Effectiveness and Safety of the TRIO Optimal Health Management Program in Patients With Type 2 Diabetes Mellitus Initiating Basal Insulin Therapy: Prospective Observational Real-World Study

Introduction

Background

The prevalence of type 2 diabetes mellitus (T2DM) in China is rapidly increasing due to lifestyle changes and an aging population. As per the 2018 American Diabetes Association criteria, the estimated prevalence of total diabetes and prediabetes among Chinese adults escalated to 12.8% and 35.2%, respectively, between 2015 and 2017 [Li Y, Teng D, Shi X, Qin G, Qin Y, Quan H, et al. Prevalence of diabetes recorded in mainland China using 2018 diagnostic criteria from the American Diabetes Association: national cross sectional study. BMJ. 2020;369:m997. [FREE Full text] [CrossRef] [Medline]1]. Despite the wide range of medication options available for antidiabetic treatment, glycemic control rates among patients with T2DM remain suboptimal [Ramachandran A, Jain SM, Mukherjee S, Phatak S, Pitale S, Singh SK, et al. Suboptimal glycemic control among subjects with diabetes mellitus in India: a subset analysis of cross-sectional wave-7 (2016) data from the international diabetes management practices study (IDMPS). Ther Adv Endocrinol Metab. 2020;11:1-16. [FREE Full text] [CrossRef] [Medline]2]. A national survey conducted in 2018 revealed that only 32.9% (10,071/30,609) of patients with diabetes received treatment, with only half of them (15,336/30,609, 50.1%) achieving adequate glycemic control [Wang L, Peng W, Zhao Z, Zhang M, Shi Z, Song Z, et al. Prevalence and treatment of diabetes in China, 2013-2018. JAMA. 2021;326(24):2498-2506. [FREE Full text] [CrossRef] [Medline]3].

Diabetes, a chronic disease necessitating long-term treatment and self-management, presents significant challenges for patients who spend most of their treatment time outside of hospitals [Silva JAD, Souza ECF, Echazú Böschemeier AG, Costa C, Bezerra HS, Feitosa E. Diagnosis of diabetes mellitus and living with a chronic condition: participatory study. BMC Public Health. 2018;18(1):699. [FREE Full text] [CrossRef] [Medline]4]. When lifestyle intervention and oral antidiabetic drugs (OADs) fail to provide optimal control, patients with type 2 diabetes are required to initiate injectable therapies, mostly basal insulin (BI), according to 2020 Chinese guidelines for T2DM management [Jia W, Weng J, Zhu D, Ji L, Lu J, Zhou Z, et al. Standards of medical care for type 2 diabetes in China 2019. Diabetes Metab Res Rev. 2019;35(6):e3158. [CrossRef] [Medline]5].

Previous randomized controlled trials (RCTs) and observational studies have shown the efficacy of BI in controlled trials [Ji L, Kang ES, Dong X, Li L, Yuan G, Shang S, et al. Efficacy and safety of insulin glargine 300 U/mL versus insulin glargine 100 U/mL in Asia Pacific insulin-naïve people with type 2 diabetes: the EDITION AP randomized controlled trial. Diabetes Obes Metab. 2020;22(4):612-621. [FREE Full text] [CrossRef] [Medline]6] and real-world settings [Zhang P, Chen M, Zhang H, Luo Y, Zhu D, Li X, et al. Effectiveness and safety of basal insulin therapy in type 2 diabetes mellitus patients with or without metformin observed in a national cohort in China. BMC Endocr Disord. 2022;22(1):26. [FREE Full text] [CrossRef] [Medline]7]. Maintaining a delicate equilibrium between achieving optimal blood glucose control and mitigating hypoglycemia risks is pivotal. This involves the appropriate titration of insulin and diligent self-monitoring of blood glucose levels, both of which are integral to sustaining effective glycemic management. Consequently, establishing an optimal diabetes management framework encompassing health education, consistent professional follow-up, and comprehensive self-monitoring tools becomes imperative for effectively managing patients with type 2 diabetes [Kaufman N. Digital therapeutics: leading the way to improved outcomes for people with diabetes. Diabetes Spectr. 2019;32(4):301-303. [FREE Full text] [CrossRef] [Medline]8]. The lack of comprehensive and patient-centered approaches in current health care systems further compounds the burden of diabetes management. Time constraints and resource availability often limit traditional health education and face-to-face interactions with health care providers. As a result, there is a growing need for innovative solutions to bridge these gaps and provide ongoing support to individuals with type 2 diabetes [Adhikari M, Devkota HR, Cesuroglu T. Barriers to and facilitators of diabetes self-management practices in Rupandehi, Nepal—multiple stakeholders' perspective. BMC Public Health. 2021;21(1):1269. [FREE Full text] [CrossRef] [Medline]9-Iregbu S, Spiers J, Duggleby W, Salami B, Schick-Makaroff K. Nigerian health care providers and diabetes self-management support: their perspectives and practices. Qual Health Res. 2023;33(1-2):92-105. [FREE Full text] [CrossRef] [Medline]11]. The emergence of digital tools such as mobile apps and WeChat miniprograms has increased application in diverse therapeutic domains, such as attention-deficit/hyperactivity disorder, cancer, asthma, and insomnia tools to augment patient self-management [Wang C, Lee C, Shin H. Digital therapeutics from bench to bedside. NPJ Digit Med. 2023;6(1):38. [FREE Full text] [CrossRef] [Medline]12].

