Effectiveness of continuous glucose monitoring systems on glycemic control in adults with type 1 diabetes: A systematic review and meta-analysis

The DPLG system evaluated by Benhamou et al. (2019) achieved the largest improvement in TIR (+14.5%, P<0.001).

“The DPLG system evaluated by Benhamou et al. (2019) achieved the largest improvement in TIR (+14.5 %, P-value<0.001), followed by similar benefits with other closed-loop systems.”

• cites Benhamou et al. (2019) sourceDisputed

  The Results report a 9.2% improvement in time in range for DBLG1 versus sensor-assisted pump therapy (p<0.0001), not 14.5%. The Discussion describes this improvement as similar to those in previous trials, rather than the largest among closed-loop systems. (LLM, conf=0.88)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

Benhamou et al. (2019) reported an 8.3% reduction in CV (P < 0.001).

“The study by Benhamou et al. (2019) reported an 8.3 % reduction in CV (P-value<0.001)”

• cites Benhamou et al. (2019) sourceCould not verify

  The Methods list glucose coefficient of variation (CV) as a secondary outcome, but the passages available from our copy do not report an 8.3% reduction in CV or its P-value. The Results report a 9.2% increase in time in the target glucose range, which is a different outcome. I searched for CV, coefficient of variation, and the claimed 8.3% figure. (LLM, conf=0.74)

  Experimental — accuracy not yet measured

Visser et al. (2021) quantified an almost 21% reduction in hypoglycemia fear scores with rtCGM compared to isCGM (P = 0.003).

“The study by Visser et al. (2021) quantified an almost 21 % reduction in hypoglycemia fear scores with rtCGM compared to isCGM (P-value = 0.003).”

• cites Visser et al. (2021) sourcePartly supported

  The Results and Table 2 (Additional information; page unavailable) support lower hypoglycaemia-worry scores with rtCGM than isCGM at six months. They report scores of 15.4 versus 18.0, an adjusted difference of −2.62 points and p=0.0071—not an almost 21% reduction with p=0.003. (LLM, conf=0.91)

  Experimental — accuracy not yet measured

Studies evaluating closed-loop systems (Choudhary et al., 2022; Matejko et al., 2022) reported significant improvements in quality of life measures.

“the studies evaluating closed-loop systems (Choudhary et al., 2022, Matejko et al., 2022) reported significant improvements in quality of life measures.”

• cites Choudhary et al., 2022 sourceNot checked — no full text of the source
• cites Matejko et al., 2022 sourceSupported

Garg et al. (2011) found an 8.9% reduction in MAGE (P < 0.01).

“while Garg et al. (2011) found an 8.9 % reduction in MAGE (P-value<0.01).”

• cites Garg et al. (2011) sourcePartly supported

  The Results (Glycemic variability indexes) report that MAGE fell during the six-month unblinded period. Table 2 (Results, Glycemic excursions; page not available) gives P < 0.001 for the blinded-versus-unblinded comparison, supporting P < 0.01, but it does not report an 8.9% reduction; the displayed MAGE values do not yield that percentage for either group. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

Glycemic variability is linked to oxidative stress, endothelial dysfunction, and cardiovascular outcomes independent of mean glucose levels.

“new literature evidence links glycemic variability to oxidative stress, endothelial dysfunction, and cardiovascular outcomes independent of mean glucose levels [[51], [52], [53], [54]].”

• cites [51] Martinez, 2021, Glycemic variability and cardiovascular disease in patients with type 2 diabetes, BMJ Open Diabetes Res Care sourceSupported
• cites [52] Tateishi, 2022, Impact of glycemic variability on coronary and peripheral endothelial dysfunction in patients with coronary artery disease, J Cardiol sourceNot checked — no full text of the source
• cites [53], [54]Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

In the rtCGM versus isCGM meta-analysis, heterogeneity was low across all outcomes (I2 ranging from 12.9% to 23.5%).

“Heterogeneity was low across all outcomes (I2 ranging from 12.9 % to 23.5 %), suggesting consistency in the observed advantages of rtCGM over isCGM.”

In studies of automated insulin delivery systems, TIR absolute increases ranged from +4.1% to +14.5%.

“TIR showed the most observable benefits, with absolute increases ranging from +4.1 % to +14.5 % across studies.”

In direct head-to-head comparisons, rtCGM was associated with a greater reduction in time below range than isCGM (mean difference 1.03%, 95% CI 1.45 to −0.61%, P<0.001; abstract reports −1.28%).

