Efficacy of a glucose meter connected to a mobile app on glycemic control and adherence to self-care tasks in patients with T1DM and LADA: a parallel-group, open-label, clinical treatment trial

ABSTRACT Objective: The main aim of the study was to evaluate the patients’ glycemic control and adherence to self-care tasks. Materials and methods: Patients with type 1 diabetes mellitus (T1DM) or latent autoimmune diabetes of the adult (LADA) using a multiple daily injection (MDI) regimen with carbohydrate counting (n = 25, Subgroup B) or fixed insulin dose (n = 25, Subgroup C) were allocated to use the application (app) for 12 weeks. Both subgroups were compared with each other and against a control group (n = 25, Group A) comprising patients with T1DM or LADA treated with continuous subcutaneous insulin infusion (CSII) in a parallel-group, open-label, clinical treatment trial. All patients had glycated hemoglobin (A1C) levels measured and were asked to fill out the Diabetes Self-Management Profile (DSMP) questionnaire at study start and end. The patients were instructed to measure capillary glucose six times daily in study weeks 4, 8, and 12. Results: Mean A1C levels decreased 0.725% in Subgroup C in intragroup analysis (p = 0.0063), and had a mean variation of 0.834% compared with Group A (p = 0.003). Mean DSMP scores increased 5.77 points in Subgroup B in intragroup analysis (p = 0.0004) and increased by a mean of 6.815 points in relation to Group A (p = 0.002). Conclusion: OneTouch Reveal improved both A1C levels and DSMP scores in patients with T1DM or LADA compared with standard treatment (CSII).


INTRODUCTION
A 2019 projection from the International Diabetes Federation (IDF) ranked Brazil third among all countries in highest number of cases of type 1 diabetes mellitus (T1DM), with an estimated prevalence of 51,500 cases among children and adolescents (1). Concerningly, only 13.2% of all patients with T1DM have glycated hemoglobin (A1C) levels within target goals (2). Complications of T1DM can be prevented by strict patient adherence to self-care tasks and achievement of intensive glycemic control (3).
Adherence can be defined as the extent to which a person's behavior corresponds with the agreed recommendations from a health care provider (4). Adherence to treatment of chronic diseases is lower than 50% (5)(6)(7)(8) and requires three additional medical visits per patient per year, increasing the average annual cost of individual treatment by $2,000 (6).
For optimal adherence in diabetes, the patients must follow difficult treatment regimens. For example, patients with T1DM and some of those with latent autoimmune diabetes of the adult (LADA) must be treated with continuous subcutaneous insulin infusion (CSII) or with multiple daily injection (MDI) regimens of fixed insulin doses or insulin doses conditioned to carbohydrate counting (9). The Diabetes Self-Management Profile (DSMP) questionnaire is a self-reported measure of adherence (10) (Appendix 1) that has been translated into Brazilian Portuguese and validated for the Brazilian population by Teló and cols. (2). Questions in the DSMP contemplate different aspects of diabetes selfmanagement, covering five essential self-care domains, namely, exercise, hypoglycemia, diet, blood glucose monitoring, and insulin (11,12).
Regular physical activity has been associated with a reduced risk of cardiovascular disease in patients with diabetes, decreasing all-cause mortality (13,14). However, insulin therapy in these patients must be managed with special care to reduce the risk of exerciserelated hypoglycemia.
Regarding MDI regimens, the use of modern ultralong-acting insulin analogues along with carbohydrate counting help reduce the risk of hypoglycemia; patients on CSII therapy also have important tools to prevent this complication (9,15).
In terms of diet, carbohydrate counting is currently considered the gold standard for estimation of mealtime insulin dose in patients with T1DM or LADA (9). Modern CSII has integrated bolus calculators that automatically determine how much insulin should be administered based on food intake (9).
The Brazilian Diabetes Society (SBD) recommends 4-8 capillary self-monitoring blood glucose (SMBG) measurements daily for patients with T1DM or LADA (9). One of the various tools currently available for SMBG is the OneTouch Reveal (LifeScan, Wayne, PA, USA) mobile phone application (app). This app connects with the OneTouch Select Plus Flex (LifeScan) meter allowing patients to enter insulin doses, carbohydrate intake, and physical activity data, automatically transferring capillary glucose readings from the meter to the app. Synchronization between the app and the meter further allows patients to share the data with their physicians.
The Diabetes Association of Ourinhos (ADO) is a nongovernmental organization affiliated with the Juvenile Diabetes Association. All patients treated at ADO receive care covered by the Brazilian Unified Health System (SUS) and following the guidelines recommended by SBD (9). Carbohydrate counting is recommended to all patients with T1DM or LADA receiving care at ADO, but the effectiveness of this approach depends on the patient's commitment to the treatment. Those who fail to cope with the demands of carbohydrate counting are placed on an MDI regimen with fixed insulin dose. Obtaining a CSII device through SUS requires time; therefore, many patients at ADO follow carbohydrate counting while waiting for the device.
Patients with T1DM and LADA face significant challenges to meet the considerable demands of selfcare management. Based on these considerations, this study focused on the glycemic control and adherence to self-care tasks of patients with T1DM or LADA treated with an MDI regimen using the OneTouch Reveal self-monitoring application versus standard treatment (CSII, control group).