Objectives

The potential of digital therapeutics for diabetes has garnered recognition from different organizations, such as the Centers for Disease Control and Prevention and the Digital Therapeutics Alliance [Holmes D. Pharmacotherapy: a smarter way to treat obesity. Nat Rev Endocrinol. 2017;13(11):626. [CrossRef] [Medline]13]. Although some prior studies have demonstrated successful reductions in patients’ blood glucose levels and body weight through digital diabetes programs up to a hemoglobin A_1c (HbA_1c) reduction of 0.49% [Hou C, Carter B, Hewitt J, Francisa T, Mayor S. Do mobile phone applications improve glycemic control (HbA1c) in the self-management of diabetes? A systematic review, meta-analysis, and GRADE of 14 randomized trials. Diabetes Care. 2016;39(11):2089-2095. [CrossRef] [Medline]14], many studies were limited by including patients with prediabetes [Sepah SC, Jiang L, Ellis RJ, McDermott K, Peters AL. Engagement and outcomes in a digital diabetes prevention program: 3-year update. BMJ Open Diabetes Res Care. 2017;5(1):e000422. [FREE Full text] [CrossRef] [Medline]15], including patients treated with mostly premixed insulin [Lin J, Li X, Jiang S, Ma X, Yang Y, Zhou Z. Utilizing technology-enabled intervention to improve blood glucose self-management outcome in type 2 diabetic patients initiated on insulin therapy: a retrospective real-world study. Int J Endocrinol. 2020;2020:1-8. [FREE Full text] [CrossRef] [Medline]16], or evaluating user engagement outcomes rather than clinical outcomes [Böhm AK, Jensen ML, Sørensen MR, Stargardt T. Real-world evidence of user engagement with mobile health for diabetes management: longitudinal observational study. JMIR Mhealth Uhealth. 2020;8(11):e22212. [FREE Full text] [CrossRef] [Medline]17]. Consequently, limited evidence remains regarding the effectiveness of health management mobile apps specifically designed for patients with T2DM initiating BI.

Accordingly, we have developed a personalized health management program combined with artificial intelligence named “TRIO” for initiating BI therapy in patients with type 2 diabetes and aim to evaluate its effectiveness and safety. Unlike conventional acronyms or abbreviations, TRIO does not represent a longer phrase but rather represents the 3 essential pillars in comprehensive diabetes management: physician, nurse, and patient. This program combines traditional health education with a mobile app to enhance diabetes management. Through this evaluation, we aspire to contribute to understanding practical approaches to optimizing glycemic control and promoting patient well-being.

Methods

Study Design and Population

This prospective, 3-month observational program aimed to evaluate the effectiveness and safety of TRIO, an optimal health management program for patients with T2DM initiating BI therapy in a real-world setting. Participants were recruited from outpatient departments and, at discharge, from inpatient departments between December 1, 2019, and December 31, 2021, from 594 hospitals in China. Patients were assessed for their suitability by the following criteria: (1) aged between 18 and 85 years; (2) patients with T2DM who were inadequately controlled by OADs at the time of enrollment (ie, baseline HbA_1c level of ≥7%); (3) initiating BI therapy during the program period, meaning they had not used BI within 12 weeks prior to enrollment; (4) absence of mental disorders or communication impairments; and (5) absence of severe illnesses or limitations regarding follow-up. Patients who fulfilled these eligibility requirements were enrolled upon their willingness to provide informed consent. On the first day of enrollment, patients received health education from nurses, including knowledge about diabetes and insulin, psychological support for a healthy life with diabetes, and how to inject and store insulin. In addition, the physicians assisted patients in drawing up a self-management plan and helped them set individualized fasting plasma glucose (FPG) and postprandial glucose (PPG) goals. Patients were also asked to follow the WeChat official account of the TRIO program, through which knowledge about diabetes management would be sent. Follow-up assessments were conducted via phone calls at 1, 2, 4, 8, and 12 weeks. The frequency and follow-up methods were tailored to each patient’s FPG level. If the FPG level was 7 mmol/L, a phone call was not scheduled for the next follow-up visit, and only a WeChat message was sent (Figure 1).

‎

Ethical Considerations

This study was approved by the ethics committee of Nanjing Drum Tower Hospital (institutional review board review approval document, code: 2019-231-01), which was the principal research institute representing other subcenters. Implied consent was obtained from all participants when they registered on the TRIO WeChat official account following the principles of the Declaration of Helsinki, as the privacy policy included a clause allowing anonymized data to be used for research purposes. Participant privacy and anonymity were achieved through the elimination of any patient identifiers such as name or dates of birth, which were anonymized and deidentified before extraction and securely stored in compliance with data protection regulations. No compensation was provided to participants, as this study involved a secondary analysis of existing data. No identifiable images of participants were included in the study or supplementary materials, eliminating the need for additional image consent.

