This review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines published in 2020 [24], and a search protocol retrospectively registered on July 5, 2024 on PROSPERO (CRD42024560431) formed the basis of the review process.

We searched PubMed, Web of Science, Embase, Cochrane Library, China National Knowledge Infrastructure, Wanfang, and the Chinese Biomedical Literature Database using the terms “Diabetes Mellitus, Type 2,” “Internet,” and “RCT” from inception until December 25, 2023. The search strategy used Boolean logic to combine medical subject headings and text word searches. We also screened references from relevant previous systematic reviews as additional sources, and the languages were limited to Chinese and English. The complete search strategy can be found in Multimedia Appendix 1.

Inclusion criteria were formulated based on the populations, interventions, comparisons, outcomes, and study design (PICOS) framework:

Exclusion criteria included (1) studies involving patients with gestational diabetes or type 1 diabetes, (2) review or discussion papers, and (3) studies with insufficient data.

Two reviewers (YX and NX) independently performed the screening and selection of studies. Disagreements were resolved through discussion or consultation with a third reviewer (ZW). First, we imported the retrieved literature into the NoteExpress software (Beijing Aiqinhai Software Company) and used it to identify and remove duplicate literature. Then, we evaluated whether the titles and abstracts were eligible. In addition, clearly incompatible studies were excluded after screening the titles and abstracts. Finally, full texts of qualified studies that strictly adhered to the inclusion and exclusion criteria [25] were screened to determine their final eligibility for the review.

Data extraction from the studies included was performed by 2 independent reviewers. Characteristics of the studies (author, year, country, and sample size) and intervention characteristics (equipment used by the intervention, intervening measures, control measures, duration, and outcomes) were included.

The quality of the included studies was assessed using the Cochrane risk-of-bias tool, administered by 2 independent assessors. This tool evaluates the quality of study methodology across seven core dimensions: (1) the process of randomization sequence generation, (2) the concealment of allocation, (3) blinding of participants and researchers, (4) blinding during outcome assessment, (5) completeness of outcome data and handling of missing data, (6) selective reporting of study results, and (7) the potential for other sources of bias. The risk of bias for each domain was categorized into 3 levels: low, unclear, or high. Disagreements between the assessors were resolved through discussion, or, if necessary, consultation with a senior reviewer (ZW).

The meta-analysis was conducted using RevMan 5.3 [26], a desktop version of Review Manager software used for Cochrane intervention and flexible reviews. For continuous outcomes, we calculated the mean difference (MD) as a metric; while for dichotomous variables, the risk ratio was used and visualized by forest plots. In each analysis, I² was used to measure statistical heterogeneity among studies. According to the values of P and I², the fixed-effect model (P>.10, I²<50%) or random-effects model (0<P<.10, I²≥50%) were selected [27].

A total of 5177 papers were retrieved from 7 databases. After removing 3460 duplicate records, 1360 articles were deleted due to unqualified titles and abstracts. A more detailed full-text screening was conducted on the remaining 357 articles, and 345 of them were excluded for not meeting the established criteria. As a result, the final systematic review included 12 meta-analysis studies [28-39]. The detailed research screening process and results are shown in Figure 1.

The research includes trials conducted in 5 countries: China [28-34,36], the United States [35], India [37], the United Kingdom [38], and Italy [39]. These trials were published between 2015 and 2023. The sample size ranged from 64 to 285 participants, and the intervention duration varied from 1 to 12 months. Among the 12 studies, 6 used Bluetooth technology to achieve automatic upload of blood glucose values [28,29,32,34,38,39], one study adopted an implantable blood glucose sensor for automatic data transmission [33], and another 5 studies used local area network technology to automatically upload blood glucose measurement values [30,31,35-37]. Multimedia Appendix 2 [28-39] presents the author, year, country, sample size, equipment used by the intervention, intervening measure, control measure, duration, and outcomes.

Based on the Cochrane criteria, a risk-of-bias assessment is presented in Figures 2 and 3 [28-39]. A total of 12 studies were included, all of which were assigned a literature quality grade of B [28-39]. Randomized methods were reported in detail in 9 (75%) of the 12 studies [28-30,34-39]. The remaining 3 studies did not provide detailed randomization methods. In terms of allocation concealment, 5 studies explicitly mentioned the methods used [30,32,35-37]. In all studies, participants were not blinded due to the specificity of the intervention method.

Most of the points in the distribution of HbA_1c, an outcome measure, were clustered within the funnel plot; however, the distribution was not entirely symmetrical, suggesting the possibility of publication bias in the results for HbA_1c in digital diabetes management (refer to Figure 8). We also performed a sensitivity analysis to assess how the results varied when individual studies were excluded and to evaluate the stability of the findings. When we removed a study from the model and recalculated the pooled estimates for the remaining studies, heterogeneity either disappeared or decreased, and the pooled results remained consistent, indicating the robustness of our findings.

