Effects of Ecological Momentary Interventions on Glycated Hemoglobin, Body Mass Index, and Quality of Life in People with Type 2 Diabetes: A Review

Summary

This systematic review and meta-analysis evaluate the effectiveness of ecological momentary interventions on glycemic control, weight management, and quality of life in people with type 2 diabetes.

Abstract

The global prevalence of type 2 diabetes mellitus (T2DM) continues to rise. Effective blood glucose and weight management are crucial for disease control and improving quality of life. Ecological momentary interventions (EMIs), which provide real-time personalized support based on ecological momentary assessments (EMAs) of relevant variables, offer a promising approach to promote adaptive self-management in daily life. We systematically searched five databases from inception to August 2025. Primary outcomes included glycated hemoglobin (HbA1c) and body mass index (BMI); secondary outcomes covered physical and mental health-related quality of life (HRQoL). Fourteen studies, including a total of 1,873 participants, were included. Compared to controls, EMIs significantly reduced HbA1c (MD = -0.71, 95% CI: -1.02 to -0.40, p < 0.001) and BMI (MD = -1.08, 95% CI: -1.90 to -0.26, p = 0.01), with substantial between-study heterogeneity (I² = 84% and 92%, respectively). Subgroup analyses showed more consistent effects for multi-domain comprehensive EMIs, while the effects of single-domain EMIs varied depending on the intervention content. EMIs were also associated with modest improvements in physical (SMD = 0.37, 95% CI: 0.17 to 0.56, p < 0.001; I² = 47%) and mental (SMD = 0.42, 95% CI: 0.15 to 0.68, p = 0.002; I² = 0%) HRQoL. These findings suggest that EMIs, especially multi-domain comprehensive strategies, appear to be a promising intervention for T2DM management, yet the overall certainty of current evidence is limited by marked heterogeneity, restricted statistical power in subgroups, and some risk of bias, warranting cautious interpretation of the results. Future research should prioritize high-quality designs and comprehensive outcome assessments.

Introduction

It is estimated that there were approximately 589 million people with diabetes worldwide in 2024, and this number is projected to rise to 853 million by 2050¹. Type 2 diabetes mellitus (T2DM) accounts for about 90–95% of all diabetes cases, while diabetes imposes an annual economic burden of nearly 1.015 trillion US dollars globally², placing substantial pressure on healthcare systems worldwide³. Effective blood glucose management is essential to reduce the disease burden of T2DM and improve the quality of life among people with T2DM. Notably, over 60% of individuals with T2DM also have overweight or obesity⁴; elevated body mass index (BMI) exacerbates insulin resistance and complicates glycemic control⁵. Poor management of blood glucose and body weight will significantly increase the risk of microvascular and macrovascular complications, potentially leading to life-threatening clinical outcomes^6,7. However, achieving optimal glycemic and weight control fundamentally relies on effective daily self-management behaviors by individuals, including but not limited to blood glucose monitoring, dietary adjustments, physical activity, and medication adherence⁸. These practices are vital for reducing disease-related risks and optimizing long-term outcomes.

In recent years, technology has increasingly been used to support self-management and improve treatment adherence among people with T2DM. Compared with traditional interventions, currently available telephone counseling, short message service (SMS) or other digital platforms have improved the accessibility and continuity of healthcare services to a certain extent^9,10,11. However, when individuals face immediate decision-making or challenges such as low motivation or limited knowledge in daily life, fixed interaction models still lack the ability to provide real-time, context-relevant support^12,13. In particular, some programs still rely on retrospective self-reports or clinical assessments with long intervals, which are inadequate for capturing the dynamic and contextual nature of the T2DM experience, thus failing to provide personalized support¹⁴. Therefore, within the evolving context of telemedicine, how to better meet the needs of people with T2DM for personalized, real-time feedback to promote their effective participation in diabetes self-management remains an urgent issue to be addressed^15,16.

Ecological momentary interventions (EMIs) refer to a behavioral support strategy that uses digital technology to deliver real-time and personalized feedback in individuals’ natural environments^17,18. Its key feature is the ability to provide adaptive interventions when unhealthy or high-risk behaviors occur, based on data captured by ecological momentary assessments (EMAs)—thereby overcoming the limitations of predetermined treatment protocols¹⁹. In this study, we clarified a operational definition for EMI in the present study based on core descriptions of EMI from existing literature^17,20, with three necessary minimum requirements for included studies: (1) Interventions are delivered via digital or electronic platforms in participants’ daily living environments and are accessible during participants’ routine activities, rather than being limited to clinical or supervised settings; (2) Information is obtained via EMAs, which capture momentary changes in individuals’ relevant states in daily life, and these data are used to guide the delivery of interventions when support is needed; (3) Interventions must provide adaptive and personalized feedback or behavioral support in response to collected EMA data, rather than relying solely on pre-set, non-tailored content that is unrelated to individual assessment results. Theoretically, by seamlessly integrating into daily life and dynamically adjusting support strategies according to real-time changes, EMIs can strengthen the relevance and effectiveness of interventions²⁰. It holds promise in facilitating the shift from habitual behaviors to goal-directed behaviors and improving self-management of the disease^21,22. As a form of digital health intervention, EMIs may be especially suitable for individuals with basic digital literacy and access to smart devices, who can benefit from real-time support for daily T2DM management decisions, particularly in settings where access to regular in-person support is limited²³. However, they are likely less appropriate for individuals who are uncomfortable using smart devices, and for those with concerns about data privacy¹⁷.