Data Collection

Baseline information was collected by interviews at the hospital enrollment, including demographics, disease characteristics, medical history, physical examination, BI types, starting dosage, and concomitant antidiabetic drugs (bolus insulin, glucagon-like peptide-1 receptor agonists, or OAD) used with BI. Laboratory tests including HbA_1c and FPG were obtained in hospitals at baseline, while glycemic control regarding HbA_1c and self-monitoring blood glucose (SMBG), including fasting blood glucose (FBG), dosage, and hypoglycemia information during follow-up time, were self-reported. Self-reported data were collected through phone calls by nurses or uploaded via a smart blood glucose device or input into the TRIO WeChat official account by the patients.

Outcomes

• Primary effectiveness end points: The primary effectiveness end point of our analysis is the change in HbA_1c levels from baseline to month 3.
• Secondary effectiveness end points: In addition to the primary end point, we also examined various other measures related to glycemic control. These secondary end points include changes in FBG levels from baseline to month 3, the achievement of target HbA_1clevels (<7%), the achievement of target FBG levels (<7 and <6.1 mmol/L), and an assessment of changes in BI dose.
• Safety end points: Our safety end points include monitoring and assessing the incidence and rates of hypoglycemia events and evaluating composite end points. The 3-month end point included the incidence and rates of hypoglycemia events and a composite end point encompassing the percentage of patients reaching target HbA_1c and FBG levels without experiencing hypoglycemia events.

Statistical Methods

Continuous variables were described using mean and SD, while categorical variables were presented as frequencies and percentages. For continuous effectiveness indicators, a paired t test (2-tailed) was applied to test the significance of the change from baseline to month 3 in HbA_1c or FPG in the total population. Analysis of covariance was used to compare these changes between subgroups, including patient source (inpatient or outpatient), complication status (no or yes), duration of diabetes (<5 years or ≥5 years), baseline HbA_1c (7%-8%, 8%-9%, 9%-10%, or ≥10%), and FBG goal setting (≥6.1 mmol/L or <6.1 mmol/L). Least square (LS) mean (SE) and LS mean difference with 95% CIs were provided. For binary effectiveness outcomes (HbA_1c <7%, FBG <7 mmol/L, or FBG <6.1 mmol/L), multivariable logistic regression models were applied to explore the association of subgroups with outcomes and variables included in the model were the same as analysis of covariance model. Hypoglycemic incidence and rate were evaluated by SMBG, uploaded by the smart glucose blood device or manual input to the TRIO platform by patients. Hypoglycemic incidence (percentage of patients with SMBG ≤3.9 mmol/L or SMBG <3.0 mmol/L) was analyzed using logistic regression, and odds ratio (OR) with 95% CI was used as the effect size; hypoglycemic rate (numbers of events per patient-year) was investigated by Poisson regression, and risk ratio (RR) with 95% CI were used as the effect size for this analysis. Composite end points including HbA_1c <7% without SMBG ≤3.9 mmol/L, FBG <7 mmol/L without SMBG ≤3.9 mmol/L, and FBG <6.1 mmol/L without SMBG ≤3.9 mmol/L were also explored using logistic regression. All the analyses were conducted using SAS 9.4 (SAS Institute, Inc), and a 2-sided P value of <.05 was considered statistically significant.

Results

Participant Recruitment

Between December 1, 2019, and December 31, 2021, a total of 225,764 patients were recruited from 594 hospitals. A total of 26,333 patients were excluded because of violating inclusion or meeting exclusion criteria, leaving 199,431 patients remaining at baseline. Among them, 81,297 patients were lost to follow-up within the first 3 months, resulting in 118,134 patients who completed the 3-month follow-up with measurements of either HbA_1c or FPG (Figure 2).

‎

Baseline Characteristics

The patients’ mean (SD) age at baseline was 57.3 (12.5) years, with 42.8% (85,337/1,99,431) of participants being women. The average BMI was 24.8 kg/m². At baseline, the mean HbA_1c, FPG, and PPG levels were 9.6%, 9.5 mmol/L, and 12.7 mmol/L, respectively. The mean duration of diabetes was 7.3 years. The most common complications or comorbidities observed in our patient cohort were hypertension (55,637/1,96,023, 28.4%), hyperlipidemia (32,240/1,96,023, 16.4%), and peripheral neuropathy (63,145/1,96,023, 32.2%). Most patients were taking BI with oral antihyperglycemic drugs (OADs) (1,59,930/1,98,969, 80.4%), with some also using prandial insulin concurrently.

