Use of the Glycemia Risk Index at Hybrid Closed-Loop Initiation to Predict Combined International Glucose Targets at 12 Months: Results From the CIRDIA Study Group

Abstract

Background:

Hybrid closed-loop (HCL) therapy helps reaching efficacy and safety glucose targets (ESGT+) in persons with type 1 diabetes (PwT1D). We analyzed here the glycemia risk index (GRI) in PwT1D at HCL initiation (M₀) and at 12 months (M₁₂) and determined whether M₀GRI value and/or M₀GRI zone (A-B-C-D-E) could identify people reaching M₁₂ESGT+.

Methods:

This was a retrospective study. Consecutive PwT1D who started HCL in a CIRDIA center were included after written consent. Glucose parameters were manually extracted from platforms at M₀ and M₁₂. ESGT+ meant reaching time in range (TIR) > 70% and glucose management indicator < 7% and time below range (TBR)^<70 < 4% and TBR^<54< 1%. Glycemia risk index was calculated and receiver-operating characteristic (ROC) analyses were performed to study the relation between M₀GRI and M₁₂ESGT+/M₁₂ESGT–.

Results:

M₁₂ data were available for 128 PwT1D. M₀GRI predicted M₁₂ESGT mostly for low and high M₀GRI values. An M₀GRI < 41 had a 90% specificity, a 36% sensitivity, and a 74% positive predictive value for M₁₂ESGT+. Sensitivity increased to 80% but specificity dropped to 56% for M₀GRI < 61 and M₀GRI ≥ 61 had a 78% negative predictive value. All PwT1D with M₀GRI 0 to 20 (zone A) reached M₁₂ESGT+. Then, the percentage of M₁₂ESGT+ people dropped about 25% per M₀GRI zone (A-B-C-D) and to 11% for zone E.

Conclusions:

M₀GRI was significantly associated with M₁₂ESGT status but mostly when in zones A-B or D-E. Hybrid closed-loop training should focus on PwT1D with M₀GRI ≥ 41 (90% of M₁₂ESGT– persons), but reaching M₁₂ESGT+ is possible with M₀GRI in zones C-D-E (64% of M₁₂ESGT+ persons) and even D-E (20% of M₁₂ESGT+ persons).

Keywords

glucose targets glycemia risk index (GRI)hybrid closed loop type 1 diabetes

Introduction

The glycemia risk index (GRI) is a new glucose parameter providing information on the hypoglycemia and hyperglycemia risk for insulin-treated people using continuous glucose monitoring (CGM). The GRI gives more weight to hypoglycemia and to extreme glucose values, either < 54 or > 250 mg/dL. Glycemia risk index can be expressed either as a single number ranging from 0 (= minimal risk) to 100 (= maximal risk) or as a dot on a GRI grid displaying both the hypoglycemia and hyperglycemia components of the GRI.¹ The GRI grid was originally divided into five zones stratifying the hypo/hyperglycemia risk from the lowest (zone A: 0-20) to the highest (zone E: 81-100).

It was shown that GRI can detect hypoglycemia/hyperglycemia risk much better than usual glucose parameters, such as glycated hemoglobin (HbA1c): Oriot et al² observed among 216 persons with type 1 diabetes (PwT1D) and a mean HbA1c level at 6.6 ± 0.3%, that 39% had actually a GRI ≥ 41 and were classified in high-risk GRI zones C, D, or E. Besides, in 2019, an international consensus³ defined ambulatory glucose profile (AGP) targets in PwT1D, including time in range (TIR) (70-180 mg/dL) > 70%, glucose management indicator (GMI) <7%, time below range (TBR) < 70 mg/dL (TBR^<70) < 4% and TBR^<54 (< 54 mg/dL) < 1%. We have previously looked at combined efficacy and safety glucose targets (ESGT = TIR>70% and GMI <7% and TBR^<70 <4% and TBR^<54 <1%) before and after hybrid closed-loop (HCL) initiation and we have shown⁴ that a GRI threshold <26 has a high specificity (92%) and a very good negative predictive value (93%) to identify PwT1D reaching ESGT (ESGT+), suggesting that if the GRI value was automatically calculated on CGM platforms, a GRI value at or above 26 could easily target PwT1D who need further assistance to reach ESGT.

Nevertheless, there is no published data on the accuracy of the baseline GRI to predict full success (ESGT+) one year after initiation of an HCL system. In this study, we aimed to determine whether the GRI value at HCL initiation could help predicting who, among PwT1D, is going to reach ESGT+ one year later.

