Culturally Congruent Latino-Adapted Telemonitoring of Underrepresented Adults With Type 2 Diabetes: The CULTURA-DM2 Trial

This comparative effectiveness RCT enrolled 240 adult (aged ≥18 years) H/L patients with a diagnosis of type 2 diabetes. Patients were randomized to DTM, which used digital devices to share information between patients, caregivers, and clinicians, or COM, the most frequently used option for managing type 2 diabetes patients in the United States. Both comparators were consistent with recommendations in the ADA’s Standards of Medical Care in Diabetes—2018 (27).

Our sample size of 240 (120 per group) yielded 80% power to detect a 0.5% difference in the pre-/post- intervention change in A1C (using a two-tailed t test with α = 0.05) over 6 months. Our power calculation was based on the meta-analysis by Polisena et al. (12). We computed a sample size–weighted median of 12 SDs for A1C, yielding a value of 1.2%.

Given the SD of the pre-/post-intervention 6-month change in A1C of 1.2% in each group, a 0.45% difference in this 6-month change yields an effect size (ES) of 0.42. This ES is realistic, falling between two sets of reductions in the literature: the Diabetes Control and Complications Trial (in which a 1% decline in A1C reduced microvascular complications by 30%) (28) and RCTs/meta-analyses citing reductions of 0.3% as “clinically significant” (9–13). Falling between these limits, our reduction of 0.45% was clinically significant/meaningful because patients could see improvements in their A1C values.

Patients were followed for 6 months. To minimize resentful demoralization (the perception that the group receiving the new technology [tablet and Bluetooth- enabled peripherals] was getting the “better” treatment), we used the Zelen randomized consent form approach (29). Although a traditional RCT is the study design of choice for most interventions, the Zelen approach is particularly useful in studies of lifestyle interventions, which are often subject to factors such as patient preference and nonadherence (30–32).

Patients were enrolled in the study at nine endocrinology and primary care outpatient practices and an outpatient practice at a large safety-net hospital in the New York Metropolitan Area from September 2019 to August 2022 by bilingual, culturally congruent enrollment specialists. Before the COVID-19 pandemic, all patients were approached in person during visits to one of the designated clinic sites. After a 4-month pause in recruitment because of COVID-19 safety concerns, the study switched to telephone recruitment.

It was not feasible to blind the research nurse to patient participation in the DTM or COM group. To limit selection bias and ensure true random selection, our Biostatistics Unit used a random number generator to randomly select 400 patients from the pool of 4,092 eligible patients.

Eligible patients were approached (or called) for consent to have longitudinal data collected, including surveys at specified time points over 6 months of participation. Patients who were approached in person signed consent forms; those approached by telephone gave verbal consent.

Team members obtained randomization numbers from the Biostatistics Unit for patients to determine whether they were allocated to the DTM arm or COM arm. For patients randomized into the DTM arm, the recruiter approached them to obtain a second consent to receive DTM. Informed consent included a detailed description at a fourth-grade reading level, in the patient’s preferred language, of study, risks, and benefits, as well as images of the equipment to be used.

Both comparator cohorts received a phone call from enrollment specialists within 72 hours of enrollment to schedule baseline surveys and blood work, if needed. Enrollment was conducted in patients’ language of choice (English or Spanish). Patients were reimbursed $50 for completing surveys and blood work at baseline and 6 months.

Patients randomized to DTM, in addition to comprehensive outpatient visits, received a U.S. Food and Drug Administration–approved computerized monitoring device (Health Recovery Solutions), connecting them through a wireless tablet loaded with Health Recovery Solutions PatientConnect software, as well as eight diabetes educational videos developed by The Permanente Medical Group, Inc. Patients also received a Bluetooth-enabled glucose meter, blood pressure cuff, and weight scale and had weekly nursing telehealth visits with a culturally congruent nurse.