Outpatients had a slightly higher mean age of 57.7 (SD 12.2) years than inpatients, with a mean age of 57.0 (SD 12.6) years. Gender distribution revealed that 56.3% (39,802/70,704) of outpatients were male, while 57.6% (74,064/1,28,645) of inpatients were male. Moreover, the duration of diabetes was slightly longer in inpatients, with a mean of 7.4 years (SD 6.8) than in outpatients, with a mean of 7.0 years (SD 6.3). Both groups displayed a similar average BMI of 24.8 kg/m² and exhibited comparable values for blood pressure and lipid levels such as triglycerides, total cholesterol, low-density lipoprotein, and baseline HbA_1c. Meanwhile, baseline FPG (10.3 vs 9.0 mmol/L) and PPG (14.0 vs 12.2 mmol/L) were slightly higher in outpatients than in inpatients. A notable difference was observed in the percentage of patients with complications or comorbidities, with 31.3% (39,950/1,27,610) of inpatients having hypertension compared with 22.9% (15,687/68,413) of outpatients and 38.1% (48,580/1,27,610) of inpatients having peripheral neuropathy compared with 21.3% (14,565/68,413) of outpatients. Inpatients also had a higher proportion of patients being treated with BI in combination with prandial insulin, accounting for 20.3% (26,037/1,28,342) of inpatients as opposed to 15.5% (10,924/70,627) of outpatients (Table 1). No clinically significant differences in baseline characteristics were observed between patients who remained in the study at month 3 and those who were lost to follow-up (Table S1 in

Multimedia Appendix 1

Loss to follow-up, dosage regimen, fasting blood glucose (FBG) target rate, incidence of hypoglycemia, and levels of FBG during the TRIO optimal health management program.

Initial Regimens and Adherence to BI Treatment

During the intervention, the majority of patients initiated insulin glargine (1,77,331/1,98,969, 89.1%) as their primary treatment, while a smaller proportion started with insulin determir (6588/1,98,969, 3.3%), neutral protamine Hagedorn insulin (766/1,98,969, 0.4%), or insulin degludec (8939/1,98,969, 4.5%). Alongside BI treatment, 41% (57,242/1,39,739), 28.2% (39,473/1,39,739), and 9.7% (13,545/1,39,739) of patients were concurrently taking 1, 2, and ≥3 OADs, respectively. The most commonly used OADs were metformin (69,027/1,39,739, 49.4%) and α-glucosidase inhibitors (46,960/1,39,739, 33.6%), followed by sodium-glucose cotransporter-2 inhibitors (22,911/1,39,739, 16.4%), dipeptidyl peptidase-4 inhibitors (20,722/1,39,739, 14.8%), sulfonylureas (7266/1,39,739, 5.2%), glinides (6257/1,39,739, 4.5%), and thiazolidinediones (5504/1,39,739, 3.9%). Sulfonylureas were more frequently prescribed to outpatients (3571/48,696, 7.3%) than inpatients (3695/91,043, 4.1%), while sodium-glucose cotransporter-2 inhibitors were more commonly used in inpatients (16,615/91,043, 18.2%) than outpatients (6296/48,696, 12.9%; Table S2 in

Multimedia Appendix 1

Loss to follow-up, dosage regimen, fasting blood glucose (FBG) target rate, incidence of hypoglycemia, and levels of FBG during the TRIO optimal health management program.

Glycemic Outcomes

At the end of the 3-month management intervention period, the mean HbA_1c level among 44,847 participants with eligible self-reported mean HbA_1c measurements was 6.89% (SD 0.90). This represented a mean decrease in HbA_1c by –2.59% (SE 0.01; P<.001) from baseline (Table 2). Regarding FBG levels, at the end of month 3, the mean FBG level among 60,365 participants with eligible self-reported FBG measurements was 6.81 (SD 1.4) mmol/L, indicating an average decrease in FBG by –2.77 (SD 0.01) mmol/L (P<.001) from baseline (Table 2 and Figure S1 in

Multimedia Appendix 1

Loss to follow-up, dosage regimen, fasting blood glucose (FBG) target rate, incidence of hypoglycemia, and levels of FBG during the TRIO optimal health management program.

Table 3

In the subgroup analyses, patients with complications experienced slightly lower HbA_1c (LS mean difference: 0.07, 95% CI 0.05-0.09) than those with no complications. Considering the duration of diabetes, patients with a duration of ≥5 years exhibited a lesser decrease in HbA_1c (LS mean difference: 0.09, 95% CI 0.07-0.11) than those with a duration of <5 years. A higher baseline HbA_1c level was also associated with a more significant reduction. Compared with patients with baseline HbA_1c in the range of 7% to 8%, patients with a baseline HbA_1c in the range of 8% to 9% had the lesser decrease (LS mean difference: –0.83, 95% CI –0.86 to –0.79), while patients with a baseline HbA_1c of 10% or higher had the highest decrease (LS mean difference: –4.04, 95% CI –4.08 to –4.00). Also, patients with initial FBG goal setting of <6.1 mmol/L had a greater decrease in their HbA_1c (LS mean difference: –0.36, 95% CI –0.38 to –0.34) than those with FBG goal setting of ≥6.1 mmol/L (Table 2). In the subgroup analysis exploring factors related to FBG reduction, consistent results with HbA_1c reductions found that more significant reductions were seen in patients with no complications, diabetes duration of <5 years, and an initial FBG goal setting of <6.1 mmol/L, except for baseline HbA_1c. Patients with lower baseline HbA_1c exhibited higher reductions in FBG from baseline (Table 2).

