We conducted a qualitative study by carrying out focus groups at the Massachusetts General Hospital (MGH) Chelsea HealthCare Center, one of the MGH Community Health Care Centers that serve a patient population in the greater Boston area that is majority Latino or Spanish-speaking [28]. Eligible participants had a T2D diagnosis (based on diagnostic codes, confirmed with electronic health record review), were ≥18 years old, were taking at least 1 glucose-lowering medication, and reported Spanish as their preferred written and spoken language. We excluded patients with auditory or communication difficulties.

We searched the MGH Research Patient Data Registry to identify individuals who received care at MGH within the previous 2 years and who were ≥18 years old, had a diagnosis of T2D, and had registered Spanish as their preferred language. A total of 3709 patients met this criteria, of whom 500 lived in the city of Chelsea. We mailed out opt-out letters to 350 of these participants (selected randomly), followed by telephone calls to screen for the eligibility criteria described above. Among 30 eligible patients willing to participate in the study, 22 unique participants attended one of the focus groups 1‐5. A total of 10 participants who attended focus groups 1‐3 also attended either focus group 4 or 5. The reason for participation in multiple focus groups was to engage patients in the iterative process of content adaptation, which took place in two steps: (1) identification of barriers to diabetes medication adherence of relevance to the study population (focus groups 1‐3) and (2) adaptation of SMS text message content addressing the relevant barriers. Enrollment occurred between March and November of 2023. Given that the original REACH content was created with input from communication experts who edited the text messages to be readable and understandable (ie, written at the sixth-grade reading level or below) [26], the adaptation of REACH barrier content was also guided by an emphasis on reading appropriateness for the study population. Thus, for eligible patients interested in participating in the study, we conducted a health literacy screening to tailor the focus group activities to the literacy level of the study population. We used the 1-item Brief Health Literacy Screening Tool (BRIEF) questionnaire (“How confident are you filling out medical forms by yourself?”), which has been validated for Spanish-speaking populations as a quick assessment of health literacy [29].

Participants received US $50 for focus group attendance. The Mass General Brigham Institutional Review Board approved all aspects of the study (protocol 2022P002678). At the beginning of each focus group, participants received an IRB-approved information sheet, which outlined the objectives of the study, procedures, benefits and risks, confidentiality, as well as an emphasis on voluntary participation. All participants agreed to participate in the focus groups.

At the start of each focus group, participants were asked to fill out a sociodemographic and diabetes history questionnaire. The data that were collected through this questionnaire were age, sex, ethnicity, country of birth, preferred language, educational attainment, household income, health insurance, and diabetes history (age at diagnosis, diabetes medications, and complications).

The development of the REACH barrier content has been described in studies by Nelson et al [25,26]. The REACH study team conducted a thorough review of published studies describing patient-reported barriers to adherence to glucose-lowering medications. The literature review yielded 68 barriers to taking oral glucose-lowering medications and 7 barriers to taking insulin; similar barriers were collapsed, resulting in 36 adherence barriers [25]. A 2-step interactive procedure was subsequently used to rank these barriers among adults with T2D. Each of the 36 barriers was mapped to the IMB constructs, and content experts drew on identified studies to develop SMS text messages addressing each barrier [26]. For instance, the barrier “My daily medicine routine is too complicated to keep track of” generated 21 SMS text messages to address this concern, such as “to help keep things simple, take your medications at the same time you’re doing something else, like brushing your teeth or eating meals.”

We used descriptive statistics (means, SDs, and proportions) to summarize participant characteristics. For the qualitative analysis, an “open-coding” approach (inductive) was used first to identify themes in the transcripts (in Spanish) from focus groups 1‐3. Themes were then categorized as barriers and facilitators to diabetes medication adherence. Second, we used deductive coding to map the codes into the IMB construct. Coding was conducted by native Spanish-speaking research team members. Analyses were conducted using Microsoft Excel and NVivo 14 (Lumivero).

A total of 22 participants attended focus groups 1‐5. The first 3 groups were attended by 18 participants; 10 of these participants also attended either focus group 4 or 5. Table 1 summarizes the demographic and clinical characteristics of the study participants. The mean age was 63.2 (SD 11) years, 55% (10/22) were female, 59% (13/22) were categorized as having high health literacy (based on the 1-item BRIEF questionnaire), although 55% (12/22) reported an educational attainment of elementary or middle school only. Over two-thirds (16/22 73%) of participants were born outside of the United States, largely in Central America. Participants had a mean (SD) diabetes duration of 20.3 (12) years. In addition, 59% (13/22) of participants reported insulin use.