MATERIALS AND METHODS
A total of 75 patients receiving care at ADO were allocated in a parallel-group, open-label, clinical treatment trial with a control group. The intervention lasted 12 weeks.

Eligibility criteria
The eligibility criteria included a diagnosis of T1DM or LADA for at least 1 year and exclusive use of insulin analogues. Patients on MDI with carbohydrate counting and those on CSII therapy were required to be on these treatment regimens for at least 3 months.

Exclusion criteria
The exclusion criteria for the study were pregnancy, current use of a mobile phone app other than OneTouch Reveal, visual impairment preventing proper use of the

Sample size definition
A review of the medical data of patients registered at ADO in 2018 identified 127 patients meeting the eligibility criteria. Among these, 32 were receiving CSII therapy, while the remaining were on an MDI regimen (37 counting carbohydrates and 58 on a fixed insulin dose regimen).
During the recruitment phase in January 2019, all 32 patients on CSII were invited to participate in the study, and the 25 patients who agreed comprised the control group. After selection of these patients, we recruited, following alphabetical order of patients who agreed to participate, the first 25 patients on MDI with carbohydrate counting and the first 25 patients on MDI with fixed insulin doses; these 50 patients comprised the intervention group ( Figure 1).
The therapeutic regimen and type of CSII or insulin analogues used by the subjects were determined during routine follow-up appointments. Treatment regimens in place at the time of enrollment (CSII, MDI with carbohydrate counting, or MDI with fixed insulin doses) were established prior to patient recruitment during routine follow-up. A free and informed consent form was signed by all participants older than 18 years. Legal guardians signed the consent forms for patients younger than 18 years, while these patients signed assent forms with ageappropriate language.

Ethical aspects
All expenses related to blood drawing and A1C tests performed at the study start and end, along with costs of meals and transportation for the participants at Preintervention evaluation Table 1 shows the baseline characteristics of the study participants. Preintervention assessments included measurement of A1C levels (by high-performance liquid chromatography) and evaluation with the DSMP questionnaire.