Regarding target HbA_1c of <7% at month 3, patients enrolled from inpatient showed a slight advantage over outpatient (OR 1.16, 95% CI 1.11-1.21; P<.001), and patients with complications had lower odds of achieving HbA_1c target than those with no complications (OR 0.85, 95% CI 0.81-0.89; P<.001). Those with ≥5 years of duration of diabetes had lower success rates than those with <5 years of duration (OR 0.83, 95% CI 0.79-0.87; P<.001). Lower baseline HbA_1c levels were associated with better outcomes. Moreover, setting FBG target goal level of <6.1 mmol/L at the beginning of the treatment demonstrated a higher possibility of reaching the HbA_1c target of <7% at month 3 (OR 1.8, 95% CI 1.7-1.9; P<.001) (Table 3). Regarding the FBG target of <7 and <6.1 mmol/L at month 3, patients exhibited consistent results as in HbA_1c target of <7%. Inpatients; patients with no complications; and patients with diabetes duration of 5 years, lower HbA_1c target, and initial FBG goal setting of <6.1 mmol/L were associated with higher odds of achieving the target (Table 3).

Insulin Dose and Satisfaction

Total insulin dose (U/d/kg) change was –0.01 (SD 0.06), from baseline (mean 0.23, SD 0.09) to month 3 (mean 0.22, SD 0.09; Table 4). Patients recruited from inpatients, with complications, having diabetes duration of ≥5 years, with FBG goal setting of <6.1 mmol/L and higher baseline HbA_1c, had a higher starting dose of BI. Among 36,037 patients with both baseline and 3-month BI dosages, 7% (2546/36,037) remained unchanged, and 50.1% (18,047/36,037) lowered the dosage per kilogram (Table 4). Possible reasons for the lack of titration, such as patients reaching FBG targets or experiencing hypoglycemic events, were explored in Table S3 in

Multimedia Appendix 1

Loss to follow-up, dosage regimen, fasting blood glucose (FBG) target rate, incidence of hypoglycemia, and levels of FBG during the TRIO optimal health management program.

Incidence and Event Rate of Hypoglycemia

Hypoglycemia incidence (≤3.9 mmol/L) in all patients was 27.1% (3317/12,227). Inpatients had a higher incidence than outpatients (OR 1.24, 95% CI 1.07-1.45; P=.005). Patients with complications experienced more hypoglycemia (OR 1.25, 95% CI 1.07-1.45; P=.005). Longer diabetes duration (≥5 years) was associated with lower hypoglycemia incidence (OR 0.60, 95% CI 0.52-0.69; P<.001; Table 5). Higher baseline HbA_1c levels correlated with increased hypoglycemia risk. HbA_1c ≥10% had the highest incidence (OR 1.47, 95% CI 1.23-1.76; P<.001; Table 5). For hypoglycemic events defined by SMBG levels of ≤3.9 mmol/L, a total of 7619 events occurred, yielding an event rate of 2.49 events per person-year (Table 6). Notable trends included higher hypoglycemia rates for inpatient sources compared with outpatient sources (RR 1.25, 95% CI 1.08-1.46; P=.004), higher rates in patients with complications compared with those with no complications (RR 1.25, 95% CI 1.07-1.47; P=.006), and a lower rate in patients with a diabetes duration of ≥5 years (RR=0.67, 95% CI 0.59-0.77; P<.001). Furthermore, elevated baseline HbA_1c levels (≥10%) were associated with a higher hypoglycemia rate (RR 1.37, 95% CI 1.15-1.63; P=.001). An FBG goal setting of <6.1 before initiating BI was not associated with increased hypoglycemic incidence (OR 0.97, 95% CI 0.77-1.22; P=.79) or rate (RR 0.92, 95% CI 0.76-1.1; P=.35). For hypoglycemia defined as SMBG <3.0 mmol/L, 15.1% (1852/12,227) of patients with 2954 events were recorded, resulting in an incidence rate of 0.97 events per person-year. Similar trends were observed in relation to complications, duration of diabetes, and baseline HbA_1c levels. An FBG goal of <6.1 was not related to increased incidence (OR 0.97, 95% CI 0.77-1.22; P=.79) or rate (RR 1.01, 95% CI 0.80-1.27; P=.95) of hypoglycemia (Table 6).