“Real-time CGM showed advantages over intermittently scanned CGM for TIR (+5.63 %, P-value<0.001) and hypoglycemia reduction (−1.28 %, P-value<0.001).”

• Also stated in: § Comparison of rtCGM vs. isCGM technologies · boosted

  “A significant advantage was observed for hypoglycemia reduction, with rtCGM demonstrating a 1.03 % greater reduction in time below range compared to isCGM (95 % CI: 1.45 to −0.61 %, P-value<0.001).”

Among included studies, HbA1c reduction varied from no significant change to significant reductions up to −1.0%.

“The magnitude of HbA1c reduction varied across studies, ranging from no significant change to significant reductions of up to −1.0 %.”

Closed-loop and automated insulin delivery systems generally demonstrated larger HbA1c reductions (−0.5% to −1.0%) compared to standalone CGM systems.

“Studies investigating closed-loop and automated insulin delivery systems generally demonstrated larger HbA1c reductions (−0.5 % to −1.0 %) compared to standalone CGM systems, suggesting additional benefits with automated insulin delivery (Fig. 2).”

The largest TIR improvements were observed with closed-loop/automated insulin delivery systems, with increases up to 18.5%.

“The largest TIR improvements were observed with closed-loop/automated insulin delivery systems, with increases of up to 18.5 %.”

In network meta-analysis pairwise comparisons, transitioning from MDI + SMBG to SAP yielded the largest TIR improvement (+6.8%, 95% CI 4.2-9.4%).

“Pairwise comparisons showed significant benefits for each progressive technology tier, with the largest TIR improvement seen when transitioning from MDI + SMBG to SAP (+6.8 %, 95 % CI: 4.2–9.4 %), and the most significant hypoglycemia reduction observed when moving from isCGM to rtCGM (−1.28 %, 95 % CI: 1.77 to −0.79 %).”

Automated insulin delivery systems consistently demonstrated improvements in glycemic outcomes compared to standard therapy.

“These systems consistently demonstrated improvements in glycemic outcomes.”

In network meta-analysis pairwise comparisons, moving from isCGM to rtCGM yielded the most significant hypoglycemia reduction (−1.28%, 95% CI 1.77 to −0.79%).

“Pairwise comparisons showed significant benefits for each progressive technology tier, with the largest TIR improvement seen when transitioning from MDI + SMBG to SAP (+6.8 %, 95 % CI: 4.2–9.4 %), and the most significant hypoglycemia reduction observed when moving from isCGM to rtCGM (−1.28 %, 95 % CI: 1.77 to −0.79 %).”

Hypoglycemia outcomes in automated insulin delivery studies generally showed either no significant difference or modest reductions.

“Hypoglycemia outcomes generally showed either no significant difference or modest reductions with automated systems.”

In network meta-analysis, automated hybrid closed-loop systems achieved the highest overall effectiveness ranking among CGM technologies (SUCRA 0.92).

“Automated closed-loop systems achieved the highest ranking in network meta-analysis (SUCRA = 0.92).”

• Also stated in: § Network meta-analysis of CGM technologies · boosted

  “Automated hybrid closed-loop systems demonstrated the highest overall effectiveness (SUCRA value: 0.92), ranking first for HbA1c reduction (probability of being best: 0.71) and TIR (probability of being best: 0.82).”

The systematic review and meta-analysis included 27 studies with a total of 2975 participants.

“Twenty-seven studies with total of 2975 participants were included.”

• Also stated in: § Study characteristics · plain

  “Our literature search identified 27 studies including 2975 participants that met our inclusion criteria and were included in the meta-analysis”

Some studies reported reductions in hypoglycemia exposure up to 72% in time <3.0 mmol/L.

“Several studies reported significant reductions in hypoglycemia exposure, with decreases of up to 72 % in time <3.0 mmol/L.”

Four studies provided direct head-to-head comparisons of rtCGM versus isCGM.

“Direct head-to-head comparisons of rtCGM versus isCGM technologies were available from four studies (Table 4).”

Among studies reporting hypoglycemia outcomes, 20 of 24 (83%) reported reductions in time in hypoglycemia.

“For hypoglycemia outcomes, most studies (20/24, 83 %) reported reductions in TIH, despite these findings, the magnitude and statistical significance varied across studies.”

In direct head-to-head comparisons, rtCGM was associated with a higher time in range than isCGM (mean difference +5.36%, 95% CI 3.41-7.31%, P<0.001; abstract reports +5.63%).

