Abstract
Background:
To evaluate whether sustained use of continuous glucose monitoring (CGM) is associated with long-term patterns in glycemic control, safety indicators, and health care utilization among adults with type 2 diabetes (T2D) in routine clinical care.
Method:
A two-year retrospective chart review was conducted at a tertiary care center in Saudi Arabia. Adults with T2D who initiated second-generation CGM (FreeStyle Libre 2) and maintained use for 24 months were included. Glycemic, metabolic, and clinical outcomes were assessed at baseline, 12 months (T12), and 24 months (T24).
Results:
Among 222 adults (mean age 48.7 years; 48.2% female), sustained CGM use was associated with improvements across multiple glycemic parameters. Glycated hemoglobin (HbA1c) declined from 8.2% at baseline to 7.8% at T24, accompanied by reductions in mean glucose, self-monitoring of blood glucose frequency, glycemia risk indices, and glucose variability. %TIR70-180 increased, time spent in hyperglycemia decreased, and time below range remained low throughout follow-up. Directionally similar glycemic improvements were observed across subgroups, including individuals with obesity, those treated with oral agents only, and those receiving insulin-based regimens. Beyond glycemic outcomes, body weight decreased by approximately 2 kg over 24 months. Diabetes-related emergency department visits declined from five participants (2.3%) at T0 to two (0.9%) at T12 and one (0.5%) at T24.
Conclusions:
Two-year use of CGM in routine care was associated with favorable trends in glycemic control, glycemic stability, self-monitoring behavior, and health care utilization. Long-term CGM integration may be feasible and potentially beneficial across diverse T2D populations, although prospective studies are needed to clarify the causal effects.
Keywords
Introduction
Type 2 diabetes (T2D) is a prevalent and growing health challenge worldwide, necessitating effective glycemic control to prevent complications.1 -4 The Middle East bears a particularly high burden of this disease, with Saudi Arabia among the most affected countries; national projections indicate that nearly one in four adults may be living with diabetes by 2045. 5 Achieving and maintaining glycemic targets in T2D is often challenging due to the progressive nature of the disease and challenges with adherence to lifestyle and medication. 6 Traditionally, self-monitoring of blood glucose (SMBG) via fingerstick has been a cornerstone of diabetes self-management, but SMBG can be cumbersome and provides limited data, potentially contributing to suboptimal glycemic control in many patients. 7
Advances in continuous glucose monitoring (CGM) technology have opened new possibilities for optimizing diabetes management.8 -11 Continuous glucose monitoring enables patients to track interstitial glucose levels and trends; depending on the system, this is achieved either through automatic data transmission (real-time CGM [rtCGM]) or by user-initiated scanning of a sensor (intermittently scanned CGM [isCGM]).12,13
Early randomized trials of CGM in insulin-treated T2D, such as the REPLACE study, demonstrated significant reductions in hypoglycemia compared with SMBG, although they did not initially show immediate improvements in glycated hemoglobin (HbA1c).14,15 However, accumulating evidence from subsequent trials and real-world studies has demonstrated that CGM can indeed lead to significant reductions in HbA1c and overall improved glycemic control in both type 1 diabetes and T2D.16 -20 Among users of isCGM, higher scanning frequency has been associated with greater reductions in HbA1c and improvements in metrics such as time in range (%TIR70-180) and reduced time spent in hyperglycemia.21 -23
In addition to improved glycemic averages, the use of CGM has been associated with significant clinical and patient-centered benefits. Studies report that CGM adoption is associated with fewer acute diabetes events (such as hospitalizations for severe hyperglycemia or diabetic ketoacidosis) and reduced time spent in hypoglycemia.15,24,25 Users of CGM also frequently report improved treatment satisfaction and quality of life, likely due to a reduced fingerstick burden and increased confidence in glucose management.14,17
Despite broad global experience with CGM, long-term real-world evaluations in diverse T2D populations remain limited. Most prior studies have shorter follow-up durations or focus primarily on individuals using intensive insulin regimens. Evidence from the Middle East, where diabetes prevalence and complication rates are high, is also scarce. To address these gaps, we conducted a two-year retrospective study of adults with T2D who initiated CGM. The aim was to examine whether sustained CGM use over 24 months was associated with changes in glycemic control, body weight, CGM-generated glucose profiles, hypoglycemia indicators, and diabetes-related health care utilization. We hypothesized that continued CGM use in routine care would be accompanied by favorable trends across these metabolic and clinical measures in this real-world cohort.