DTM included three main components: 1) daily self-monitoring of vital signs, 2) weekly telehealth visits with a nurse, and 3) eight educational videos, with the goal of building knowledge and diabetes self-efficacy. These videos were vetted by our Community Advisory Board to ensure representation of the language, foods, and culture of our patient population. All components were delivered in patients’ language of choice (English or Spanish).

Patients were trained by the enrollment specialists and nurses to use the Health Insurance Portability and Accountability Act–compliant tablet and Bluetooth- enabled peripherals to measure daily vital signs in English or Spanish. The nurse reinforced measurement of daily vital signs during weekly visits. If data (e.g., blood glucose and blood pressure) were significantly outside the normal range, the nurse could initiate a telehealth visit and/or either contact the patient’s practitioner to discuss or revise the treatment plan, including medication adjustment (e.g., insulin), or, for urgent matters, instruct the patient to call 911. An enrollment specialist delivered equipment to each patient’s home and trained the patient and/or caregiver to use the tablet and peripherals. If technological problems arose or if the patient needed further training that could not be accomplished over the phone, the enrollment specialist would return to the home for additional training.

DTM patients were scheduled to attend weekly video telehealth visits with the nurse, during which vital signs, symptoms of hypoglycemia and hyperglycemia, and behaviors (e.g., food choices) were discussed, as well as the diabetes management videos. The first telehealth visit was designed to facilitate a trusting relationship between the patient and provider, to discuss whether concepts should be reviewed in English or Spanish, and to reinforce the use of the tablet.

Patients were given unlimited access to the diabetes educational videos and asked to view one of the eight videos per week. The nurse reviewed each video with patients during weekly visits; key concepts were reinforced using the teach-back method (33). Each video was presented as one of three types: learn/aprenda, do/actúe, or succeed/triunfe (Supplementary Appendix S1).

COM, an evidence-based best practice based on the ADA’s 2018 Standards of Care (27), is the most frequently experienced option for patients with type 2 diabetes in the United States. COM includes a review of personal and family medical history, social history, medications, screenings, physical examination, and laboratory assessments (27). Patients are instructed to monitor their blood glucose levels within their physician’s recommendations and to have routine clinic visits every 3 months.

Patients randomized to COM, in addition to comprehensive outpatient visits, received a glucose meter (ReliOn Premier BLU Blood Glucose Monitoring System), which saves up to 450 test results in memory. To capture all fingerstick tests (beyond 450 values), the enrollment specialist was instructed to reset the monitor every 90 days to allow for continued monitoring and capturing of data. COM patients also received ReliOn Prime blood glucose test strips, a ReliOn lancing device, and ReliOn 33-gauge lancets. Supplies received by each patient were in accordance with physician recommendations regarding the frequency of blood glucose testing. COM patients also received monthly phone calls regarding their health care utilization.

Six months into enrollment, the New York Metropolitan Area became the epicenter of the COVID-19 epidemic, and H/L residents suffered from disproportionately high rates of hospitalization and death related to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection (34). The pandemic necessitated adapting enrollment processes to ensure patient and staff safety. New enrollments were paused for 4 months (from 9 March to 7 July 2020). The 43 patients who were enrolled before 9 March 2020 remained in the study, receiving contactless delivery of supplies. DTM patients continued to receive telehealth visits, and COM patients continued to receive monthly phone calls. Enrollment at the 10 sites resumed on 7 July 2020 but was conducted by phone to ensure safety. A substudy was published on the impact of the COVID-19 pandemic on subjects enrolled in this study (35). Unfortunately, after the pandemic began, adherence data from COM group participants’ glucose meters could not be collected because entering their homes was deemed unsafe.