“Real-time CGM showed advantages over intermittently scanned CGM for TIR (+5.63 %, P-value<0.001) and hypoglycemia reduction (−1.28 %, P-value<0.001).”

• Also stated in: § Comparison of rtCGM vs. isCGM technologies · plain

  “Similarly, TIR was significantly higher with rtCGM (mean difference: +5.36 %, 95 % CI: 3.41–7.31 %, P-value<0.001), equivalent to 77 additional minutes daily in the target range.”

In adults with type 1 diabetes, CGM significantly reduced HbA1c compared with standard monitoring (pooled mean difference −0.38%, 95% CI 0.49 to −0.27%, P<0.001).

“CGM significantly reduced HbA1c compared to standard monitoring (mean difference: 0.38 %, 95 % CI: 0.49 to −0.27 %).”

• Also stated in: § Effect of CGM on HbA1c · boosted

  “Among the studies comparing CGM to standard care, the pooled mean difference in HbA1c was −0.38 % (95 % CI: 0.49 to −0.27 %, P-value<0.001), indicating a clinically meaningful improvement.”

Most included studies showed statistically significant improvements in HbA1c with CGM use.

“Most of the included studies have showed statistically significant improvements in HbA1c with CGM use.”

Eight studies evaluated automated insulin delivery systems compared to standard therapy approaches.

“Eight studies evaluated automated insulin delivery systems compared to standard therapy approaches (Table 4).”

The most substantial HbA1c reduction among automated insulin delivery studies was observed with closed-loop systems compared to MDI + CGM (−1.0%, P<0.001) in Choudhary et al. (2022).

“The most substantial HbA1c reduction was observed with closed-loop systems compared to MDI + CGM (−1.0 %, P-value<0.001) in the study by Choudhary et al. (2022).”

• cites Choudhary et al. (2022) sourceNot checked — no full text of the source

Our systematic review and meta-analysis provides significant and detailed evidence that CGM technologies significantly improve glycemic control across multiple dimensions.

“Our systematic review and meta-analysis provides significant and detailed evidence that CGM technologies significantly improve glycemic control across multiple dimensions.”

Across the included studies, treatment satisfaction consistently improved.

“Fourteen studies reported patient-related outcomes, with consistent improvements in treatment satisfaction, quality of life, and reduced fear of hypoglycemia.”

Across the included studies, fear of hypoglycemia was consistently reduced.

“Fourteen studies reported patient-related outcomes, with consistent improvements in treatment satisfaction, quality of life, and reduced fear of hypoglycemia.”

Across the included studies, quality of life consistently improved.

“Fourteen studies reported patient-related outcomes, with consistent improvements in treatment satisfaction, quality of life, and reduced fear of hypoglycemia.”

Fourteen studies reported patient-related outcomes.

“Fourteen studies reported patient-related outcomes, with consistent improvements in treatment satisfaction, quality of life, and reduced fear of hypoglycemia.”

Ten studies reported improvements in glycemic variability across various metrics including CV, MAGE, and SD.

“For glycemic variability, ten studies reported improvements across various metrics including Coefficient of Variation (CV), Mean Amplitude of Glycemic Excursions (MAGE), and Standard Deviation (SD) of glucose values.”

We found significant improvements in TIR that are equivalent to nearly two additional hours daily in target range.

“We found consistent HbA1c reductions with CGM use, significant improvements in TIR that are equivalent to nearly two additional hours daily in target range, and meaningful reductions in hypoglycemia exposure across various definitions.”

We found meaningful reductions in hypoglycemia exposure across various definitions.

“We found consistent HbA1c reductions with CGM use, significant improvements in TIR that are equivalent to nearly two additional hours daily in target range, and meaningful reductions in hypoglycemia exposure across various definitions.”

We found consistent HbA1c reductions with CGM use.

“We found consistent HbA1c reductions with CGM use, significant improvements in TIR that are equivalent to nearly two additional hours daily in target range, and meaningful reductions in hypoglycemia exposure across various definitions.”

Sensitivity analyses excluding studies before 2010 showed consistent effect directions (HbA1c reduction: 0.35% vs. −0.38%).

“However, sensitivity analyses excluding studies before 2010 showed consistent effect directions as, HbA1c reduction: 0.35 % vs. −0.38 %, suggesting that technological improvements have improved magnitude rather than altered the effectiveness patterns of CGM technologies.”

The network meta-analysis demonstrated a clear hierarchy of effectiveness, with automated closed-loop systems achieving the highest overall ranking, followed by rtCGM, and isCGM.

