Effect of a Mobile Health Technology–Based Diabetes Education Program on Glucose Control in Patients With Type 2 Diabetes Initiating Premixed Insulin: A Prospective, Multicenter, Observational Study

Due to the scarcity of medical resources in China, very often patients with diabetes do not get enough diabetes education from the clinic, while those in rural areas or ethnic minority groups have less access to education and support. Results of previous studies have indicated that mobile health applications (mHealth apps) are feasible tools for improving self-management of diabetes (1,2). Specifically, compared with treatment with oral hypoglycemic agents, the initiation of insulin involves more comprehensive education such as injection skills and insulin dose adjustment. However, little is known about the role of mHealth apps for patients initiating insulin. Therefore, in the current study, we investigated the effect of the Lilly Connected Care Program (LCCP), an app-based diabetes education program, on glycemic control in patients with type 2 diabetes initiating premixed insulin. The study is registered at chictr.org.cn (clinical trial reg. no. 1900021191).

The study protocol has been published previously (3). Briefly, we enrolled patients with type 2 diabetes aged ≥18 to <80 years, with uncontrolled HbA_1c (≥8% [64 mmol/mol] to <15% [140 mmol/mol]) after treatment with one or more oral antihyperglycemic medications, who were willing to initiate premixed insulin analog and could use their social media (WeChat) accounts independently. After the prescription of premixed insulin analog, eligible subjects were invited to participate in the 12-week LCCP program by their health care professionals (HCPs). On valid registration on the LCCP program through WeChat, the participant could freely download the app and use the resources provided by the platform without any expense. HCPs could interact with their patients and provide advice on diabetes management including treatment adjustments. Data from patients across 250 hospitals in China recruited to the LCCP platform from 1 April 2019 to 30 June 2021 were analyzed.

Patient engagement was assessed by their behaviors in the “Mini Course” and “Daily Quiz” sections during the program. There are 60 diabetes education courses, presented in the form of audio and text, on the LCCP platform. Patients could choose any of the courses of interest. Self-reported HbA_1c levels at baseline and after 12 weeks were obtained through the program.

Overall, 9,426 patients were included for the study (male, 53.3%, and mean ± SD age 50.9 ± 12.5 years, duration of diabetes 4.4 ± 5.9 years, and initial insulin dose 28.0 ± 10.1 units/day). After 12-week intervention, mean HbA_1c decreased from 9.8 ± 1.5% (84 ± 16.4 mmol/mol) to 7.4 ± 1.2% (57 ± 13.1 mmol/mol) at week 12 (P < 0.001), with attainment of HbA_1c goal (<7% [53 mmol/mol]) increased from 0 to 36%. The proportion of subjects experiencing hypoglycemia reduced from 10.1% at week 1 to 4.4% at week 12.

Approximately 61% of the patients with type 2 diabetes had completed at least one education course during the study. Approximately 61% of the patients participated in the Daily Quiz during the study. The proportion of patients taking 0–4 courses, 5–29 courses, and ≥30 courses was 74%, 8%, and 18%, respectively. The mean HbA_1c reduction increased progressively with ascending categories of completed courses (0–4 courses, −2.4% [−26.2 mmol/mol]; 5–29 courses, −2.7% [−29.5 mmol/mol]; ≥30 courses, −2.9% [−31.7 mmol/mol]; P_trend < 0.001) (Fig. 1A). After adjustment for covariates including age, sex, BMI, education background, duration of diabetes, baseline insulin dosage, oral antidiabetes agents, baseline HbA_1c, and baseline frequency of blood glucose monitoring, the number of completed courses was significantly related to HbA_1c at week 12.

Moreover, the mean HbA_1c reduction was significantly associated with the amount of time spent on learning in the Mini Course (0–30 min, −2.4% [−26.2 mmol/mol]; 30–150 min, −2.6% [−28.4 mmol/mol]; ≥150 min, −2.8% [−30.6 mmol/mol]; P_trend < 0.001) (Fig. 1B). Patients participating in the Daily Quiz section >20 times during the study reported greater mean HbA_1c reductions (−2.8 vs. −2.5 vs. −2.4% [−30.6 vs. −27.3 vs. −26.2 mmol/mol]; P_trend < 0.001) at week 12 compared with those participating <20 times or without any participation (Fig. 1C). In addition, individuals who preferred the audio + visual learning mode for engagement of education courses had significantly greater mean HbA_1c reduction compared with those who preferred the visual-only learning mode (audio + visual, −2.7% [−29.5 mmol/mol]; visual only, −2.4% [−26.2 mmol/mol]; P < 0.001) (Fig. 1D).

The timely initiation of insulin is associated with significantly improved metabolic control in subjects with type 2 diabetes inadequately controlled with oral antihyperglycemic medications (4). Indeed, HbA_1c was greatly improved even in patients taking only 0–4 courses (mean reduction: −2.4% [−26.2 mmol/mol]), which is mostly explained by the addition of premixed insulin. In this context, we provide evidence that an app-based educational platform is feasible and effective for glucose management in patients initiating insulin. This is one of the largest studies to date focusing on the effect of mobile health technology–based diabetes education. The strength of the study also included the multicenter design across China. Nevertheless, certain limitations should be noted. First, the observation period of the study was short, and further evaluation is needed for the long-term effects of the education program. However, given the scarcity of education from HCPs in clinical practice in China and the sustained access to diabetes education provide by the platform, we believe that the app-based program may exert beneficial effects in the long run. Second, due to the observational design of the study, potential confounding caused by unmeasured variables such as socioeconomic factors, which have been reported to affect clinical outcomes (5), is almost inevitable. Third, as the study was conducted across China, self-reported HbA_1c was used as the primary end point instead of central laboratory–measured HbA_1c. Therefore, recall bias could not be excluded. Finally, only subjects with type 2 diabetes who initiated insulin treatment were enrolled in the study. Whether our findings could be applied to other patient groups remains to be determined.

In summary, the 12-week LCCP program along with the initiation of insulin significantly decreased HbA_1c in patients with type 2 diabetes. Patient engagement with diabetes education in the program was significantly associated with improvement of glucose control, implying that measures should be taken to increase patient adherence to online educational platforms for better diabetes care.