Methods
Study Design and Setting
This retrospective chart review study was conducted at a major tertiary care hospital in Saudi Arabia. The study was designed and reported in accordance with STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. 26 All procedures adhered to the principles outlined in the Declaration of Helsinki and its subsequent amendments. 27 The study protocol was reviewed and approved by the Institutional Review Board of the hospital (approval no. [IRB# 1394]). Being a retrospective analysis of de-identified data, the need for individual informed consent was waived. No experimental intervention was applied; instead, the study leveraged real-world clinical data obtained from electronic medical records (EMRs) and device downloads.
Eligibility Criteria
Adults with a documented diagnosis of T2D and receiving active treatment (oral antihyperglycemic therapy, insulin therapy, or combination therapy) were eligible if they had initiated CGM, had used the device for no more than two months before the study evaluations, and subsequently maintained CGM use for two years with complete clinical and CGM data.
Participants were excluded if they had unstable diabetes requiring recent hospitalization, inadequate CGM recording (<70% active time), missing required outcome data, or comorbidities expected to substantially influence glucose or weight regulation (such as chronic supraphysiologic corticosteroid use, untreated thyroid disease, or end-stage renal disease).
Study Period and Index Time Points
All participants initiated personal use of the CGM system (FreeStyle Libre 2; FSL2) in 2023, with the initiation date serving as the individual index date for analysis. Baseline (T0) was defined as the laboratory value obtained closest to the CGM initiation date within a prespecified window of 90 days before device start (−90 to 0 days). The first follow-up (T12) occurred approximately 12 months after the device was initiated. Clinical assessments and CGM metrics for this interval were aggregated within a ±3-month window around the one-year mark to accommodate routine variability in outpatient scheduling. The second follow-up (T24) was defined similarly, representing approximately 24 months after initiation.
Prescription records, pharmacy dispensation logs, and clinician documentation were systematically reviewed to verify the exact date of the CGM prescription and activation and to confirm sustained device use over the study period. Any temporary discontinuation was documented and categorized as continuous use (no interruption), or approximately two- or four-week gaps per year.
Standard Clinical Care and CGM Use Protocol
All participants received standard diabetes care in accordance with the center’s clinical protocols throughout the study period. Initiation of the CGM system was part of the usual care for patients deemed likely to benefit from more intensive glucose monitoring. As part of standard practice, all patients beginning CGM received structured device training. This included instruction on sensor insertion and replacement, the use of the reader or smartphone application for scanning, interpretation of glucose data and trend arrows, review of ambulatory glucose profile reports, and recommendations for scanning frequency. Device adherence was monitored through documentation and the availability of sensor data.
Data Sources and Measurements
Clinical and laboratory information extracted from the EMR included demographics, diabetes duration, comorbidities, and complete medication profiles, including insulin use, oral antihyperglycemic agents, and glucagon-like peptide-1 receptor agonist (GLP-1 RA). Glycated hemoglobin values were obtained from standardized laboratory assays. For participants with multiple HbA1c measurements within the prespecified baseline or follow-up windows, the value closest to the index time point was used. Body weight and body mass index (BMI) were recorded at each clinical visit. The EMR was also reviewed for documented hypoglycemic events and diabetes-related emergency department (ED) visits during the year preceding baseline and during each follow-up year. The patient-reported or device-documented frequency of SMBG was recorded to assess changes in fingerstick use after initiation of CGM.
Continuous glucose monitoring data were obtained through the LibreView platform (https://www.libreview.com/) or through in-clinic FSL2 downloads. LibreView stores de-identified glucose data from FreeStyle Libre sensors and provides clinicians with standardized reports across predefined timeframes. Participants consented electronically to the de-identification and aggregation of their glucose and device-related data within the platform.
Standard CGM metrics were calculated for each assessment interval, including TIR (70-180 mg/dL), time below range (level 1: 54-69 mg/dL; level 2: <54 mg/dL), time above range (level 1: 181-250 mg/dL; level 2: >250 mg/dL), mean glucose, glucose management indicator (GMI), glucose variability (coefficient of variation), and sensor active time. The glycemia risk index (GRI), along with its hypoglycemia (Chypo) and hyperglycemia (Chyper) components, was also calculated as previously documented in the literature.28,29
Baseline and follow-up hypoglycemia data originated from different sources. Baseline hypoglycemia reflected SMBG-based or clinician-documented episodes recorded in the EMR, while follow-up values were derived from CGM-detected hypoglycemia. All data were de-identified prior to extraction, and study conduct adhered to ethical standards and local regulatory requirements.