This RCT was based on the intention-to-treat (ITT) principle that included all subjects randomized. Outcomes included A1C (the primary outcome), diabetes quality of life (measured with the Diabetes Quality of Life Brief Clinical Inventory) (36), blood pressure, weight, medication adherence (measured with the Adherence and Refill Medication Scale [ARMS-D]) (37), hypoglycemia episodes, diabetes self-efficacy (measured with the Stanford Diabetes Self-Efficacy Scale [DSES]) (38), emotional functioning (measured with the five-item validated Problem Areas in Diabetes [PAID-5] questionnaire) (39), inpatient utilization, unscheduled type 2 diabetes physician visits, and sick days. Data were analyzed using SAS, v. 9.4, statistical software (SAS Institute, Cary, NC).

Usability variables of adherence and engagement included 1) patient adherence to vital signs measurement, including blood pressure, blood glucose, and weight; 2) total number of telehealth visits completed (maximum 25); 3) patient and caregiver usability, measured with the Telemedicine Satisfaction and Usefulness Questionnaire (40); 4) caregiver app utilization (for patients with caregivers); 5) patient and caregiver engagement with technology; and 6) patient engagement in both groups (measured with the 10-item Patient Activation Measure [PAM] [41,42]).

Study data were collected and managed using the REDCap (Research Electronic Data Capture) tool (43). The data sets generated during and/or analyzed in the study are available from the corresponding author upon reasonable request.

The primary objective was to compare change in A1C from baseline to 6 months between groups. We performed an ANCOVA adjusting for caregiver status (yes or no). In keeping with the International Council for Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, Topic E9, statistical principles for clinical trials (chapter V, section 5.7) with a stratification variable, the interaction term for treatment × caregiver status was included in the primary analysis and as a main effect for treatment and caregiver status (44).

The secondary analysis of A1C over time from baseline to 3 months to 6 months was analyzed with a mixed-model, repeated-measures ANCOVA adjusting for caregiver status (yes or no). Other variables were analyzed using ANCOVA or, where appropriate, a Mann-Whitney U test, as detailed below.

This study conformed to all Consolidated Standards of Reporting Trials (CONSORT) guidelines for reporting. As seen in the CONSORT diagram (Supplementary Appendix S2), of 971 patients screened for eligibility, 65 were excluded; the remainder were approached for permission to collect data (n = 906), of which 662 declined. The remaining 244 were randomized to DTM or COM. Because we used the randomized consent design (Zelen) approach, we approached those randomized to DTM for permission to follow them using telehealth, and 120 consented, resulting in 120 randomized to DTM and 120 randomized to COM.

Study groups were equivalent on key baseline variables; 65.4% were female, and the mean age was 55.7 years (range 21–88 years). Most (62.3%) reported having a total household annual income <$25,000, and more than half (51.3%) reported being uncomfortable or very uncomfortable speaking English (Table 1). Most (74%) reported having a high school education or less.

For our primary outcome, change in A1C from baseline to 6 months, mean A1C values decreased from 8.32 to 7.55% in the DTM group and from 8.32 to 8.07% in the COM group. The adjusted mean A1C for DTM minus A1C for COM was −0.42 (95% CI −0.80 to −0.03).

Daily caregiver phone interaction for the DTM and COM groups was 82 and 81%, respectively, and live daily caregiver interaction was 92 and 84%, respectively. Neither caregiver status (P = 0.37) nor the between-treatment × caregiver interaction (P = 0.67) was significant. We therefore removed the interaction terms, resulting in a model that found a significant study group effect (P = 0.03), with DTM showing a significant improvement compared with COM.

Change in A1C over time (from baseline to 3 months to 6 months) was analyzed with mixed-model, repeated-measures ANCOVA. There was a significant group × time interaction (P <0.02) (a significant difference between groups for change in A1C over time). Specifically, within the DTM group, there was a significant difference with respect to A1C between baseline and month 3 (P <0.01) and between baseline and month 6 (P <0.01) but not between month 3 and month 6 (P = 1.00). Within the COM group, there were no significant differences in A1C between baseline and month 3 (P <0.59), between baseline and month 6 (P <0.25), or between month 34 and month 6 (P = 1.000) (Figure 1).