Multimedia Appendix 2 shows the list of 40 barriers to diabetes medication adherence that were assessed for relevance in focus groups 1‐3. Barriers are shown in order of higher to lower proportion of participants reporting each barrier as relevant sometimes or always. Figure 2 shows the list of 40 barriers, categorized into the IMB model and ranked according to the proportion of participants who reported the barrier as relevant sometimes or always. Among the top 10 barriers, 50% (n=5) corresponded to the information category, 20% (n=2) to the social motivation category, 20% (n=2) to behavioral skills, and 10% (n=1) to personal motivation.

Table 2 outlines the key themes that emerged in focus groups 1‐3 based on participant quotes, grouped into barriers to diabetes medication adherence, and mapped into the IMB model. In total, 5 themes emerged in response to the question “What are some difficulties that come to mind when taking your diabetes medications?” In order of relevance (higher or lower number of times mentioned), most participants agreed that medication burden was one of the most important barriers to diabetes medication adherence, followed by fear of side effects or of new medication, difficulty or confusion with medication changes, forgetting to take medications, and frustration with diagnosis or lack or improvement.

Multimedia Appendix 6 shows the 5 themes that emerged in response to the question, “What motivates you to take your diabetes medications.” In order of relevance (higher to lower), fear as motivation was an important facilitator, followed by belief in treatment benefit, personal responsibility, and will to take medications, family support and medication organization, and getting used to the medication.

Focus groups 4‐5, which centered on eliciting feedback on the translated text message language content, generated rich discussion. A total of 111 text messages (related to barriers 1‐8) were reviewed in focus group 4 and 100 text messages (related to barriers 9‐18) were reviewed in focus group 5. Participants were generally enthusiastic about the text message content and provided feedback on wording that seemed confusing, too complex, or less likely to be understood by broad Spanish-speaking audiences. For instance, the word “comprimidos” was changed to “pastillas” and the preferred word for refills was “reposiciones” instead of “prescripciones.” In total, 12 messages were excluded since patients did not like the tone, felt that they were difficult to understand, or felt that they were not relevant. Some examples of these excluded text messages included “If taking your diabetes medications is unpleasant, try to push through. Remember that not taking them may make you feel worse” and “What advice would you give your friends who forget to refill their medications? Use those tips for yourself too!”

In the final step of REACH-Español content adaptation, 3 patient participants who had attended one of the focus groups were invited to participate in the beta testing process over the course of 4 weeks. At the end of beta testing, the following patient stakeholder input was provided:

In this study, we outline the multistage process used to adapt the REACH text message content [26], which addresses patient-reported barriers to diabetes medication adherence, to a Latino population. The adaptation process centered on the theory-based IMB skills model for medication adherence and engagement of patient stakeholders to identify barriers to diabetes medication adherence with specific relevance to Latino adults with T2D (with Spanish-speaking preference). We found that among the top 10 most commonly reported barriers to diabetes medication adherence among Latino adults with diabetes, 50% (n=5) corresponded to information gaps, 20% (n=2) to social motivation, 20% (n=2) to behavioral skills, and 10% (n=1) to personal motivation. In contrast, our thematic analysis showed that personal motivation barriers (medication burden and fear of side effects) were the highest concern for focus group participants, followed by behavioral skills (forgetting to take medications), and information barriers (difficulty or confusion with medication changes and frustration with diagnosis or lack of improvement). These findings suggest that using the IMB model as a framework to identify barriers to diabetes medication adherence is useful in this population and that different qualitative approaches (open discussion vs ranking exercise) can provide complementary data to inform the adaptation process of tailored interventions.

By engaging focus group participants in an interactive barrier ranking exercise conducted through card sorting, we were also able to select highly relevant barriers to diabetes medication adherence and exclude those with lower salience from the study. Through this exercise, we found that the priority of some barriers differed in this population compared with the original REACH intervention. For instance, “I’m not sure what my diabetes medicine is supposed to do” was reported as relevant by 61% (11/18) of participants in this study, whereas this barrier was relevant only for 16% (38/237) of participants in the REACH development study [25]. Similarly, “I am not sure why my doctor sometimes changes my dose or type of medicine” was relevant to 50% (9/18) of participants in this study, whereas only 15% (36/237) of participants in the REACH development study reported that this barrier was relevant. These findings underscore the importance of stakeholder-driven adaptation approaches given that core intervention components (in this case, the list of barriers), differed across populations.