An increasing number of studies have begun to explore the application of EMIs in T2DM management. Most of these interventions focus on providing immediate guidance on lifestyle (e.g., dietary calorie control, exercise intensity adjustment) by capturing real-time dynamic data (e.g., blood glucose, diet, physical activity) from individuals^24,25,26. They are considered promising for promoting glycemic control and weight management among people with T2DM, providing objective and quantifiable evidence for evaluating the effectiveness of EMIs. Some studies have also assessed the impact of EMIs on patient-reported outcomes to fully understand their overall benefits. However, there has been no systematic integration and evaluation of these findings to date. Moreover, existing EMI studies exhibit significant differences in T2DM intervention strategies and content design, and there is a lack of discussion on their clinical feasibility. Thus, this systematic review and meta-analysis aim to assess the effectiveness of EMIs on clinically relevant outcomes in people with T2DM, providing evidence for the optimization and promotion of EMIs in T2DM management.

Protocol

The systematic review and meta-analysis were conducted following the PRISMA guidelines (Supplementary Table 1)²⁷. The protocol for this review was registered in PROSPERO: No. CRD420251132100.

1. Search strategy

1. Database selection: Perform a systematic approach across five major electronic databases: PubMed, Cochrane Library, Embase, Web of Science, and Scopus. Ensure the search encompasses all records from database inception until August 2025.
2. Search term development: Utilize a combination of medical subject headings (MeSH) terms and free-text keywords. Include key concepts such as “type 2 diabetes mellitus,” “ecological momentary intervention,” and “randomized controlled trial.” Combine these terms using appropriate Boolean operators (AND, OR).
3. Search execution: Tailor the constructed search syntax to the specific requirements of each database. Provide the detailed search strategy in Supplementary Table 2.

2. Application of eligibility criteria

1. Inclusion criteria: (1) Population: adults (age ≥18 years) diagnosed with T2DM, regardless of gender; (2) Intervention: EMIs delivered via technology-enabled tools (e.g., smartphones, mobile applications, sensors, wearable devices) that provide real-time feedback and support in routine primary care or home settings; (3) Comparison: usual care, standard care, or interventions without EMI-related components (e.g., traditional self-monitoring and guidance without real-time feedback); (4) Outcomes: glycated hemoglobin (HbA1c) is the gold standard for assessing long-term glycemic control in people with T2DM², and BMI is a core indicator for evaluating weight management and metabolic status that is closely associated with T2DM progression^2,28. Health-related quality of life (HRQoL) is an important patient-reported outcome for evaluating the impact of interventions on health and well-being²⁹. Therefore, select HbA1c and BMI as primary outcomes, and define patient-reported HRQoL as the secondary outcome. (5) Study type: randomized controlled trials (RCTs).
   ​NOTE: To more comprehensively reflect the potential benefits of the intervention, we plan to extract other patient-reported subjective outcomes related to self-management for further exploratory supplementary analyses. Accordingly, require eligible studies to report at least one of the following outcomes: HbA1c (measured in % or mmol/mol), BMI (kg/m²), or patient-reported outcomes (e.g., diabetes self-care activities or HRQoL); 
2. Exclusion criteria: Exclude studies that meet any of the following criteria: (1) report​ target outcomes but provide no corresponding statistical results for reference, or have no available full text; (2) report duplicate data; (3) enroll pregnant or lactating women with T2DM.

3. Study selection procedure

1. Management of records: Import all identified literature records into reference management software (Table of Materials) to facilitate the removal of duplicate records.
2. Screening process: Perform this task blindly, with two authors independently reviewing article titles and abstracts, excluding those that do not meet the inclusion criteria, and then reading the full texts of eligible articles for further verification. Resolve any discrepancies arising during this process through consensus with a third author.

4. Data extraction method

1. Extraction form: Extract data using a standardized, piloted form created in an electronic spreadsheet (Table of Materials).
2. Data points: Use variables for analysis that are consistent with the predefined outcomes specified in the inclusion criteria. Organize each article according to the basic publication information (year, country, author/s), participant characteristics (age, the number of female participants, sample size, and group distribution), recruitment strategies, details of the intervention groups (devices used for EMA, types and frequency of data collected, delivery platform, framework, methods, content, duration, and human involvement for EMI), and outcome measures.