Study Outcomes
The primary outcomes were the changes in HbA1c and body weight from baseline (T0) to each follow-up time point after initiation of CGM. Secondary outcomes were predefined to assess additional dimensions of glycemic control, metabolic status, safety, and health care utilization. Glycemic secondary outcomes included changes in mean sensor glucose, GMI, GRI and its Chypo and Chyper components, glucose variability, and CGM-derived time-in-range metrics: %TIR70-180, %TBR, and %TAR.
Safety-related secondary outcomes included the frequency of clinically documented hypoglycemic events and the number of diabetes-related ED visits. Emergency department visit counts were analyzed as raw patient counts; because observation windows differed slightly across periods, these values reflect descriptive trends rather than standardized event rates. Behavioral and adherence-related outcomes included changes in SMBG frequency and patterns of sensor discontinuation, categorized as uninterrupted use, approximately two-week cumulative interruptions per year, or approximately four-week cumulative interruptions per year.
Statistical Analysis
All analyses were performed using R (version 4.3.0; R Foundation for Statistical Computing, Vienna, Austria) and jamovi (version 2.7; The jamovi project, Sydney, Australia). Continuous variables were expressed as mean (standard deviation, SD) or median (interquartile range, IQR), and categorical variables as frequencies and percentages.
All comparisons were performed using non-parametric methods due to the distributional characteristics of metabolic and CGM-derived variables. Continuous outcomes measured across three time points (baseline T0, T12, T24) were assessed using the Friedman test, the non-parametric analogue of repeated-measures analysis of variance. When the overall Friedman test was significant, pairwise timepoint comparisons (T0 vs T12, T0 vs T24, and T12 vs T24) were conducted using the Durbin-Conover procedure. Continuous outcomes were compared across two time points using the Wilcoxon signed-rank test.
Paired categorical changes in treatment modality across T0, T12, and T24 were evaluated using the Stuart-Maxwell test for marginal homogeneity. Changes in the CGM sensor discontinuity categories were also examined using the Stuart-Maxwell test. Paired categorical changes in GLP-1 RA use were assessed using McNemar’s test.
Prespecified subgroup analyses examined whether the effects of CGM varied by baseline characteristics or device-use patterns. Participants were stratified by baseline HbA1c (<8.0% vs ≥8.0%), baseline BMI (<30 vs ≥30 kg/m2), and treatment regimen (oral therapy only vs any insulin use). Sensor-use continuity was assessed by comparing participants with uninterrupted use to those with approximately two- or four-week cumulative interruptions per year. Outcome changes were evaluated within each subgroup and compared between groups as exploratory analyses to identify potential differences. To assess the potential impact of treatment changes, particularly the use of GLP-1RA, stratified analyses were conducted comparing clinical outcomes between GLP-1RA users and non-users at baseline, year 1, and year 2.
Associations between baseline characteristics and two-year changes in metabolic and glycemic outcomes were evaluated using Spearman’s rank-order correlation coefficients. All statistical tests were two-sided, with significance defined as P < .05.
Results
Participant Characteristics
Of the 327 adults screened for CGM initiation, 105 were excluded, resulting in a final cohort of 222 participants included in all analyses (Figure 1). The mean age was 48.7 (±7.9) years, and 48.2% were female. The average duration of diabetes was 7.7 (±5.0) years, with a mean BMI of 30.2 ± 3.4 kg/m2 and a mean baseline HbA1c of 8.25 ± 0.39%. Dyslipidemia (56.8%) and hypertension (41.4%) were the most frequent comorbidities. A total of 39 participants (17.6%) received oral-only therapy, whereas 183 (82.4%) were managed with insulin-based regimens, with or without adjunctive oral agents. Estimated average glucose at baseline was 187 (±10.5) mg/dL, with participants experiencing a median of two low glucose events every 28 days (IQR: 2). Self-monitoring frequency at baseline was six checks per week (IQR: 3), and five participants (2.3%) reported at least one diabetes-related emergency visit in the preceding year. Baseline clinical and metabolic characteristics are presented in Table 1.
Selection of the study population.
Baseline Clinical and Metabolic Characteristics (N = 222).