For blood pressure, a mixed-model, repeated-measures ANCOVA showed a decrease over 6 months in mean systolic blood pressure of 8.7 mmHg in the DTM group compared with an increase of 2.4 mmHg in the COM group (P = 0.01). Mean diastolic blood pressure decreased by 3.2 mmHg in the DTM group compared with an increase of 1.0 mmHg in the COM group (P = 0.03).

The percentage decrease of systolic blood pressure ≥10 mmHg (which ADA guidelines associate with a significant drop in mortality) was also significantly different between groups (DTM 43.0% vs. COM 16.3%, P <0.01) (45). Similarly, the percentage decrease of diastolic blood pressure ≥5 mmHg (also associated with a significant drop in mortality) was significantly different between groups (DTM 47.7% vs. COM 18.6%, P <0.01) (45) (Table 2).

Hypoglycemia episodes (defined as blood glucose <70 mg/dL, measured by fingerstick or reported symptoms) were analyzed using a Mann-Whitney nonparametric U test. Medians were 3 and 2 hypoglycemic events, for the DTM and COM groups, respectively, yielding a P value of 0.1796, indicating no significant between-group difference.

On ARMS-D scores, there were no significant differences found in 6-month mean difference between groups (DTM −0.15 vs. COM −0.26, P = 0.53) or between 6-month scores (DTM 14.1 vs. COM 14.8, P = 0.18).

On the DSES, scores in the DTM group increased from 7.3 at baseline to 8.2 at 6 months (P <0.01), and scores in the COM group increased from 7.6 at baseline to 8.0 at 6 months (P = 0.19). However, self-efficacy was not significantly different between groups; change in DSES from baseline to 6 months was 0.83 in the DTM group versus 0.28 in the COM group (P = 0.07) (Table 3).

Between-group differences in DQOL scores were not significant (P = 0.65); scores decreased (i.e., quality of life improved) over the 6-month period for the DTM and COM groups by 3.1 and 2.5, respectively.

For changes in weight from baseline to 6 months, the DTM group lost a mean 2.4 lb, whereas the COM group gained a mean 0.1 lb (P = 0.13).

For emotional functioning, analysis of scores on the PAID-5 scale (39) found no significant differences between groups between baseline and 6 months. Both group’s scores improved almost identically (DTM −2.2 and COM −2.1, P = 0.75).

There were 73 inpatient utilization records of encounters from 50 unique patients, of which 35 records were emergency department (ED) visits and 38 were hospitalizations. The mean number of ED visits was 0.17 (median 0, range 0–3). The mean number of hospitalizations was 0.13 (median 0, range 0–4).

In the DTM group, 23 patients were hospitalized for a total of 109 days (mean 4.74 days); in the COM group, 26 patients were hospitalized for a total of 107 days (mean 4.12 days). The DTM group had 20 hospitalizations, whereas the COM group had 18. There was no significant difference between groups with respect to the proportion of patients who had at least one hospital visit (DTM 10.83% vs. COM 10.92%, P <0.9820).

The DTM group experienced 20 ED visits, whereas the COM group had 15 ED visits. There was no significant difference between groups for the proportion of patients who had at least one ED visit (DTM 11.67% vs. COM 10.92%, P <0.8562). There were also no statistically significant differences for cumulative lengths of stay as determined by the Mann-Whitney U test.

Unscheduled sick visits and sick days were assessed monthly through self-report and analyzed using χ² tests comparing group rates. The proportion of patients who reported at least one sick day was significantly different between the DTM and COM groups (22 [18.3%] and 6 [5%], respectively; P = 0.0013).

For blood glucose monitoring (BGM), the majority of participants in the DTM group (64.2%) were highly adherent (performing 91–180 measurements over 6 months); 6.7% were moderately adherent (46–90 measurements), 5.8% were minimally adherent (1–45 measurements), and 23.3% were nonadherent (0 measurements). For blood pressure monitoring, the majority of participants (62.5%) were highly adherent, with moderate, low, and nonadherence rates of 8.3, 5.8, and 23.3%, respectively. Most patients (51.7%) were highly adherent in weight monitoring. The proportions of patients with moderate, low, and nonadherence rates were 13.3, 11.7, and 23.3%, respectively.