Patients' allocation
Following the baseline evaluation, the patients were divided into the following groups ( Figure 1): • Group A: CSII therapy (no intervention, control group), n = 25. • Subgroup B: MDI regimen with carbohydrate counting, n = 25. • Subgroup C: MDI regimen with fixed insulin doses, n = 25. Patients with macrovascular complications (in %) None Abbreviations: anti-GAD, anti-glutamic acid decarboxylase antibody; U 100 Glargine = 100 units of glargine insulin per mL of solution. Notes: For patients younger than 18 years, guardians with the highest level of education were considered. The same patient could present more than one concomitant microvascular complication. Subjects could have no positive autoantibody or more than one positive autoantibody. The analysis excluded patients without investigation of other autoimmune diseases at recruitment (for example, those with positive antibodies but no histological confirmation of celiac disease or atrophic gastritis). The same patient could present two or more concomitant autoimmune conditions.
Adherence in diabetes and a mobile app The participants in Subgroups B and C were instructed to use the OneTouch Select Plus Flex meter and the OneTouch Reveal app for 12 weeks. The main researcher delivered the meters and instructed all 50 patients on MDI regimens on how to operate both meter and app. The first 3 weeks of the research were reserved for patients' acquaintance with meter and app (run-in period).
Data from the devices were downloaded on study weeks 4, 8, and 12, and the patients were instructed to perform six capillary glucose measurements daily each week prior to these three time points. Each of these weeks was labeled a "focus week".
In addition to the visits for data download, the participants maintained medical appointments with their usual endocrinologists at ADO throughout the study, and only data collected in the second focus week were evaluated in these appointments. Changes in insulin regimen implemented during these appointments followed SBD guidelines (9).
For ethical reasons, continuous glucose monitoring (CGM) data from Group A were only used for insulin adjustment and were not considered in the present analysis.

Postintervention evaluation
Only data downloaded at focus weeks 4 and 12 were analyzed ( Table 2). Blood drawing for A1C measurement and evaluation with the DSMP questionnaire were again performed at the end of week 12.
Since the DSMP results can be significantly biased by patients' wish to please the research team (1,16), the questionnaires on both occasions were administered by staff members not assisting the patients.

Primary outcome
The primary outcome was a statistically significant (p < 0.05) reduction in A1C levels between the initial and final evaluations, considering patients in Group B + C in intragroup analyses and compared with Group A.

Secondary outcome
The secondary outcome was a statistically significant increase in DSMP scores between the initial and final evaluations considering patients in Group B + C in intragroup analyses and compared with Group A.
Normality was tested with the Anderson-Darling, Kolmogorov-Smirnov, Shapiro-Wilk, and Ryan-Joiner tests. Differences between paired samples with normal distribution were analyzed with paired Student's t test. The z-proportion test compared proportions of success between the groups. Values were compared across different groups using independent samples t test. When the normality assumption was violated, we applied the Wilcoxon-Mann-Whitney test for independent samples, proportion test, or Wilcoxon test for paired data.
The independent samples t test, Wilcoxon rank sum test, and proportion test were used to analyze correlations between (1) Group B + C versus Group A,

RESULTS
Two patients in Subgroup C withdrew participation during the study. Eight other patients were excluded, including five for whom insulin analogues and/or supplies were not properly delivered by governmental agencies. Another patient in Subgroup C was excluded due to pregnancy occurring during the study. One patient in Subgroup B was excluded after receiving a CSII device by the governmental agency. Finally, another patient in Group A abandoned CSII during the study.
The final statistical analysis included the 65 remaining patients (Group A, n = 23; Subgroup B, n = 22; Subgroup C, n = 20) (Figure 1). Table 1 shows the baseline characteristics of the study sample. Table 2 shows postintervention data including the initial and final focus weeks. Group A compared with Group B + C showed a trend (p = 0.05) toward more patients with > 70% of blood glucose levels in the target range (70-180 mg/dL) in the first focus week. This finding no longer held true in the final focus week, when no differences were observed in the percentage of patients with > 70% of blood glucose levels within the target range in comparisons between Group A versus      Figure 2 shows the data related to the primary outcome. On intragroup analysis, the mean A1C level decreased by 0.725% in Subgroup C (p = 0.0063). A mean variation of 0.421% occurred between the initial and final A1C levels in Group B + C compared with Group A (p = 0.08). Similarly, no significant variation in A1C levels occurred when Subgroup B was compared with Group A (p = 0.843). In contrast, significant mean variations from initial to final A1C levels occurred between Subgroup C versus Group A (0.834%, p = 0.003) and Subgroup C versus Subgroup B (0.789%, p = 0.009). Figure 3 shows the data related to the secondary outcome. On intragroup analysis, the variation in DSMP scores was not significant in Group A or Subgroup C, while the scores increased significantly in Subgroup B