We also found that the iterative process of translating and modifying the original intervention content with stakeholder input (bilingual research team, focus group participants) enriched the adaptation process by incorporating language diversity and nuance. By eliciting input from participants through interactive focus group activities, we were able to capture both the thematic and linguistic or cultural relevance of the content being translated. For instance, we excluded certain text messages that did not resonate with the audience. We revised the language with modifications considered more broadly understood across Spanish-speaking Latin American and Caribbean populations. Choosing language wording with a broader reach was particularly important in this study given the heterogeneity of the Spanish language across Latin America and the Caribbean. Our observations are consistent with previous studies that have highlighted the importance of engaging bilingual stakeholders (including patients) throughout the language adaptation process [30,32,33].

Although barriers to diabetes medication adherence have been previously reported in the Latino population [8,10,11,17,19,34], most studies have not used a theoretical framework to identify such barriers. The IMB model has been empirically validated as a framework to identify barriers to diabetes medication adherence [23], but only 1 intervention based on the IMB model has documented its use in Latin American adults with diabetes [35]. Our study supports the use of the IMB framework to better ascertain barriers to diabetes medication adherence among Latino adults, which can be helpful in the development of interventions to address these barriers. Our qualitative analysis also highlights the importance of treatment complexity in this patient population and underscores the need for interventions that can identify and reduce treatment burden and regimen complexity. Although these challenges to diabetes self-management have been well described [36-39], scalable interventions that can reduce the treatment burden are lacking, particularly in the Latino population.

Digital health technology can improve diabetes medication adherence behavior [21,40] and HbA1c [21,40,41] and is recommended by the American Diabetes Association to support diabetes self-care [42]. However, among the 90 (approximately) digital health apps currently available for diabetes self-management, only one-third are available in Spanish, of which 94% (90/96) have a readability level above that recommended for patient education material [43]. Some digital medication adherence support platforms are currently under study in Latino and Spanish-speaking populations [44,45]. However, none of these platforms provide educational content tailored to patient-reported barriers to diabetes medication adherence. The REACH-Español platform adapted in this study offers this key innovation, in addition to health literacy–informed content, which could facilitate broader engagement of prospective participants and long-term medication adherence.

Our study has the key strength of having adapted an intervention with iterative input from stakeholders who could be representative of the population most likely to use the intervention. This user-centered approach is recommended in diabetes digital health research [46,47] to ensure that end user perspectives are represented in the development of interventions and content is tailored to user needs. However, our study also has several limitations. First, the sample size was relatively small, although we achieved thematic saturation for barrier selection after conducting 3 focus groups. Second, we recruited participants from a single health center and did not have representation from every Latin American and Caribbean country, which could limit the generalizability of our findings. In addition, although the barriers to diabetes medication adherence selected in this study were based on relevance to the focus group participants, the order of relevance may differ in other Latin American populations living elsewhere in the United States or with a different insulin usage profile. For instance, “I have trouble paying for my medicine” was not a highly ranked barrier but this may be an important barrier in other areas of the country with lower insurance coverage. Thus, should the REACH-Español content be adapted to other contexts, using the IMB model to identify population-specific barriers could be helpful. Finally, the focus of this study was to adapt the barrier-specific text message content. While text message content pertaining to diet and physical activity was directly translated from English to Spanish, the content did not undergo the stakeholder-driven adaptation process described here, which will be the subject of a future study.

In this study, we report the multiphase, stakeholder-driven adaptation of the REACH digital health platform text message content, which addresses barriers to diabetes medication adherence, to a Latino population with T2D. We found key barriers to diabetes medication adherence that were relevant to this population, largely related to information gaps. These findings can contribute to the fields of patient-important outcomes and treatment burden to guide respectful and meaningful patient engagement in the management of T2D. Furthermore, engaging a diverse group of bilingual stakeholders (bilingual research team, focus group participants) early and throughout the iterative adaptation process was integral to ensuring content and linguistic relevance of the text messages in the REACH-Español platform. The feasibility, acceptability, usability, and preliminary efficacy of REACH-Español is under study in a 6-month pilot randomized controlled trial, which could broaden our understanding of the role of tailored, digital health interventions on improving diabetes self-care among Latino adults with T2D.