Table 7 shows the composite end points of patients reaching the target with no hypoglycemia events during a 3-month period. For HbA_1c <7% without SMBG ≤3.9 mmol/L at month 3, patients with complications (OR 0.68, 95% CI 0.54-0.84; P=.001) and those with a duration of diabetes for ≥5 years (OR 0.76, 95% CI 0.62-0.93; P=.008) had significantly lower odds of reaching the target with no hypoglycemia events. Furthermore, higher baseline HbA_1c levels in the ranges of 9%-10% (OR 1.26, 95% CI 1.03-1.55; P=.02) and ≥10% (OR 1.47, 95% CI 1.23-1.76; P<.001) were associated with increased odds of achieving the target. Also, those who set an FBG goal of <6.1 mmol/L at initiation had significantly higher odds of reaching the composite end point (OR 1.35, 95% CI 1.03-1.79; P=.03). None of these factors were related to the composite end points for achieving FBG <7 mmol/L without SMBG ≤3.9 mmol/L at month 3. Finally, for achieving FBG <6.1 mmol/L without SMBG ≤3.9 mmol/L at month 3, only the duration of diabetes for ≥5 years was associated with a significantly lower possibility of achieving this composite end point.

Discussion

Principal Findings

The TRIO program, a large-scale health management initiative using a digital WeChat platform for patients with T2DM initiating BI treatment, demonstrated its effectiveness in improving glycemic control 3 months after initiating BI. Prior to enrollment in TRIO, patients with T2DM exhibited suboptimal blood glucose control, with elevated baseline HbA_1c (9.6%) and FPG (9.5 mmol/L), a high prevalence of diabetic complications, and long diabetes duration (7.3 years). Following the 3-month TRIO management intervention, notable reductions in HbA_1c (–2.59%) and FBG (–2.77 mmol/L) were observed in the total population, accompanied by heightened proportions of achieving HbA_1c <7% (55.6%) and FBG target <7.0 mmol/L (61.3%) across diverse subgroups, such as patients from inpatient or outpatient care, patients with or with no complications, and patients with different length of diabetes duration, baseline HbA_1c, and FBG goal setting. This study also highlights the potential for setting a lower FBG target (<6.1 mmol/L) at the initiation of BI compared with the traditional <7.0 mmol/L target. By setting a more rigorous FBG target of <6.1 mmol/L, better glycemic control was achieved without increased risk of hypoglycemia. These results hold promise for digital health tools such as TRIO in improving the overall management of T2DM in real-world clinical settings.

TRIO has shown effectiveness and safety in patients with T2DM initiating BI after OAD failure with the assistance of WeChat digital platform, which is consistent with previous single-arm studies incorporating digital tools conducted in patients with prediabetes [Sepah SC, Jiang L, Ellis RJ, McDermott K, Peters AL. Engagement and outcomes in a digital diabetes prevention program: 3-year update. BMJ Open Diabetes Res Care. 2017;5(1):e000422. [FREE Full text] [CrossRef] [Medline]15,Buch A, Yeshurun S, Cramer T, Baumann A, Sencelsky Y, Zelber Sagi S, et al. The effects of metabolism tracker device (Lumen) usage on metabolic control in adults with prediabetes: pilot clinical trial. Obes Facts. 2023;16(1):53-61. [FREE Full text] [CrossRef] [Medline]18] and patients treated with premixed insulin and BI [Hou C, Carter B, Hewitt J, Francisa T, Mayor S. Do mobile phone applications improve glycemic control (HbA1c) in the self-management of diabetes? A systematic review, meta-analysis, and GRADE of 14 randomized trials. Diabetes Care. 2016;39(11):2089-2095. [CrossRef] [Medline]14,Lorig K, Ritter PL, Turner RM, English K, Laurent DD, Greenberg J. Benefits of diabetes self-management for health plan members: a 6-month translation study. J Med Internet Res. 2016;18(6):e164. [FREE Full text] [CrossRef] [Medline]19,Hermanns N, Ehrmann D, Finke-Groene K, Krichbaum M, Roos T, Haak T, et al. Use of smartphone application versus written titration charts for basal insulin titration in adults with type 2 diabetes and suboptimal glycaemic control (My Dose Coach): multicentre, open-label, parallel, randomised controlled trial. Lancet Reg Health Eur. 2023;33:100702. [FREE Full text] [CrossRef] [Medline]20]. For instance, the Omada Health Program investigated digital Diabetes Prevention Program engagement among patients with prediabetes, showing a reduction of –0.33 mmol/L in HbA_1c levels over 3 years [Sepah SC, Jiang L, Ellis RJ, McDermott K, Peters AL. Engagement and outcomes in a digital diabetes prevention program: 3-year update. BMJ Open Diabetes Res Care. 2017;5(1):e000422. [FREE Full text] [CrossRef] [Medline]15]. In our TRIO study, with a larger sample size, we achieved a greater HbA_1c reduction of –2.58 mmol/L. Furthermore, in a 12-week German trial involving individuals with type 2 diabetes on BI, a smartphone app (My Dose Coach) was compared with a written titration chart. The intervention group using the app exhibited a noteworthy reduction in HbA_1c levels compared with the control group (–0.31%; P=.04), with safety outcomes remaining unaffected. These findings suggest that app-assisted titration can enhance glycemic control in patients with type 2 diabetes who use BI [Hermanns N, Ehrmann D, Finke-Groene K, Krichbaum M, Roos T, Haak T, et al. Use of smartphone application versus written titration charts for basal insulin titration in adults with type 2 diabetes and suboptimal glycaemic control (My Dose Coach): multicentre, open-label, parallel, randomised controlled trial. Lancet Reg Health Eur. 2023;33:100702. [FREE Full text] [CrossRef] [Medline]20].