5. Risk of bias assessment

1. Assessment tool: Assess the selected studies using the revised Cochrane risk-of-bias tool for randomized trials (RoB 2.0)³⁰.
2. Assessment domains: Appraise each RCT against five specific domains for bias: (1) process of randomization; (2) deviations from planned interventions; (3) missing outcome data; (4) outcome measurement; and (5) selection of the reported result³⁰.
3. Judgment process: Assign a bias value of ‘low risk’, ‘some concerns’, or ‘high risk’ for the overall rating³⁰. To minimize errors or bias, conduct the assessment independently by two authors. Resolve discrepancies between the two authors with a third author after evaluating the risk of bias in the specific study.

6. Certainty of evidence assessment

1. Assessment tool: Evaluate the certainty of evidence for all pooled clinical outcomes using the grading of recommendations assessment, development, and evaluation (GRADE) framework³¹.
2. Evidence certainty assessment domains: Appraise each outcome against five predefined domains for evidence downgrading: (1) risk of bias, informed by the aforementioned RoB 2.0 assessment results; (2) inconsistency, defined as substantial unexplained between-study heterogeneity; (3) indirectness, referring to mismatches between included studies and the core research question; (4) imprecision, determined by total sample size and the width of 95% confidence intervals (CIs); and (5) publication bias, assessed via funnel plot symmetry and sensitivity analyses³¹.
3. Evidence certainty judgment process: Initially rate evidence from the included RCTs as high certainty for all outcomes. Assign a final grade of high, moderate, low, or very low after full evaluation across all five domains³¹. Complete the assessment independently with two authors, and resolve any discrepancies through consensus discussion with a third author.

7. Data synthesis and analysis plan

1. Meta-analysis software: Use specialized meta-analysis software (Table of Materials) to conduct the meta-analysis for evaluating the efficacy of EMIs.
2. Effect measures: Summarize the primary outcomes (HbA1c and BMI) using the mean difference (MD). Convert HbA1c values to a uniform unit of mmol/mol, and express BMI in kg/m². Represent secondary outcomes (HRQoL) involving various scales using the standardized mean difference (SMD). Report all effect estimates with 95% CIs.
3. Heterogeneity and model selection: Evaluate the heterogeneity level among the studies via the I² test. Establish the threshold for statistical significance at a two-tailed probability (p) value < 0.05. Apply the fixed-effect model for the combined analysis if I² <50% (acceptable heterogeneity). Apply the random-effect model for analysis if I² ≥50%.
4. Additional analyses: Perform sensitivity analyses by sequentially excluding individual studies to evaluate their influence on the overall outcomes, according to the most recent cochrane handbook for systematic reviews of interventions³². Compare fixed-effect and random-effect estimates to assess the role of small-study effects in the intervention effect estimate. Attempt to perform subgroup analyses to investigate potential sources of heterogeneity and detect differential effects based on study design characteristics, drawing on prior relevant research and the intervention characteristics of EMIs.
   NOTE: Group studies by the core intervention domain of EMIs (single-domain vs. multi-domain), and perform further exploratory analyses based on EMA frequency (high vs. low), human involvement (i.e., whether the intervention was fully automated; yes vs. no), and EMI duration (≤3 months vs. >3 months). Extract all variables utilized in these analyses directly from the included studies.
5. For outcomes that cannot be pooled via meta-analysis (e.g., insufficient number of eligible studies, incomplete data for effect size extraction), provide a descriptive summary of study findings. Present this summary solely as supplementary exploratory content that carries no evidential weight to support clinically guiding conclusions and is not reported as a formal outcome.

Results

Study selection
After comprehensive searches across various databases, 2,691 articles were identified. Following the removal of duplicate records (n = 633), 2,058 remaining records were checked for titles and abstracts, and 2,029 were removed. We downloaded 29 full-text articles during the final screening and a more detailed investigation. Ultimately, only 14 articles satisfied the predetermined criteria (Figure 1).

Characte...

Discussion

This systematic review and meta-analysis of 14 studies aimed to evaluate the effectiveness of EMIs in the management of T2DM (Supplementary Table 4). The results indicate that EMIs significantly reduced HbA1c and BMI. Nevertheless, considerable heterogeneity was observed across studies, and subgroup analyses confirmed that intervention components were a key source of heterogeneity. Multi-domain comprehensive EMIs appear more beneficial for glycemic and BMI control, whereas among single-domain EMIs, the e...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
EndNote X9	Clarivate	Official Version	This software was utilized for managing bibliographic records and removing duplicate citations identified through the systematic database searches. It facilitated the initial organization of references prior to the screening process.
Review Manager 5.3 (RevMan 5.3)	The Cochrane Collaboration	Official Version	This program served as the primary tool for conducting the meta-analysis. It was used to perform all statistical calculations, generate forest plots for outcome analyses, and create risk of bias figures.
Microsoft Excel	Microsoft	Official Version	A standardized data extraction form was created in this spreadsheet software. It was used to systematically collect and tabulate relevant data from each of the included primary studies.