Methods

This was a rolling retrospective study conducted in the CIRDIA (“Inter-Regional Center for Automatized Insulin in Diabetes”), a multisite (n = 7 at the time of the study) out-of-hospital HCL initiation center. The CIRDIA setting was previously described⁴: it involves highly trained diabetologists, mostly in private practice, who had completed a university degree on HCL systems, and the structure is based on the French-speaking Diabetes Society (SFD) recommendations.⁵ Dedicated CIRDIA education material was given to each patient.

Consecutive non-pregnant adult PwT1D who had a first HCL initiation in a CIRDIA site between May 1, 2023 and March 31, 2024 were included after they had signed an informed consent. No ethic committee approval was necessary, as it was a retrospective study on an ongoing basis. AGP metrics for the 14 previous days were manually extracted from HCL platforms at HCL initiation (M₀) and 12 ± 1 months later (M₁₂), and we calculated for both timepoints (M₀ and M₁₂), the percentage of PwT1D reaching TIR > 70%, GMI < 7%, TBR^<70 < 4%, TBR^<54 < 1% and the combined criteria (ESGT+). The GRI value was calculated using an online electronic calculator⁶ for each patient at M₀ and M₁₂, and GRI grids were edited with a GRI zone allocation for each participant at both timepoints.

No difference was made between HCL systems, noting that at the time of the study, only three systems were available and reimbursed in France: Medtronic 780G with G4 sensors (“780G”), Control-IQ with Tandem TSlimX2 pump and Dexcom G6 sensors (“CIQ”), CamAPS with Ypsopump (Ypsomed) and Dexcom G6 sensors (“CamAPS”).

Statistical analyses were performed using GraphPad Prism 10, GraphPad Software, LLC. Results are expressed as mean ± standard deviation (SD) or numbers (percentages). Results across time periods were compared using Mann-Whitney (unpaired data) or Wilcoxon tests (paired data). Spearman’s correlation coefficient (r_s (95% confidence interval [CI]) was determined to correlate the GRI value at M₀ (M₀GRI) to the GRI value at M₁₂ (M₁₂GRI). Receiver–operating characteristic (ROC) analyses were used to determine the sensitivity, the specificity, and the area under the curve (AUC) of M₀ GRI values to predict the percentage (95% CI) of participants reaching ESGT+ at M₁₂ (M₁₂ESGT+). A P-value < .05 was considered as significant.

Results

Between May 1, 2023 and March 31, 2024, 145 PwT1D had HCL initiation in a CIRDIA center. Twelve-month (M₁₂) data were available for 128 PwT1D noting that 56% of the participants were included in one single center and 80% in three of the seven CIRDIA centers. Because of French regulations, all the participants had to be on insulin pump therapy for at least six months (9.2 ± 5.5 years) at the time of HCL initiation and all of them were using CGM. Twelve patients (9.4%) had been using a low glucose suspend system for more than three months at M₀. The distribution of HCL systems was: 780G (77.3%), CIQ (13.3%), and CamAPS (9.4%).

At M₁₂, the mean TIR significantly improved compared with M₀ (from 52.9 ± 16.0% to 72.7 ± 11.6%; P < .0001), the TBR^<70 decreased (from 3.0 ± 3.1% to 2.0 ± 1.8%; P = .001) while the TBR^<54 did not change significantly (from 0.4 ± 0.8% to 0.3 ± 0.7%). Overall, 56 participants (44%) reached ESGT+ criteria at M₁₂ (vs eight participants = 6% at M₀). The mean GRI value dropped from 56.4 ± 21.0 at M₀ to 30.1 ± 14.1 at M₁₂ (P < .0001) with a decrease in both the hypoglycemic component (from 2.5 ± 2.6 to 1.7 ± 1.6; P = 0.02) and the hyperglycemic component (from 30.7 ± 14.7 to 15.7 ± 8.7; P < .0001). Noteworthily, between M₀ and M₁₂, the GRI hyperglycemic component improved in 91% of the participants, whereas the GRI hypoglycemic component improved only in 49% of the patients. Table 1 shows the characteristics of participants at HCL initiation (M₀) for the entire population and according to whether or not they reached ESGT criteria at M₁₂ (M₁₂ESGT+ or M₁₂ESGT–).

Table 1.

Characteristics of the Population at Hybrid Closed-Loop Initiation (M₀) According to Efficacy and Safety Glucose Targets (ESGT) Status at 12 Months (M₁₂).