TRIO has demonstrated that digital tools including health education and self-management modules added on BI might provide additional benefits to effectiveness in glycemic control than medication alone. The ORBIT study is an observational registry conducted in China with patients with T2DM who were inadequately controlled by OADs and initiated BI [Ji L, Zhang P, Zhu D, Li X, Ji J, Lu J, et al. Observational registry of basal insulin treatment (ORBIT) in patients with type 2 diabetes uncontrolled with oral antihyperglycaemic drugs: real-life use of basal insulin in China. Diabetes Obes Metab. 2017;19(6):822-830. [CrossRef] [Medline]21], with similar baseline HbA_1c (9.6%, SD 2%) but higher baseline FBG (11.7, SD 4.0 mmol/L) and shorter diabetes duration (6.4, SD 5.3 years) than those in our study. Notably, the change in HbA_1c from baseline to month 3 demonstrated a more improvement in the TRIO group (–2.59%) than in the ORBIT group (–2%), as well as the attainment of the HbA_1c target of <7% at month 3 (55.6% vs 35.9%). Despite the relatively lower reduction in FBG levels in the TRIO study due to lower baseline FBG levels, a larger proportion of TRIO patients successfully reached the FBG target of <7 mmol/L (37,017/60,377, 61.3%) than those in the ORBIT study (2078/5571, 37.3%). Another significant study in this field, the First Basal Insulin Evaluation (FINE) Asia study, was a multinational, prospective, observational approach to assess BI’s efficacy in patients with uncontrolled type 2 diabetes (HbA_1c ≥8%) [Tsai ST, Pathan F, Ji L, Yeung VT, Chadha M, Suastika K, et al. First insulinization with basal insulin in patients with type 2 diabetes in a real-world setting in Asia. J Diabetes. 2011;3(3):208-216. [FREE Full text] [CrossRef] [Medline]22].

In TRIO, baseline HbA_1c and FBG were as high as 9.6% and 9.5 mmol/L, respectively (Table 1), which suggests delayed initiation of BI. The American Diabetes Association and the European Association for the Study of Diabetes suggest that BI should be promptly considered after the apparent “failure” of lifestyle modifications, including diet and exercise in combination with metformin, particularly when HbA_1c levels remain at or exceed 7% for a span of 2-3 months [Nathan DM, Buse JB, Davidson MB, Ferrannini E, Holman RR, Sherwin R, et al. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care. 2009;32(1):193-203. [FREE Full text] [CrossRef] [Medline]23]. However, consistent with our findings, delay in injectable therapies was universal [Goodall G, Sarpong EM, Hayes C, Valentine WJ. The consequences of delaying insulin initiation in UK type 2 diabetes patients failing oral hyperglycaemic agents: a modelling study. BMC Endocr Disord. 2009;9(1):19. [FREE Full text] [CrossRef] [Medline]24,Kim SG, Kim NH, Ku BJ, Shon HS, Kim DM, Park TS, et al. Delay of insulin initiation in patients with type 2 diabetes mellitus inadequately controlled with oral hypoglycemic agents (analysis of patient- and physician-related factors): a prospective observational DIPP-FACTOR study in Korea. J Diabetes Investig. 2017;8(3):346-353. [FREE Full text] [CrossRef] [Medline]25]. Timely initiation of insulin such as BI after the failure of oral treatment is associated with better glycemic control [Chen P, Ma X, Chen H, Wang K, Zhou L. Delays in insulin initiation among patients with type 2 diabetes mellitus in southeast China: a retrospective, real-world study. Diabetes Metab Syndr Obes. 2020;13:3059-3068. [CrossRef]26].