Abbreviations: BMI, body mass index; HbA1c, glycated hemoglobin; eAG, estimated average glucose; MDI, multiple daily injections.
Twenty-six patients had hypothyroidism, two had hyperthyroidism. Values are mean ± SD unless otherwise noted.
Longitudinal Changes in Glycemic and Metabolic Outcomes
HbA1c, glucose measures, and body composition
Changes over time were observed across glycemic and anthropometric measures (Table 2). Median HbA1c decreased from 8.2% at T0 to 7.9% at T12 and 7.8% at T24 (all pairwise P < .001). Median average glucose declined from 188 mg/dL at T0 to 184 mg/dL at T12 and 178 mg/dL at T24 (P < .001). Median body weight decreased from 85 kg at T0 to 83 kg at both T12 and T24, corresponding to BMI reductions from 30.3 kg/m2 at T0 to 29.8 kg/m2 at T12 and 29.4 kg/m2 at T24 (all P < .001).
Longitudinal Changes in Glycemic Control and Glycemia Risk Indices (N = 222).
Abbreviations: BMI, body mass index; GMI, glucose management indicator; GRI, glycemia risk index.
P1 = baseline versus first year; P2 = baseline versus second year; P3 = first year versus second year. Values are median (interquartile range, IQR).
Glycemia risk indices
From T12 onward, CGM-derived risk indices changed over time (Table 2). Median Chypo and median Chyper decreased during follow-up (P < .001). Correspondingly, the median overall GRI declined from 38.7 at T12 to 36.5 at T24 (P < .001).
Continuous glucose monitoring metrics
Continuous glucose monitoring metrics demonstrated directional changes over time (Table 3; Figure 2). The median %TIR70-180 increased from 65% at T12 to 66% at T24 (P < .001). Median TBR54-69 decreased from 3% to 2%, while median %TBR<54 remained 0% (P < .001), with a modest reduction in variability (P = .011). Median %TAR181-250 declined from 27% to 25.5%, and median %TAR>250 decreased from 6.1% to 5.95% (both P < .001). Median glucose variability decreased from 32% to 30% (P < .001). Median sensor active time remained stable, with 76% at T12 and 75% at T24. The median GMI decreased from 7.9% at T12 to 7.8% at T24 (P < .001).
Continuous Glucose Monitoring Metrics Over Time.
Abbreviations: %TIR 70-180, time in range (70-180 mg/dL); %TAR 181-250, time above range (181-250 mg/dL); %TAR > 250, time above range (>250 mg/dL); %TBR 54-69, time below range (54-69 mg/dL); %TBR < 54, time below range (<54 mg/dL); CGM, continuous glucose monitoring.
Values are median (interquartile range, IQR).
The distribution of time in glucose ranges over a two-year period.
Treatment modalities and GLP-1 RA use
Treatment distributions remained generally stable over time (Table 4). Oral-only therapy represented 17.6% of participants at T0, 15.8% at T12, and 17.1% at T24. The proportion using multiple daily injections (MDI) plus oral therapy decreased from 30.2% at T0 to 25.2% at T12 and 24.3% at T24. The proportion of patients receiving basal insulin plus oral therapy increased from 33.3% to 41.9% at T12 and to 43.7% at T24. Twice-daily insulin use remained uncommon. Glucagon-like peptide-1 receptor agonist use increased from 36.0% at T0 to 41.9% at T12 (P = .002) and 43.2% at T24 (P = .001), with no significant difference between T12 and T24 (P = .450).
Treatment Modalities and GLP-1 Use Over Time (N = 222).
Abbreviations: MDI, multiple daily injections; GLP-1, glucagon-like peptide-1.
P1 = baseline versus first year; P2 = baseline vs second year; P3 = first year versus second year. Treatment modality: Stuart-Maxwell test; GLP-1 use: McNemar’s test.
Behavioral adherence, safety events, and health care utilization
Self-monitoring of blood glucose frequency declined from a median of six checks per week at T0 to 0 at T12 and T24 (P < .001). Sensor discontinuation remained low, with 95.5% maintaining uninterrupted use at T12 and 92.8% at T24 (P = .091). Emergency visits declined from five participants (2.3%) at T0 to 2 (0.9%) at T12 and 1 (0.5%) at T24. Median low glucose events were similar between T0 and T12 (2 vs 2, P = .06) and decreased at T24 (median 1, P < .001) (Table 5).
Behavioral Adherence, Safety Events, and Health care Utilization (N = 222).