Subgroup analysis of BMI based on the length of the intervention revealed that digital diabetes management for 3 months or less resulted in more significant improvements in BMI (refer to Table 1).

This meta-analysis evaluated the effectiveness of digital diabetes management technology based on digital SMBG for home-based patients with T2DM. The findings demonstrated significant improvements in HbA_1c, FBG, PBG, and BMI. Key features of effective digital health interventions include frequent patient engagement in SMBG and timely, personalized guidance provided by specialized HCPs.

Studies have shown that while the compliance rate for blood glucose control among patients with T2DM during hospitalization can reach as high as 80%, this rate significantly decreases by 25% to 30% once they are discharged and transition to home self-management [40,41]. Consequently, there is an urgent need to support home-based patients with T2DM in managing blood glucose fluctuations [42]. This study demonstrates that compared with traditional home blood glucose management, digital diabetes management technologies can significantly improve blood glucose levels in home-based patients with T2DM. Twelve studies [28-39] reported improvements in HbA_1c levels among home-based patients with T2DM following the use of digital diabetes management technologies, 7 studies reported that digital diabetes management technologies effectively improve FBG levels [28-34], and 5 studies examined the impact on PBG levels [29-33]. In conclusion, digital diabetes management technologies play a pivotal role in blood glucose management for home-based patients with T2DM and should be widely implemented in clinical practice. Ongoing focus on blood glucose monitoring is essential for maintaining stable blood sugar levels in these patients [43].

At the same time, the meta-analysis results of this study suggest that digital diabetes management technologies can reduce the BMI of home-based patients with T2DM, consistent with the findings of a systematic review by Marcolino et al [44]. However, the heterogeneity in this study is relatively high. Sensitivity analysis, conducted by excluding studies one at a time, indicated that the variability may be due to differences in intervention durations. Subgroup analysis based on intervention duration showed that when the intervention lasted less than 3 months, it effectively improved BMI. In contrast, when the intervention duration exceeded 6 months, no similar improvement was observed. Given the relatively small sample size, future large-scale, double-blind randomized controlled trials are needed to further validate the impact of digital diabetes management technologies on BMI.

With continuous innovation in sensor, microelectronics, and data analysis technologies, researchers have been actively investing in research and development. Meanwhile, the market’s strong demand for convenient and efficient blood glucose monitoring has spurred the rapid development of continuous glucose monitoring (CGM) technology. Medical staff assist patients in accessing blood glucose data via supporting mobile apps to formulate treatment plans. CGM technology offers continuous, dynamic blood glucose data, enabling a comprehensive grasp of blood glucose fluctuations, allowing timely detection of asymptomatic hypoglycemia and hyperglycemia, and contributing to optimized treatment plans and prevention of diabetes complications.

Nonetheless, CGM technology has limitations. Its equipment is costly, making long-term use unaffordable for some patients. In addition, frequent blood glucose checks may cause psychological stress in some, negatively affecting blood glucose control. Therefore, SMBG continues to play an indispensable role in diabetes management. SMBG offers patients immediate blood glucose readings at specific time points, enabling prompt adjustments to diet, exercise, or insulin dosage to maintain glucose levels within a safe range and prevent hypoglycemia or hyperglycemia. Furthermore, compared to CGM, SMBG devices are simpler, more affordable, and thus more accessible to a broader patient population, especially those with limited financial resources or relatively stable glucose fluctuations. SMBG also promotes active patient participation, increasing their awareness of disease management and enabling them to better understand the impact of daily factors on blood glucose levels. This process fosters improved lifestyle adjustments, enhances confidence in glucose control, and strengthens compliance with self-management regimens.

Among the 12 included randomized controlled trials, only 2 studies evaluated the impact of SMBG frequency on blood glucose levels [28,36], so a meta-analysis was not conducted. Both studies demonstrated that more frequent home blood glucose monitoring improved blood glucose control in home-based patients with T2DM. We observed that the frequency of SMBG in the intervention group was significantly higher than in the control group, and the digital diabetes management technology notably enhanced blood glucose control. Tomah et al [45] reported that higher SMBG frequency correlated with greater improvements in HbA_1c, suggesting that the frequency of measurements may directly influence clinically relevant outcomes, with 4 to 8 measurements per day being the most effective. Similarly, Mannucci et al [46] found that SMBG was more effective when a structured monitoring plan was implemented, and HCPs used the collected data. None of the 12 randomized controlled trials included in this study considered the frequency of contact with HCPs as an outcome measure to assess improvements in blood glucose levels. As a result, a meta-analysis could not be conducted. However, an analysis of the intervention strategies in these studies revealed that the frequency of contact with HCPs was higher in the intervention group than in the control group. In addition, a systematic review found that SMBG without support from HCPs did not yield significant clinical benefits [47]. This lack of benefit may be attributed to treatment inertia, where poor adherence to lifestyle changes and prescribed medications is common among patients with T2DM [48,49]. When feedback from HCPs is minimal, patients may feel anxious and uncertain when faced with poor SMBG control. In conclusion, digital diabetes management technologies, along with SMBG supported by HCPs, can significantly improve blood glucose control in home-based patients with T2DM.