The proportion of weekly scheduled telehealth visits that were kept was 83% (20.7 of 25). The average length of a visit was 38.1 minutes (range 1.6–113.9 minutes). The mean number of minutes spent per patient watching the eight educational videos was 1:46 minutes/week. Videos were relatively short in length, ranging from 1:17 to 3:00 minutes each.

The mean Telemedicine Satisfaction and Usefulness Questionnaire score (>56 is considered a good experience score for patients) for the 65 respondents was 63.1 ± 5.95, and the median score was 64 (range 48–70). Overall, 75.4% of respondents had a score >56. Baseline PAM scores were 65.3 for the DTM group and 65.2 for the COM group (P = 0.74). At 6 months, PAM scores decreased to 60.2 and 62.4, respectively (P = 0.35).

Participants in the DTM group achieved significant reductions in A1C and improvements in self-efficacy and systolic and diastolic blood pressure over the 6-month study period compared with those in the COM group. This is the first study to exclusively enroll underserved H/L patients in an RCT to receive either DTM or COM.

These findings are consistent with those of the IDEATel study, although over a much shorter time frame than in the IDEATel study, in which patients were managed over a 5-year period (20,21). Both studies included all patients diagnosed with type 2 diabetes, without regard to their degree of blood glucose control. The IDEATel study enrolled underserved patients identifying as White, Black, or Latino, and a prerequisite for study enrollment was insurance coverage (i.e., current Medicare beneficiaries); however, the current study exclusively enrolled H/L patients, many of whom were uninsured.

Access issues this population typically faces may have been at least partially addressed by the nature of the DTM intervention. Even during the height of the COVID-19 pandemic, which severely affected underserved H/L people in New York, patients randomized to DTM were able to safely meet with the study nurse to address their self-management of diabetes.

For our primary outcome, A1C, patients receiving DTM experienced significant glycemic improvement after 3 months of intervention and sustained that improvement until the end of the 6-month study. This improvement may have been the result of high rates of adherence, the development of diabetes-related self-efficacy, and the rigorous adaptation process to ensure the feasibility and acceptability of the intervention during the first year of the study.

To identify patient-reported factors that likely had an impact on patient adherence and ultimately improvement in self-efficacy, A1C, and blood pressure, we conducted semi-structured, 1:1 interviews with 42 patients after they completed the 6-month DTM intervention. These interviews were conducted by phone in each patient’s language of choice (Spanish or English). A qualitative analysis of participant feedback identified the factors patients reported contributed to their success. With the guidance of the culturally congruent nurse, patients reported “feeling supported” and becoming mindful of their own role in taking care of their health and diabetes. As one patient reported:

Bandura (46) defined self-efficacy as “people’s belief in their ability to control their functioning and events that affect their lives,” emphasizing that self-efficacy can facilitate motivation, well-being, and personal accomplishment. He further posited that people’s beliefs in their efficacy are influenced by four primary sources, including mastery experiences, vicarious experiences, social persuasion, and emotional states. The DTM intervention incorporated all of these sources of influence; the culturally congruent educational videos portrayed people who looked like the participants succeeding in managing their diabetes. Social persuasion was provided by the culturally congruent nurse, who provided positive reinforcement to patients every week. As one patient reported:

Patients in the DTM group achieved high rates of adherence for blood glucose, weight, and blood pressure monitoring. This finding may have been the result of the education provided by culturally congruent recruitment specialists and nurses in patients’ homes, in their preferred language, and scheduled at convenient times, including nights and weekends.