Abbreviation: IQR, interquartile range.
No skin-related adverse events were identified during the two-year observation period. Specifically, no reports or documentation of contact dermatitis or other dermatologic reactions associated with sensor wear were noted in clinical records.
Subgroup Comparative Analysis
Stratification by baseline HbA1c category
Across all evaluated outcomes, only two measures differed significantly when comparing participants with a baseline HbA1c level of ≥8% versus <8% (Supplemental Figure S1). Participants with higher baseline HbA1c showed more significant changes in HbA1c (P < .001). Changes in average glucose, %TIR70-180, GRI, Chyper, and Chypo were similar between groups (all P > .50). A difference was observed for low glucose events (P = .004), with greater reductions among participants with baseline HbA1c ≥ 8%.
Stratification by obesity status (BMI ≥30 vs <30 kg/m2)
No significant differences in the magnitude of change in glycemic or metabolic outcomes were observed between participants with a BMI of ≥30 kg/m2 and those with a BMI of <30 kg/m2 (Supplemental Figure S2). Changes in HbA1c, average glucose, TIR, GRI, Chypo, and low glucose events were comparable (all P > .10).
Stratification by baseline treatment regimen (oral only vs insulin-based therapy)
Comparisons between participants on oral-only therapy and those on insulin-based regimens showed minimal differences (Supplemental Figure S3). A difference was observed only for ΔHbA1c (P = .049). Changes in low glucose events were similar (P = .31).
Stratification by CGM discontinuation
Across T12 and T24, changes in metabolic and glycemic outcomes did not differ significantly between participants with uninterrupted use and those with two- or four-week annual discontinuations (Supplemental Table S1). The changes observed were similar across groups (all P > .05).
Stratification by GLP-1 receptor agonist use
No significant differences were observed in changes in glycemic, metabolic, or CGM-derived metrics between groups across all time points. This included ΔHbA1c, average glucose, body weight, BMI, TIR, GRI, glucose variability, and hypoglycemia-related outcomes (all P > .05). A non-significant trend toward more reductions in GRI values was observed among GLP-1RA users (Supplemental Table S2).
Associations Between Baseline Characteristics and Two-Year Outcome Changes
Spearman correlation analysis identified several associations between baseline characteristics and subsequent changes (Δ) in clinical outcomes over the two-year follow-up (Supplemental Figure S4). Higher baseline HbA1c was associated with greater reductions in HbA1c (P < .001) and GMI (P < .01). A higher baseline weight was correlated with greater weight loss (P < .001). Continuous glucose monitoring use duration correlated with larger improvements in %TIR70-180 (P < .01) and simultaneous increases in HbA1c (P < .001) and GMI (P < .05). Baseline low glucose event frequency was inversely associated with its change (P < .001). No other significant correlations were observed.
Discussion
Summary of Key Findings
In this two-year real-world study of adults with T2D using CGM, we observed changes over time across multiple glycemic measures, including HbA1c, mean glucose, and CGM-derived metrics. Body weight declined modestly over follow-up. %TIR70-180 increased and time in hyperglycemia decreased, while hypoglycemia remained low and showed a slight reduction, although this pattern should be interpreted as directional given differences in measurement methods, indicating generally stable glycemic control without evident safety concerns. Reliance on fingerstick glucose monitoring decreased, suggesting successful integration of CGM into routine self-management. Glycemic changes were broadly consistent across subgroups defined by obesity, baseline HbA1c, and treatment regimen, with larger HbA1c reductions among individuals with higher baseline values. Overall, these findings suggest favorable glycemic trends associated with long-term CGM use in routine care settings.
Interpretation and Comparison With Previous Studies
The approximately 0.4% median reduction in HbA1c over two years is consistent with prior evidence.16 -20,30 A 2024 meta-analysis of 12 randomized controlled trials (RCTs) reported a mean HbA1c reduction of approximately 0.3% with CGM versus SMBG, along with increased TIR and reduced hyperglycemia, 31 mirroring patterns observed in our cohort. These findings reflect within-cohort changes over time and should not be interpreted as a direct effect of CGM. They extend prior evidence by showing sustained glycemic improvements beyond the shorter follow-up periods of earlier trials.