“The network meta-analysis has demonstrated a clear hierarchy of effectiveness, with automated closed-loop systems achieving the highest overall ranking, followed by rtCGM, and isCGM.”

Automated insulin delivery systems demonstrate superior effectiveness across outcomes compared to other systems.

“Our findings on automated insulin delivery systems support previous studies findings and extend previous meta-analyses by demonstrating their superior effectiveness across outcomes.”

Network meta-analysis demonstrated a clear effectiveness hierarchy: automated closed-loop systems and rtCGM have superior benefits compared to isCGM and standard monitoring.

“Our network meta-analysis has demonstrated a clear effectiveness hierarchy, with automated closed-loop systems and rtCGM demonstrating superior benefits compared to isCGM and standard monitoring.”

CGM use resulted in increases in TIR by 7.9%, equivalent to 114 minutes daily.

“CGM use resulted in meaningful reductions in HbA1c by 0.38 %, significant increases in TIR by 7.9 %, which is equivalent to 114 min daily, and consistent decreases in hypoglycemia exposure.”

CGM improves hypoglycemia fear scores.

“Improvements in hypoglycemia fear scores, treatment satisfaction, and overall quality of life highlight and address the psychosocial impact of CGM technologies.”

CGM benefits were maintained or slightly improved with longer intervention durations.

“Importantly, CGM benefits were maintained or slightly improved with longer intervention durations, suggesting sustained effectiveness over time.”

CGM use resulted in reductions in HbA1c by 0.38%.

“CGM use resulted in meaningful reductions in HbA1c by 0.38 %, significant increases in TIR by 7.9 %, which is equivalent to 114 min daily, and consistent decreases in hypoglycemia exposure.”

Subgroup analysis by duration suggested maintained or improved benefits over time.

“However, our subgroup analysis by duration suggested maintained or improved benefits over time, longer-term effectiveness data would provide greater confidence regarding sustained benefits and adherence.”

rtCGM has clear advantages for hypoglycemia reduction estimated by −1.28% in time below range.

“The clear advantages of rtCGM for hypoglycemia reduction estimated by −1.28 % in time below range may be better for certain patients with hypoglycemia unawareness or frequent hypoglycemic events.”

Statistically significant differences between subgroups by baseline HbA1c were observed.

“While we observed statistically significant differences between subgroups, individual studies within each stratum may differ in other important characteristics such as diabetes duration, prior technology experience, socioeconomic status, or healthcare system factors that could confound the observed relationships.”

The HbA1c reduction of −0.38% observed in our results represents a significant improvement.

“The HbA1c reduction of −0.38 % observed in our results represents a significant improvement that could translate to better and optimized reductions in microvascular complication risk.”

The observed difference in HbA1c between rtCGM and isCGM is 0.23% (decreased).

“However, the slighter observed differences in HbA1c estimated by decreased 0.23 %, suggest that both technologies offer better benefits over standard monitoring, with specific technology selection could be guided by individual patient factors including hypoglycemia risk, cost considerations, and personal preferences.”

Benefits were maintained or improved with longer-term use.

“Also, we identified that patients with higher baseline HbA1c derived nearly three-fold greater absolute improvements, while benefits were maintained or improved with longer-term use.”

Patients with higher baseline HbA1c derived nearly three-fold greater absolute improvements.

“Also, we identified that patients with higher baseline HbA1c derived nearly three-fold greater absolute improvements, while benefits were maintained or improved with longer-term use.”

The improvement in TIR of 7.9% translated into 114 min daily.

“The improvement in TIR of 7.9 % translated into 114 min daily, exceeds the threshold of 5 % considered clinically significant in recent consensus guidelines.”

CGM reduces glycemic variability.

“Our assessment of glycemic variability, despite its limitation by the heterogeneous measurement strategies has consistently demonstrated reductions.”

CGM use resulted in consistent decreases in hypoglycemia exposure.

“CGM use resulted in meaningful reductions in HbA1c by 0.38 %, significant increases in TIR by 7.9 %, which is equivalent to 114 min daily, and consistent decreases in hypoglycemia exposure.”

TIR improvements with automated insulin delivery systems were up to 15%.

“The TIR improvements which were up to increased 15 % estimate, with these systems represent an important consideration for us in diabetes management, approaching glucose profiles of individuals without diabetes [[45], [46], [47]].”

CGM improves treatment satisfaction.

“Improvements in hypoglycemia fear scores, treatment satisfaction, and overall quality of life highlight and address the psychosocial impact of CGM technologies.”

