerns. These elements may collectively enhance participants’ self-efficacy in managing diabetes, subsequently alleviating their disease-related distress and other concerns throughout their diabetes journey.

We did not observe significant improvements in other aspects of self-care activities and beliefs in diabetes self-care activities. Several factors may have contributed to these null results. First, the small sample size likely resulted in insufficient statistical power to detect between-group differences in these outcomes. Secondly, the relatively short duration of our intervention (12 weeks) compared to some other DSMES programs (24 weeks) may not have been enough time to produce observable changes in these behavioral aspects of diabetes self-management. A larger-scale RCT with sufficient power is necessary to examine the effectiveness of the intervention across a broader range of diabetes-related outcomes. These findings will also inform the sample size for a future, fully powered RCT to assess CARE’s effectiveness and its potential to generalize these promising results to a broader population.

Several limitations should be noted. First, the sample size for this study was powered for HbA_1c levels; therefore, the self-efficacy and self-care activities reported in the current study may not have sufficient statistical power. The small sample size may limit the ability to detect significant differences in some outcomes. However, the primary goal of this pilot RCT was to assess the potential efficacy of the intervention and to generate pilot data for a future large-scale RCT. Second, all outcomes were self-reported and collected by community health workers over the phone and thus are subject to potential recall bias and social desirability bias. Additionally, our findings may not be generalizable to all Chinese Americans, particularly those with different sociodemographic characteristics. For instance, the majority of participants in our study had limited English proficiency, so the results may not apply to more acculturated counterparts who are fluent in English. Despite these limitations, this study is important as it serves as the first step in exploring scalable strategies to disseminate evidence-based diabetes interventions for underserved and minoritized populations. It also has the potential to serve as a care model for delivering culturally and linguistically tailored interventions to other immigrant or minoritized communities facing similar challenges in managing T2D and to lay the groundwork for a larger RCT.

This study provides promising evidence that culturally and linguistically tailored interventions may improve self-efficacy, dietary adherence, and diabetes distress among Chinese Americans. The use of accessible platforms like social media for health education shows promise in supporting diabetes management. Future research with adequate statistical power is needed to establish the effectiveness of our intervention. By addressing the cultural and linguistic needs of immigrant populations, tailored interventions have the potential to reduce health disparities and improve diabetes-related health outcomes for immigrant or minoritized communities with T2D.