For example, the REPLACE study evaluated outcomes over 6 to 12 months and did not initially demonstrate HbA1c differences between CGM and SMBG, despite reductions in hypoglycemia and improved treatment satisfaction.14,15 The MOBILE trial reported greater HbA1c reductions and substantial increases in %TIR70-180 with rtCGM. 30 More recently, the 12-month Steno2tech RCT in adults with inadequately controlled, insulin-treated T2D demonstrated a significantly greater between-group reduction in HbA1c, favoring CGM over SMBG (−0.9%, 95% confidence interval [CI] −1.4 to −0.3), despite both groups receiving structured diabetes self-management education and usual specialist care. 32
In the same trial, rtCGM use was associated with a clinically meaningful 12-month between-group increase in %TIR70-180 of 15.2% (95% CI 4.6-25.9), with glycemic outcomes assessed using blinded rtCGM at both baseline and follow-up, supporting the durability of CGM-associated improvements in glucose profiles. 32 Consistent with these findings, the 2Go-CGM RCT reported a significant within-group increase in %TIR70-180 from 37% to 53% over 12 weeks among participants randomized to rtCGM, while no significant change was observed in the SMBG group. 33 Although the baseline-adjusted between-group difference in %TIR70-180 in 2Go-CGM did not reach conventional statistical significance, both groups experienced improvements in HbA1c, likely reflecting the supportive treat-to-target care model implemented in the trial. 33
Although several CGM-derived metrics were statistically significant, the absolute changes were modest and unlikely to be clinically meaningful at the individual level. These findings underscore the distinction between statistical significance and clinical relevance, particularly in large data sets with low within-individual variability. Accordingly, these changes likely reflect stability of glycemic control rather than substantial treatment effects. In T2D, where control often deteriorates over time, the maintenance of HbA1c reduction and stable TIR over 24 months may be clinically meaningful at the population level, although not attributable to CGM alone.
The modest weight reduction of approximately 2 kg warrants contextualization, as CGM use alone has not generally been associated with significant weight change. 34 The observed weight loss likely reflects concurrent therapeutic changes and clinical care adjustments during follow-up. These pharmacologic changes may have contributed to changes in HbA1c, mean glucose, and body weight. 35 In addition, CGM-derived feedback may have supported behavioral modifications, as prior studies suggest that CGM-informed self-management can facilitate changes in dietary and physical activity. Thus, the metabolic improvements observed are likely multifactorial and should not be attributed to CGM use alone.36 -38 Beyond glycemic outcomes, the Steno2tech trial provides robust contemporary RCT evidence that CGM use may be associated with reductions in treatment intensity, reporting a −10.6 units/day (95% CI −19.9 to −1.3) decrease in total daily insulin dose and a −3.3 kg (95% CI −5.5 to −1.1) reduction in body weight compared with SMBG over 12 months, alongside an improvement in BMI of −1.1 kg/m2. 32
Hypoglycemic events decreased over two years despite overall improvements in glycemic indices, consistent with prior literature. Earlier studies reported reductions in hypoglycemia even without significant HbA1c changes, likely due to improved awareness of glucose trends and earlier corrective actions.14,30,39
Continuous glucose monitoring alerts for downward glucose trajectories may have contributed to avoidance of low glucose events, although causal inference cannot be established. Because baseline hypoglycemia was assessed using SMBG or clinician documentation, whereas follow-up was CGM-derived, comparisons should be interpreted as directional rather than quantitative estimates of change. This methodological difference limits direct numerical comparisons. Accordingly, strong conclusions regarding hypoglycemia reduction or safety cannot be drawn. The most conservative interpretation is that long-term CGM use was not associated with an increase in CGM-detected hypoglycemia and appeared compatible with stable or slightly improved safety profiles, although precise effect sizes cannot be inferred.
Similar interpretive caution applies to diabetes-related ED visits. The numerical reduction may reflect variability in ascertainment rather than a true change in event frequency. Nevertheless, the overall direction aligns with prior evidence suggesting that CGM use may reduce acute glycemic events in some settings.15,24,25,39 The 2Go-CGM study reported no severe hypoglycemia or diabetic ketoacidosis events in either arm, further supporting the safety of rtCGM in high-risk, insulin-treated populations. 33
Subgroup analyses provided exploratory insights but were limited by small sample sizes in certain strata, including participants treated with oral agents only and those with brief sensor discontinuations. Limited statistical power increases the risk of type II error, and non-significant p-values should not be interpreted as equivalence. Accordingly, subgroup findings should be considered descriptive and hypothesis-generating rather than confirmatory in nature.