Standard T2DM management advice recommends keeping HbA_1c levels below 7%, but the ideal FPG target for achieving this is debated [Davies MJ, D'Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G, et al. Management of hyperglycemia in type 2 diabetes, 2018. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care. 2018;41(12):2669-2701. [FREE Full text] [CrossRef] [Medline]27]. Different guidelines suggest varying FPG targets, such as 4.4-7.2 mmol/L according to the American Diabetes Association 2018 guidelines [American Diabetes Association. 6. Glycemic targets standards of medical care in diabetes—2018. Diabetes Care. 2018;41(Suppl 1):S55-S64. [CrossRef] [Medline]28] or <6.1 mmol/L according to the American Association of Clinical Endocrinology-American College of Endocrinology and the International Diabetes Federation [Garber AJ, Abrahamson MJ, Barzilay JI, Blonde L, Bloomgarden ZT, Bush MA, et al. Consensus statement by the American Association of Clinical Endocrinologists and American College of Endocrinology on the comprehensive type 2 diabetes management algorithm—2018 executive summary. Endocr Prac. 2018;24(1):91-121. [CrossRef]29]. Previous studies support an FPG target of 6.1 mmol/L, showing better outcomes. Patients with FPG goals below 6.1 mmol/L had more significant HbA_1c reductions and higher target achievement rates without an increase in hypoglycemia [Yang W, Ma J, Yuan G, Li L, Zhang M, Lu Y. Determining the optimal fasting glucose target for patients with type 2 diabetes: results of the multicentre, open-label, randomized-controlled FPG GOAL trial. Diabetes Obes Metab. 2019;21(8):1973-1977. [CrossRef]30,Yuan L, Li F, Zhou Y, Sun R, Gao G, Zhang Q, et al. Fasting glucose of 6.1 mmol/L as a possible optimal target for type 2 diabetic patients with insulin glargine: a randomized clinical trial. J Diabetes Res. 2021;2021:5524313. [FREE Full text] [CrossRef] [Medline]31]. Our results might confirm a better FPG target of <6.1 mmol/L. Patients who had an initial FPG goal setting below 6.1 mmol/L by their physician at the time of enrollment experienced both greater reductions in their HbA_1c levels (–2.64 vs –2.57%) and a higher HbA_1c target rate (67.8% vs 53.1%) than those with FPG goal setting ≥6.1 mmol/L (Tables 2 and 3). At the same time, hypoglycemic incidence and rate were comparable between the 2 groups. To enhance the management of hypoglycemic events, we recommend frequent SMBG monitoring, particularly during the initial weeks of insulin titration, to detect and address hypoglycemia promptly. Patient education on recognizing and treating hypoglycemic symptoms should be prioritized, alongside individualized insulin dose adjustments based on SMBG trends to minimize risk. Regular follow-ups are essential to reassess glycemic targets, such as FBG and HbA_1c, and to prevent overtreatment while maintaining optimal glycemic control. These measures can help balance achieving strict glycemic targets with ensuring patient safety.

Regarding the titration of BI treatment, the current Chinese guideline recommends an initial dose of 0.2 U/kg or 10 U, underscoring the importance of active insulin dose adjustment to achieve optimal glycemic control [Shi G, Zhu N, Qiu L, Yan H, Zeng L, Wang D, et al. Impact of the 2020 China Diabetes Society guideline on the prevalence of diabetes mellitus and eligibility for antidiabetic treatment in China. Int J Gen Med. 2021;14:6639-6645. [CrossRef]32]. Previous studies such as ORBIT have indicated inadequate titration, evident from a starting dose of 0.18 IU/kg/d and a final dose of 0.21 IU/kg/d, resulting in a change of +0.034 IU/kg/d. Within our program, comprehensive titration was not uniformly accomplished. Among the 36,037 patients with baseline and 3-month dosage data, 42.9% (15,444/36,037) of patients escalated their dosage during the program, while 7.1% (2546/36,037) maintained stability and 50.1% (18,047/36,037) of patients decreased their dosage. Consequently, there was a marginal numerical decline in dose by –0.01 (0.06) U/kg. Plausible explanations for this trend encompass the higher-than-recommended starting dose in our program, which even surpassed the final doses in earlier ORBIT studies. Furthermore, the pursuit of targeted FBG levels in the initial weeks and an increased incidence of hypoglycemic events among patients (as shown in Table S3 in

Multimedia Appendix 1

Loss to follow-up, dosage regimen, fasting blood glucose (FBG) target rate, incidence of hypoglycemia, and levels of FBG during the TRIO optimal health management program.

Limitations

Our study supports TRIO’s effectiveness and safety as a personalized health program, yet several limitations warrant acknowledgment. First, a significant portion of patients lacked HbA_1c follow-up data at month 3, possibly introducing a compliance bias that could overstate TRIO’s effectiveness by excluding those with less favorable HbA_1c reductions. Nonetheless, comparable baseline characteristics between compliant and noncompliant patients mitigate the risk of overestimation. Long-term data (months 3-12) are insufficient, leaving uncertainty about TRIO' s enduring effectiveness and ability to sustain patient adherence. Second, the judgment of hypoglycemia relied on SMBG, and the SMBG data were collected mainly through self-reporting on the WeChat platform, smart glucose devices, and regular phone calls by nurses, which may have a certain degree of deviation from the actual occurrence of hypoglycemia. Moreover, the absence of an RCT design and an external control group that included patients with type 2 diabetes who initiated only BI therapy without using TRIO may lead to the effect of TRIO to optimize glycemic control not solid enough. Therefore, it is necessary for future studies to design RCTs to evaluate the effectiveness and safety of intelligent blood glucose management in patients with diabetes with initiated BI therapy.

Conclusions

The TRIO program has demonstrated effectiveness in glycemic control, as reflected in HbA_1c and FBG levels, among patients with T2DM initiating BI therapy. The program has improved HbA_1c and FBG target rates and patient compliance with insulin treatments. However, it is important to acknowledge the limitations of our study, including compliance bias, insufficient long-term follow-up data, and the need for further investigation using rigorous study designs. Future research, such as RCTs, is warranted to validate our study’s findings and assess TRIO’ s generalizability in real-world populations.