Subgroup analyses demonstrated a dose-response relationship between baseline HbA1c and treatment effect, with patients having poorer initial control deriving the greatest absolute benefits.

“Subgroup analyses demonstrated a dose-response relationship between baseline HbA1c and treatment effect, with patients having poorer initial control deriving the greatest absolute benefits.”

Baseline HbA1c is related to CGM treatment effect.

“The identified relationship between baseline HbA1c and treatment effect has important implications for resource allocation and patient selection.”

CGM improves overall quality of life.

“Improvements in hypoglycemia fear scores, treatment satisfaction, and overall quality of life highlight and address the psychosocial impact of CGM technologies.”

Patients with baseline HbA1c over 8.5% achieved nearly three times the absolute HbA1c reduction (0.68%) compared to those with baseline HbA1c less than 7.5% (0.24%).

“The finding that patients with higher baseline HbA1c values over 8.5 % have achieved nearly three times the absolute HbA1c reduction by 0.68 % compared to those with good baseline control who are less than 7.5 % by estimated 0.24 %”

Glycemic management of T1DM has increasingly focused beyond HbA1c alone to include metrics such as TIR, hypoglycemia exposure, and glycemic variability.

“The glycemic management of T1DM has increasingly focused beyond HbA1c alone to include metrics such as TIR, hypoglycemia exposure, and glycemic variability, which reflect day-to-day glucose control and its impact on patients' lives.”

• cites [42] Maiorino, 2020, Effects of continuous glucose monitoring on metrics of glycemic control in diabetes: a systematic review with meta-analysis of randomized controlled trials sourceNot checked — no full text of the source
• cites [43] Rizos, 2025, Continuous glucose monitoring in type 1 diabetes, type 2 diabetes, and diabetes during pregnancy sourceNot checked — no full text of the source

Funnel plot analysis for HbA1c outcomes suggested minimal publication bias.

“Assessment of publication bias for HbA1c outcomes using funnel plot analysis suggested minimal bias (Fig. 3).”

Egger's test was not statistically significant (P = 0.078).

“The funnel plot was reasonably symmetrical, and Egger's test was not statistically significant resulting with P-value = 0.078.”

The most common source of bias was performance bias.

“The most common source of bias was performance bias, that was present in studies due to the difficulty of blinding participants to visibly different glucose monitoring devices.”

For patient-reported outcomes and glycemic variability, evidence quality was very low, with limitations across multiple GRADE domains including inconsistent measurement tools, indirectness, and imprecision.

“For patient-reported outcomes and glycemic variability, evidence quality was very low (⊕◯◯◯), with limitations across multiple GRADE domains including inconsistent measurement tools, indirectness, and imprecision.”

Technological improvements have improved the magnitude rather than altered the effectiveness patterns of CGM technologies.

“However, sensitivity analyses excluding studies before 2010 showed consistent effect directions as, HbA1c reduction: 0.35 % vs. −0.38 %, suggesting that technological improvements have improved magnitude rather than altered the effectiveness patterns of CGM technologies.”

No formal multiple testing corrections were applied despite multiple outcomes and comparisons.

“Given the multiple outcomes and comparisons analyzed, we acknowledge the possible risk for Type I error inflation. However, we did not apply formal multiple testing corrections as our primary outcomes (HbA1c, TIR, and hypoglycemia) represent clinically distinct and established endpoints in diabetes care.”

• AssumptionRests on an assumption: HbA1c, TIR, and hypoglycemia represent clinically distinct and established endpoints in diabetes care.

For hypoglycemia reduction, evidence quality was low due to risk of bias and inconsistency in hypoglycemia definitions across studies.

“For hypoglycemia reduction, evidence quality was low (⊕⊕◯◯) due to both risk of bias and inconsistency in hypoglycemia definitions across studies.”

The small adjustment from trim-and-fill suggests that possible publication bias had minimal impact on the overall findings.

“This small adjustment suggests that possible publication bias had minimal impact on our overall findings, providing additional confidence in the significance of the meta-analysis results.”

GRADE assessment demonstrated variable quality of evidence across outcomes.

“GRADE assessment demonstrated variable quality of evidence across outcomes (Supplementary Table 2).”

Evidence for automated systems was rated moderate quality.

“Evidence for automated systems was rated as moderate quality (⊕⊕⊕◯).”

Observational studies are susceptible to selection bias, confounding by indication, and unmeasured covariates that could significantly impact glycemic outcomes.