Within these constraints, patterns of change were broadly similar across baseline HbA1c, BMI, medication regimen, and sensor-use categories, consistent with prior reports suggesting that CGM benefits are applicable across diverse T2D phenotypes. 40 In our cohort, 17.6% of participants were treated with oral agents only, with outcomes that were directionally comparable to those of participants treated with insulin. This aligns with trials in non-insulin-treated T2D populations 31,41 and with a meta-analysis by Jancev et al, 31 which reported similar HbA1c reductions with CGM among individuals treated with oral agents alone and those using insulin.
Additionally, although GLP-1RA use increased during follow-up, stratified analyses did not demonstrate significant differences in the magnitude of glycemic, metabolic, or CGM-derived changes between GLP-1RA users and non-users. These findings suggest that observed changes were not solely explained by pharmacologic intensification, although residual confounding cannot be excluded. Continuous glucose monitoring use may facilitate behavioral modifications, enabling patients to adjust dietary intake, physical activity, and medication adherence in real time; however, this mechanism cannot be directly assessed in the present study. A non-significant trend toward greater improvement in GRI among GLP-1RA users was noted, which may warrant further investigation in adequately powered studies.
Participants with higher baseline HbA1c levels experienced larger absolute reductions, as expected, while those closer to glycemic targets maintained or modestly improved control without an increase in hypoglycemia. Conducting this study in a Middle Eastern population, where diabetes prevalence and complications are high, adds region-specific real-world data. Observed patterns were directionally consistent with those of Western and Asian cohorts, suggesting that similar trends may be observed across health care contexts, although causality cannot be inferred. Adherence may reflect engagement with CGM technology supported by structured clinical care, although generalizability remains uncertain. These findings complement international guideline recommendations by contributing real-world evidence from a high-prevalence region. 42
Clinical Implications
These findings have several implications for clinical practice. Continuous glucose monitoring may be considered as part of routine management for adults with T2D receiving intensive therapies; however, the observed glycemic changes should be interpreted within the context of concurrent treatment and care modifications. Rather than serving solely as a short-term intervention, CGM may function as a long-term self-management support tool, providing continuous glucose information that enables timely responses to daily glucose patterns and facilitates therapeutic adjustments. 43
For individuals at higher risk, such as those with marked glycemic variability or impaired hypoglycemia awareness, CGM alarms may offer additional situational awareness, although their specific contribution cannot be isolated. Health systems may consider CGM as a potential strategy to reduce diabetes-related emergency visits, although this requires confirmation in controlled studies. The marked reduction in fingerstick testing, coupled with sustained adherence, suggests long-term feasibility for many users. Near-real-time glucose data may enhance confidence in daily monitoring and support healthier behavioral choices. The existing literature suggests potential benefits for diabetes-related distress, underscoring the importance of structured education and follow-up.
For clinicians, these findings highlight the value of integrating CGM metrics alongside HbA1c in routine care. Measures such as TIR, TBR, and glucose variability may inform more individualized therapy adjustments, including basal insulin titration, refinement of mealtime strategies, and targeted lifestyle counseling. Incorporating CGM-derived targets, such as maintaining a TIR of greater than 70% with minimal time below range, into personalized management plans may support a more proactive and data-informed approach to T2D care within a comprehensive treatment framework rather than as an isolated intervention.
Limitations
This study has several limitations. First, the retrospective observational design without a parallel control group limits causal inference. Observed changes in glycemic outcomes, body weight, and health care utilization may reflect contemporaneous therapeutic adjustments rather than the use of CGM alone, and residual or unmeasured confounding cannot be excluded, despite within-participant comparisons. Although subgroup analyses did not demonstrate significant differences between GLP-1 receptor agonist users and non-users, the study was not specifically designed or powered to isolate treatment effects; therefore, the potential influence of evolving pharmacologic therapy, including GLP-1RA use and insulin regimen changes, cannot be fully excluded.
Second, hypoglycemia at baseline and follow-up was assessed using different measurement modalities, requiring interpretation of hypoglycemia findings as directional rather than as directly comparable incidence rates. Third, selection bias is possible because the analysis included only individuals who sustained CGM use for approximately 24 months with sufficient data, representing an adherent subgroup that may not reflect broader real-world persistence. Fourth, although changes in %TIR70-180, %TBR, glucose variability, and glycemia risk indices reached statistical significance, absolute effect sizes were modest. Given the large sample size and low within-individual variability characteristic of CGM data, some p-values may overstate clinical relevance. Finally, outcomes beyond 24 months, patient-reported experiences, and long-term diabetes complications were not assessed. Prospective or randomized studies with standardized observation windows and adjustment for evolving therapies are needed to better isolate the independent contribution of CGM to long-term outcomes.