Parameters at baseline	All population	ESGT+ at M₁₂	ESGT– at M₁₂	P
N	128	56 (44%)	72 (56%)
Gender (men/women)	62/66 (48%/52%)	28/28 (50%/50%)	34/38 (47%/53%)
Age (years)	42.9 ± 14.0	47.9 ± 14.2	41.7 ± 13.4	.01
Duration of T1D (years)	23.7 ± 12.1	25.1 ± 13.1	22.6 ± 11.2	.30
BMI (kg/m²)	27.5 ± 5.0	27.9 ± 4.8	27.1 ± 5.1	.32
HbA1c (%)	7.7 ± 0.9	7.4 ± 0.8	7.9 ± 0.9	.002
AGP parameters (%)
TAR^>250	17.4 ± 13.2	11.2 ±10.4	22.2 ± 13.1	< .0001
TAR^>180	44.0 ± 17.1	36.0 ± 17.0	50.3 ± 14.5	< .0001
TIR	52.9 ± 16.0	60.6 ± 15.7	46.9 ± 13.6	< .0001
TBR^<70	3.0 ± 3.1	3.4 ± 3.4	2.8 ± 2.8	.48
TBR^<54	0.4 ± 0.8	0.4 ± 0.7	0.5 ± 0.9	.61
GMI	7.7 ± 0.9	7.3 ± 0.7	8.1 ± 0.9	< .0001
GRI
Total	56.4 ± 21.0	45.9 ± 18.5	64.5 ± 19.4	< .0001
Hyperglycemia component	30.7 ± 14.7	23.6 ± 13.2	36.3 ± 13.4	< .0001
Hypoglycemia component	2.5 ± 2.6	2.8 ± 2.8	2.3 ± 2.4	.50
Persons reporting acute events in the previous 12 months	2 (1.6%)	0	2 (2.8%)
Severe hypoglycemia	2 (1.6%)	0	2 (2.8%)
Diabetic ketoacidosis	0	0	0
Diabetes-related hospitalization	0	0	0

Results are n (%) or mean ± standard deviation; P refers to the differences between ESGT+ and ESGT– populations. Abbreviations: AGP, ambulatory glucose profile; BMI, body mass index; ESGT+/–, efficacy (TIR > 70% and GMI < 7%) and safety (TBR^<70 < 4% and TBR^<54 < 1%) parameters reached or not reached at 12 months; GMI, glucose management indicator; T1D, type 1 diabetes; TAR, time above range; TBR, time below range; TIR, time in range.

M₁₂ESGT+ patients were older and had a lower baseline HbA1c level, a lower GMI and a higher TIR at M₀ than M₁₂ESGT– patients. However, the times in hypoglycemia (TBR^<70 and TBR^<54) were similar in both groups, just like the M₀GRI hypoglycemia component: M₁₂ESGT+ participants had a lower M₀GRI value mainly due to a lower hyperglycemia component. Figure 1 shows the individual distribution of the GRI level (Figure 1a) and of its hypoglycemia (Figure 1b) and hyperglycemia components (Figure 1c).

Figure 1.

Individual distribution of GRI level (a), GRI hypoglycemia component (b), GRI hyperglycemia (c) component in M₁₂ESGT+/M₁₂ESGT– participants.

The total insulin dose significantly increased between M₀ and M₁₂ in the total population and in both the ESGT+ population and the ESGT– population as the bolus dose increased, whereas the basal dose did not change significantly (Supplementary Table 1). However, there was no difference in the total, basal or bolus daily dose between ESGT+ and ESGT– participants neither at M₀ nor at M₁₂. Furthermore, there was no difference in the number of daily manual boluses or in the time spent on HCL between ESGT+ and ESGT– patients at M₁₂.

We tested the correlation between M₀GRI and M₁₂GRI (Figure 2a). The Spearman correlation coefficient r_s was 0.57 (0.44-0.68), P < .0001, with r_s² = 0.32 meaning that M₁₂GRI was correlated with M₀GRI but that only 32% of the variance in M₁₂GRI could be explained by the variation in M₀GRI. Figure 2b shows the ROC analysis of M₀GRI to predict M₁₂ESGT+. No difference was observed between the AUC of M₀GRI (0.7579) and the AUC of M₀TIR (0.7423), M₀GMI (0.7679), or M₀TAR^>180 (0.7340).

Figure 2.

M₀GRI and M₁₂GRI correlation (a) and ROC curve of M₀GRI for reaching efficacy and safety glucose targets (ESGT+) at 12 months (M₁₂) (b).

We also looked at the delta-GRI (M₁₂GRI-M₀GRI) and at the delta-TIR (M₁₂TIR-M₀TIR) that were tightly correlated (Supplementary Figure 1) (P < .0001) but there was no difference in the delta-GRI or in the delta-TIR according to the M₁₂ESGT status (Supplementary Table 2).