DISCUSSION
The present study showed that the use of a blood glucose meter synchronized to the OneTouch Reveal mobile phone app improved both A1C levels and DSMP scores in patients with T1DM or LADA compared with standard treatment (CSII).
This study is relevant within the Brazilian context as it identified a group of patients with diabetes receiving only insulin analogues. Another strength of the study is the unprecedented comparison of CSII as a control group versus another group using a glucose meter connected to a mobile phone app. Even though the study was conducted prospectively, a potential limitation was the inclusion of patients from a single center. Other limitations include differences in action profile among the different types of insulin analogue used in the study and in CSII devices used by the participants in the control group.
Patient recruitment by alphabetical order and definition of the amount of time within the target range based on SMBG values are also potential limitations. Finally, the intervention in the present study lasted only 3 months, while Kirwan and cols. have reported better glycemic control with an intervention lasting 9 months (17).
Defining adherence can be complex in patients with T1DM or LADA (16), since it requires the assessment of tasks, behaviors (16)(17)(18)(19), and treatment regimens that are not universal or static (16,19) in the absence of a reliable biological marker (e.g., measurement of medication levels) (16).
In assessing glycemic control, laboratory measurement of A1C alone is insufficient since the measurement may be negatively affected by interfering substances (16)(17)(19)(20). In this sense, our analysis was based on CGM and SMBG results (16). In the first focus week, we observed a trend toward more patients in Group A maintaining capillary glucose at desirable levels (70-180 mg/dL) compared with those in Group B + C. The same finding was not confirmed in the final focus week, in which the percentage of patients with glucose within the target range at 70% of the time was comparable in Group A and Group B + C. However, this absence of difference between the groups could be explained by the small study sample.

Primary outcome
The decrease in A1C levels between the initial and final analyses was especially significant in patients in Subgroup C compared with those in Group A. We are unable to evaluate the significance of this finding compared with the potential decrease in A1C levels produced by some oral antidiabetic drugs for type 2 diabetes (9).
We believe that the findings of our study reflect a better understanding of the disease and self-care strategies (21), as well as an interest in blood glucose reports and glucose diaries. These features have been mentioned in a meta-analysis by Boyle and cols. as the most desirable in mobile phone apps for patients with diabetes (22). Because OneTouch Reveal is only meant for documentation purposes and lacks a bolus calculator, the use of this app perhaps added very little to individuals on MDI with carbohydrate counting, probably explaining the absence of significant A1C variation in Subgroup C when compared with Group A.
Aligned with the favorable results found in the present study, a systematic review by Knox and cols. has shown that technology-based interventions increase the frequency of SMBG (18). Indeed, by correlating capillary blood glucose readings with information provided by adherence to self-care activities, the patients can establish cause-and-effect relationships and correct problems related to glycemic control commonly observed in clinical practice. Additionally, the chronological correlation between capillary blood glucose readings and events influencing the performance of self-care tasks can better guide the physician toward choosing specific therapies.

Secondary outcome
Self-reported questionnaires may be useful as complementary information to objective assessments such as A1C and SMBG measurement (16). In the present study, DSMP scores correlated with adherence to self-care tasks (2,16), although the questionnaire items were not adaptable to the different treatment regimens (16). We found that the use of a glucose meter synchronized to the OneTouch Reveal mobile phone application improved the perception of self-care tasks in the subgroup of patients who already used insulin on MDI with carbohydrate counting regimen, despite the absence of A1C reduction. We postulate that the meter synchronized to the application enabled a more accurate view of all DSMP items by the patients, suggesting that the application can help patients on an MDI regimen to organize the daily demands involved in controlling their disease.
In conclusion, the use of mobile phone intervention apps synched to capillary blood glucose meters could play a key role in diabetes self-management in a country like Brazil, where CGM is still not widely established. The OneTouch Reveal app used in the present study improved both A1C levels and DSMP scores in patients with T1DM or LADA compared with standard treatment (CSII). In practice, patients on MDI with fixed insulin doses are likely to benefit most from this tool.