“While observational studies may offer larger sample sizes and real-world applicability, they are susceptible to selection bias, confounding by indication, and unmeasured covariates that could significantly impact glycemic outcomes.”

Despite several RCTs demonstrating benefits of various CGM technologies, uncertainty has persisted regarding their comparative effectiveness across different outcomes, patient subgroups, and technology types.

“Despite the several RCTs demonstrating benefits of various CGM technologies, uncertainty has persisted regarding their comparative effectiveness across different outcomes, patient subgroups, and technology types [40,41].”

• cites [40] Dicembrini, 2021 sourceNot checked — no full text of the source
• cites [41] Langendam, 2012, Continuous glucose monitoring systems for type 1 diabetes mellitus sourceNot checked — no full text of the source

Most studies had moderate overall risk of bias.

“Risk of bias assessment revealed that most studies had moderate overall risk of bias, (Supplementary Table 1).”

Selection bias was low in most studies.

“Selection bias was low in most studies, indicating appropriate randomization procedures”

Performance bias was present due to difficulty blinding participants to visibly different glucose monitoring devices.

“The most common source of bias was performance bias, that was present in studies due to the difficulty of blinding participants to visibly different glucose monitoring devices.”

The funnel plot was reasonably symmetrical.

“The funnel plot was reasonably symmetrical, and Egger's test was not statistically significant resulting with P-value = 0.078.”

Attrition bias was predominantly low.

“Attrition bias was also predominantly low, indicating good follow-up and data completeness.”

Detection bias was generally low.

“Detection bias was generally low, suggesting reliable outcome assessment despite difficulties with blinding.”

Trim-and-fill method identified three potentially missing studies.

“Trim-and-fill method has identified three potentially missing studies, resulting in an adjusted pooled effect of −0.32 % (95 % CI: 0.45 to −0.19 %) compared to the original estimate of −0.38 % (95 % CI: 0.49 to −0.27 %).”

Evidence for rtCGM versus isCGM comparisons was rated low quality, limited by risk of bias and imprecision due to the small number of direct comparison studies.

“The evidence for rtCGM versus isCGM comparisons was also rated as low quality (⊕⊕◯◯), limited by risk of bias and imprecision due to the small number of direct comparison studies.”

Inability to blind participants to visibly different glucose monitoring devices resulted in performance bias in most included studies.

“First, the inability to blind participants to visibly different glucose monitoring devices resulted in performance bias in most included studies.”

CGM technologies meaningfully improve glycemic control across multiple dimensions in adults with T1D.

“This systematic review and meta-analysis provides significant evidence that CGM technologies meaningfully improve glycemic control across multiple dimensions in adults with T1D.”

The observed HbA1c reduction could translate to better reductions in microvascular complication risk.

“The HbA1c reduction of −0.38 % observed in our results represents a significant improvement that could translate to better and optimized reductions in microvascular complication risk.”

CGM effectiveness is sustained over time.

“Importantly, CGM benefits were maintained or slightly improved with longer intervention durations, suggesting sustained effectiveness over time.”

Inspection of funnel plot asymmetry and trim-and-fill method suggested minimal impact of publication bias on primary findings.

“However, the inspection of funnel plot of asymmetry and application of trim-and-fill method suggested minimal impact of publication bias on our primary findings.”

Heterogeneity likely contributed to lower quality evidence ratings for these outcomes in the GRADE assessment.

“This heterogeneity likely contributed to the lower quality evidence ratings for these outcomes in our GRADE assessment.”

rtCGM may be better for certain patients with hypoglycemia unawareness or frequent hypoglycemic events.

“The clear advantages of rtCGM for hypoglycemia reduction estimated by −1.28 % in time below range may be better for certain patients with hypoglycemia unawareness or frequent hypoglycemic events.”

Significant heterogeneity existed in outcome definitions and measurement approaches, especially for hypoglycemia and patient-reported outcomes.

“Second, significant heterogeneity existed in outcome definitions and measurement approaches, especially for hypoglycemia and patient-reported outcomes.”

The TIR improvement exceeds the threshold of 5% considered clinically significant in recent consensus guidelines.

“The improvement in TIR of 7.9 % translated into 114 min daily, exceeds the threshold of 5 % considered clinically significant in recent consensus guidelines.”

• AssumptionRests on an assumption: A TIR improvement of 5% is considered clinically significant in recent consensus guidelines.

CGM-induced reductions in glycemic variability may have implications beyond acute glycemic control.

“These improvements may have implications beyond acute glycemic control”