Then, we assessed the performance of M₀GRI zones to predict M₁₂ESGT (Table 2). We found that an M₀GRI < 41 (GRI zones A+B) had a high specificity to identify people who reached M₁₂ESGT+ since 90% (65 of 72) of M₁₂ESGT– patients had an M₀GRI value ≥ 41, but a low sensitivity since only 36% (20 of 56) of M₁₂ESGT+ participants had an M₀GRI < 41.

Table 2.

Number of Participants Reaching Efficacy and Safety Glucose Targets (ESGT) at 12 Months (M₁₂) According to the Baseline Glycemia Risk Index (GRI) Zone at Closed-Loop Initiation (M₀) and M₀GRI Zone According to the M₁₂ ESGT Status.

M₀GRI zone(GRI range)	Distribution at M₀ (n, %)(n = 128)	M₁₂ESGT+ patients (n, %) according to M₀GRI zone	M₀GRI zone in M₁₂ESGT+ patientsn (%)	M₀GRI zone in M₁₂ESGT– patientsn (%)
Zone A (0-20)	4 (3.1%)	4/4 (100%)	4/56 (7.1%)	0/72 (0%)
Zone B (21-40)	23 (18.0%)	16/23 (69.6%)	16/56 (28.6%)	7/72 (9.7%)
Zone C (41-60)	50 (39.1%)	25/50 (50.0%)	25/56 (44.6%)	25/72 (34.7%)
Zone D (61-80)	33 (25.8%)	9/33 (27.3%)	9/56 (16.1%)	24/72 (33.3%)
Zone E (81-100)	18 (14.1%)	2/18 (11.1%)	2/56 (3.6%)	16/72 (22.2%)

ESGT+/–, efficacy (TIR > 70% and GMI < 7%) and safety (TBR^<70 < 4% and TBR^<54 < 1%) parameters reached or not reached at 12 months; M₀, closed-loop initiation; M₁₂, 12 months after closed-loop initiation.

However, when we considered the three lower GRI zones (A+B+C, M₀GRI < 61), the sensitivity improved to 80% (45 of 56) but the specificity dropped to 56% (40 of 72). Figure 3 represents the M₀GRI grids according to the M₁₂ESGT status with the distribution of patients in each zone indicated in percentage.

Figure 3.

M₀GRI grids in patients reaching (ESGT+) or not reaching (ESGT–) efficacy and safety glucose targets at M₁₂.

When we considered patients in zones A and B (GRI < 41) at M₀ (n = 27, 21.1%), 20 of 27 (74%) reached ESGT+ at M₁₂ (= positive predictive value). Conversely, for patients in zones D and E (GRI ≥ 61) at M₀ (n = 51, 40%), 40 of 51 (78%) did not reach M₁₂ESGT (= negative predictive value). However, when we selected patients in GRI zone C with M₀GRI between 41 and 60 (39.1% of the whole population), 50% of them reached M₁₂ESGT+. No difference could be found between ESGT+ and ESGT– patients in zone C at M₀ (hypoglycemia or hyperglycemia component, TIR, GMI).

Discussion

The goal of this study was to determine whether baseline GRI could predict full success in reaching AGP efficacy and safety targets (TIR > 70% + GMI < 7% + TBR^<70 < 4% + TBR^<54 < 1%) one year after HCL initiation. We found that M₀GRI was significantly associated with M₁₂ESGT status but mostly for patients with baseline low GRI (zones A and B) or high GRI (zones D and E).

Our population had baseline characteristics that were comparable with the population using sensor unintegrated insulin pump therapy (n = 504) in the real-world study by Nørgaard et al⁷ who reported a mean TIR at 54.9% (vs 52.9% in our study) and a GRI at 55 (vs 56.4 in our study). And 1-year after HCL initiation, the AGP parameters in our population were very similar to the results reported by Choudhary et al⁸ in 101 629 users of the 780G system for 12 months: in this very large population, the mean TIR was 72.3% (vs 72.7% in our study) and 47.7% participants reached ESGT+ (vs 44% in our study with the same definition of ESGT+). The GRI value was not calculated in Choudhary’s study, but other studies in adults reported GRI levels on HCL therapy at 28.2,⁹ 28.6,¹⁰ or 31,⁷ that were very similar to our results (30.1 at M₁₂).