This study advocates for the promotion of digital diabetes management technologies based on digital SMBG for home-based patients with T2DM. Such technologies can help eliminate transcription errors associated with manual data entry and offer increased convenience for patients, especially older adult individuals with T2DM, who may struggle with inputting blood glucose values due to poor vision or limited dexterity. This technology provides an age-friendly solution for blood glucose management. In addition, for older adult patients who live alone or have limited mobility, HCPs can remotely monitor blood glucose data, offer timely care, and quickly detect and address hypoglycemia or hyperglycemia. This improves the continuity and responsiveness of diabetes management. Furthermore, the use of digital devices in clinical practice can reduce the workload of clinical nurses and enhance overall work efficiency. Thus, it is recommended that this technology be widely implemented in both home and clinical settings to improve the safety and efficiency of diabetes management.

The results of this meta-analysis align with those of previous studies on the effectiveness of digital diabetes management technology for improving blood glucose levels in home-based patients with T2DM. However, in many previous studies, SMBG devices required manual data entry for glucose measurements, whereas this study focuses on digital SMBG technologies capable of automatic data uploading [50-56]. To date, no other meta-analysis has specifically evaluated the application of digital diabetes management technology based on automatic SMBG for home-based patients with T2DM.

In addition, the results of this study revealed that digital diabetes management technology can significantly reduce BMI levels in home-based patients with T2DM, consistent with the findings of a systematic review by Marcolino et al [44]. However, several other systematic reviews suggest that digital diabetes management technology has no significant effect on BMI improvement [57-59]. This discrepancy highlights the need for future high-quality, large-sample studies to further investigate the impact of digital diabetes management technology on BMI. Such studies should explore potential differences in results due to variations in patient populations, intervention measures, and follow-up durations. Moreover, standardizing research methodologies and evaluation criteria will enhance the comparability and reliability of findings, providing a stronger evidence base for determining whether this technology can effectively improve BMI in patients with T2DM.

This systematic review and meta-analysis has several notable strengths. First, to the best of our knowledge, it is the first meta-analysis to evaluate digital diabetes management techniques based on digital SMBG for improving blood glucose levels and BMI in home-based patients with T2DM. Second, a comprehensive search of multiple databases was conducted, with strict adherence to established methodological tools to ensure the reliability and transparency of the research process. Third, this study included a detailed subgroup analysis of BMI outcomes, confirming that the duration of intervention significantly influences the effectiveness of digital diabetes management technologies in home-based patients with T2DM. Finally, by analyzing multiple outcome indicators, this review provides a more comprehensive understanding of the role of digital health in diabetes management, offering valuable references for clinical practice.

We also acknowledge that this meta-analysis has some limitations. First, confined to Chinese and English literature, it excluded non–English and Chinese materials, failing to reflect global research. Future reviews should cover more languages. Second, most selected studies lack blinding details, risking measurement biases. Future research must focus on blinded outcome evaluation to minimize performance biases. Third, the HbA_1c funnel plot asymmetry indicates publication bias. Future studies should use multicenter, large-sample, randomized controlled trials for reliable results. Fourth, done over a year ago, the study may miss recent research due to the field’s rapid development, and lacking timeliness. Future research needs constant updates. Fifth, 8 (67%) of the 12 studies were from China, with regional limitations as China’s medical resources and telemedicine acceptance differ from elsewhere. Future studies should include more diverse regions to boost conclusion generality.

Overall, future digital health research can fill the existing gaps and improve the approach in 5 key areas. First, for the diabetes management technology based on SMBG, continuous exploration should be carried out on its advantages in blood glucose control. In particular, the impact of SMBG supported by HCPs on blood glucose should be assessed to tap more potential. Second, the intervention measures should be elaborated in detail so as to achieve the maximum degree of reproducibility and universality in different clinical contexts. Third, a comprehensive evaluation system will be constructed, covering aspects such as cost-effectiveness, patient satisfaction, and technology availability. This system is used to measure the value of intervention measures for the sustainable development of digital health in diabetes management. Fourth, researchers should be encouraged to study the differences in the efficacy of different population subgroups or provide patient-level data, enabling future reviews to identify the patient characteristics related to treatment outcomes. Fifth, researchers should actively collect and report patients’ views on the intervention measures to improve the design basis.

This meta-analysis demonstrates that digital diabetes management technologies can effectively regulate blood glucose levels, improve glucose metabolism, and reduce BMI in patients with T2DM. However, due to the variation in intervention strategies, frequencies, and content across the included studies, future research should prioritize multicenter, large-sample, high-quality randomized controlled trials to refine and standardize these interventions. Notably, effective digital health interventions share a common characteristic: frequent patient engagement in SMBG, supported by HCPs who provide timely and personalized guidance.