Future Research Directions
Future research should further clarify the long-term clinical impact of CGM in T2D. Longer-duration controlled studies may determine whether sustained CGM use influences diabetes-related complication rates and whether observed glycemic trends persist over extended periods. Comparative evaluations may help identify which CGM features provide incremental benefit across different clinical profiles, while inclusion of more diverse T2D populations will improve generalizability. Continuous glucose monitoring should also be explored as a behavioral support tool for dietary adherence, physical activity, and early self-management skills, while accounting for concurrent therapeutic and care-related factors.
In parallel, cost-effectiveness analyses across varied health care systems, particularly in high-prevalence regions, are needed to inform resource allocation and access decisions. Additional work is warranted to optimize the integration of CGM data with digital health technologies. Combining CGM with mobile coaching platforms, decision-support algorithms, or automated trend interpretation may enhance personalization while reducing clinician burden. Further studies should examine how pattern reports, trend arrows, and risk indices, such as GRI, can guide timely treatment adjustments. Patient-centered research is also necessary to identify the drivers of sustained CGM use, including comfort, alarm burden, data fatigue, and usability, as well as to evaluate patient-reported outcomes such as satisfaction, quality of life, confidence, and perceived safety. As sensor technologies continue to evolve, including longer-wear sensors, implantable devices, and automated systems, ongoing evaluation will be essential to determine which innovations best support glucose stability, user experience, and long-term T2D management.
Conclusion
In this two-year real-world cohort, sustained use of CGM was associated with modest but directionally favorable changes in glycemic measures, higher %TIR70-180, and directionally stable or lower hypoglycemia detection, which should be interpreted cautiously. Users demonstrated adherence and a reduction in SMBG frequency, indicating that many individuals integrated CGM into daily self-management. Because medication changes and other elements of routine care likely contributed to the observed patterns, the findings should be viewed as reflecting the combined influence of CGM use within real-world clinical practice. Overall, the study supports the feasibility and potential utility of long-term CGM use in adults with T2D, while highlighting the need for prospective, controlled studies to clarify its independent effects on long-term metabolic and safety outcomes.
Supplemental Material
sj-docx-1-dst-10.1177_19322968261447256 – Supplemental material for Effectiveness of Continuous Glucose Monitoring on Glycemic and Metabolic Outcomes in Type 2 Diabetes
Supplemental material, sj-docx-1-dst-10.1177_19322968261447256 for Effectiveness of Continuous Glucose Monitoring on Glycemic and Metabolic Outcomes in Type 2 Diabetes by Ayman Al Hayek, Wael M. Al Zahrani, Haifa F. Alotaibi and Mohammed A. Al Dawish in Journal of Diabetes Science and Technology
Footnotes
Acknowledgements
Editorial assistance was provided by VivoSolve Ltd., United Kingdom.
Abbreviations
%TAR, time above range; %TBR, time below range; %TIR, time in range; BMI, body mass index; CGM, continuous glucose monitoring; Chyper, hyperglycemia risk; Chypo, hypoglycemia risk; FSL2, FreeStyle Libre 2; GLP-1RA, glucagon-like peptide-1 receptor agonists; GMI, glucose management indicator; GRI, glycemia risk index; GV, glycemic variability; HbA1c, glycated hemoglobin; MDI, multiple daily injections; SMBG, self-monitoring of blood glucose; STROBE, Strengthening the Reporting of Observational Studies in Epidemiology; T2D, type 2 diabetes.
Ethical Considerations
This study was conducted in accordance with the principles outlined in the Declaration of Helsinki. Approval was granted by our Ethics Committee (No. 1394)
Consent to Participate
Given the retrospective design of the study and use of de-identified data, the requirement for informed consent was waived. All procedures were conducted in accordance with relevant institutional guidelines and ethical standards.
Consent for Publication
Not applicable
Author Contributions
AA and MAA contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by AA, MAA, HFA, and WMA. All authors contributed to the writing of the first draft of the manuscript. AA revised the manuscript critically for important intellectual content. All authors approved the final version of the manuscript.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statements
All data generated or analyzed during this study are included in this article (tables, figures, and supplementary data).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