Other studies affirm that people with higher diabetes self-efficacy experience improvements in glycemic control, medication adherence, and quality of life (47–49). Telemedicine has also emerged as a way to use technology to help improve patient self-efficacy (50). This study confirms that underserved H/L people with diabetes not only are receptive to using telemedicine, but also experience improved glycemic control and better adherence to management plans when telemedicine is used (51–54). For our DTM-specific secondary clinical outcome measures, systolic and diastolic blood pressure, improvements may also have been the result of a high level of adherence to vital sign monitoring. Individuals in the COM group experienced an increase in both systolic and diastolic blood pressure. Observed differences in blood pressure are consistent with significant decreases in mortality found in predictions of major adverse cardiac events (45).

The 2017 systematic review by McCurley et al. (19) of culturally tailored interventions to better reach the high-risk Hispanic population found that interventions with the largest effect sizes tended to incorporate literacy modification, Hispanic foods/recipes, cultural diabetes beliefs, family/friend participation, structured community input, and innovative experiential learning. All of these factors were incorporated into the DTM intervention. Similarly, the teach-back education method was used to ensure that patients understood concepts presented in the culturally congruent educational videos.

Our use of the adaptation framework of Wingood and DiClemente (55) to systematically adapt the home telemonitoring intervention for people with type 2 diabetes in underserved H/L communities ensured a comprehensive approach that addressed many aspects and perspectives that might otherwise have been missed when caring for this population. For example, offering visits outside of the typical 9–5 workday (many patients reported working two or three jobs and therefore being unable to participate in visits during usual work hours), discussing culturally relevant food selections (i.e., American vs. Caribbean or South American eating patterns), and using familiar terminology (e.g., “azucar” [sugar] rather than “glucosa” [glucose] when referring to blood glucose) were key factors that likely made a difference, as was the provision of educational videos that presented people who looked like the intervention’s participants successfully managing their diabetes.

This study faced a number of limitations. Because of the COVID-19 pandemic, we were unable to safely visit the homes of COM participants to obtain blood glucose monitoring adherence data from their glucose meters. Therefore, we have no way of knowing whether these free supplies affected adherence rates. This limitation may explain why we saw minor (nonsignificant) improvements in quality of life and A1C in the COM group. Missing data were an issue throughout the study in both groups. Finally, it is difficult to ascertain the impact of the COVID-19 pandemic on other outcomes. Although these results are highly encouraging, it is important to replicate the study with additional underserved patient panels.

This adapted DTM intervention achieved clinically and statistically significant reductions in A1C, diabetes self-efficacy, and blood pressure over the 6-month study period compared with COM. The process of adaptation, along with consistent culturally congruent support, are likely to have been crucial components of the intervention’s success. These highly encouraging results must be replicated in larger samples of similar populations.

The authors acknowledge Jill Cotroneo of Northwell Health for administrative support and the individuals who generously volunteered their time to participate in this study.

This project was funded by the Patient-Centered Outcomes Research Institute (AD-2017C3-9185).

No potential conflicts of interest relevant to this article were reported.

All authors reviewed and edited the manuscript. R.P. wrote the manuscript and was responsible for study conceptualization, funding acquisition, methodology, project administration, implementation, and supervision. S.M. contributed to study conceptualization, methodology, and implementation. V.C.G., J.M., and N.G. contributed to implementation, patient consent, and data management. M.S.W. coordinated with the institutional review board and contributed to implementation, methodology, and data management. E.C. contributed to implementation, patient consent, data management, and data analysis. C.N.N. wrote the manuscript and contributed to data management and data analysis. V.H.P. contributed to implementation, methodology, and data analysis. A.K.M., D.G., L.F.M., J.G., A.N.M., S.I.M., R.Z., M.P., J.P., R.J.D., A.B.-Y., C.E., M.M., and W.G. contributed to the methodology. P.B. contributed to implementation. C.S. and M.K. contributed to the data analysis. M.L.L. contributed to methodology and data analysis. L.B. contributed to implementation and methodology. N.Z. contributed to implementation and patient consent. Y.T.H. wrote the manuscript and contributed to conceptualization and methodology. C.N.N. is the guarantor of this work and, as such, had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.