The GRI was initially created to report both the hypoglycemia and the hyperglycemia risks with one number.¹ The GRI formula was derived from the ranking of 225 CGM tracings by 330 highly experienced clinicians. Their analysis revealed that clinicians gave importance to both hyperglycemia and hypoglycemia and that GRI correlated better with clinicians’ opinions than did TIR, TBR or combined TIR and TBR, meaning that using GRI is preferable to using TIR especially in patients at high risk of hypoglycemia as the TIR is very weakly correlated with hypoglycemia.¹¹

Nevertheless, we found that M₀GRI was not better than M₀TIR to identify the PwT1D who reached ESGT+ at M₁₂. Furthermore, we did not observe any difference neither in the delta-GRI (M₁₂-M₀ GRI) nor in the delta-TIR (M₁₂-M₀ TIR) between the M₁₂ESGT+ and the M₁₂ESGT– participants. In our study, at baseline, all patients had been for more than six months under pump therapy and CGM, with a rather low hypoglycemic component (2.5 ± 2.6) that still improved after HCL (to 1.7 ± 1.6). The TBR^<70 and TBR^<54 targets were reached by, respectively, 83% and 84% of the participants at M₁₂ (data not shown) and the time in hypoglycemia was not the main reason why most of the participants did not reach ESGT+ at M₁₂. As a matter of fact, in our study, the GRI hypoglycemia component was not different at baseline between M₁₂ESGT+ and M₁₂ESGT– participants. This may explain why we could not find a superiority of M₀GRI compared with M₀TIR to predict full success in reaching ESGT+ one year after HCL initiation.

In addition to baseline GRI, other factors do play a role in the prediction of M₁₂ESGT+. We should probably consider the performance and/or the settings of the HCL system, the insulin formulation,¹² the time spent using the system, the number of meal boluses,¹³ the reinforcement of education by a highly trained team,⁵ the decrease in diabetes distress even if it remains present when using an HCL system.¹⁴ However, in our population, no difference was observed between M₁₂ESGT+ and M₁₂ESGT– participants regarding insulin dose, time spent on HCL or the number of manual daily boluses. Furthermore, the potential lower accuracy of CGM systems in hypoglycemia ranges is one of the limiting factors for using GRI in clinical practice.¹⁵

Our study has some limitations as for most of the participants, the glucose sensor was not the same at M₀ and M₁₂. This could have been a problem as performances of sensors were shown to differ.¹⁶ Nevertheless, difference could have been present both in over- or under estimation of glucose values and it was a real-life study. Furthermore, it could be argued that our GRI calculation was based on only 14-day AGP measurements; however, this was shown to be the ideal sampling duration for GRI estimation.¹⁷ We also analyzed therapeutic success at three months (M₃) and six months (M₆) after HCL initiation and observed that the M₁₂ ESGT status was already present in 87% of the participants at M₃ and in 89% of the participants at M₆ (data not shown). This suggests that M₁₂ ESGT status evaluated on a 14-day AGP measurement is a reliable timepoint to assess therapeutic success even if longer-term studies would be very informative. Finally, we cannot exclude differences in patients’ training between centers, but all the patients received the same dedicated education material, and this was a real-life study.

Conclusions

Our study is the first real-life study reporting the predictive value of baseline GRI to identify PwT1D reaching the combined ESGT one year after HCL initiation. The rate of people reaching M₁₂ESGT+ was 100% in GRI zone A and dropped about 25% per GRI zone increment from zone A to D and then to 11% in zone E. If training should focus on PwT1D in GRI zones C-D-E because they represent 90% of those who did not reach M₁₂ESGT+, motivation should remain intact because reaching M₁₂ESGT+ is possible even when M₀GRI is in zones C-D-E (64% of those who reached M₁₂ESGT+) and even in GRI zones D-E (20% of those who reached M₁₂ESGT+).

However, M₀GRI value was not superior to other AGP parameters, such as TIR or GMI, to predict full success in reaching ESGT+ after one year, mainly because TBR^<70 and TBR^<54 were already low at baseline in our population. Further studies on larger populations will be required to confirm these data but, in the meantime, it would be advisable for platform editors to program the display of GRI value and GRI zone, as it would allow an easy follow-up of improvement.

Supplemental Material

sj-docx-1-dst-10.1177_19322968251380291 – Supplemental material for Use of the Glycemia Risk Index at Hybrid Closed-Loop Initiation to Predict Combined International Glucose Targets at 12 Months: Results From the CIRDIA Study Group

Supplemental material, sj-docx-1-dst-10.1177_19322968251380291 for Use of the Glycemia Risk Index at Hybrid Closed-Loop Initiation to Predict Combined International Glucose Targets at 12 Months: Results From the CIRDIA Study Group by Sylvie Picard, Joëlle Dupont, Fabienne Amiot-Chapoutot, Blandine Courbebaisse, Estelle Personeni, Emmanuelle Lecornet-Sokol, François Mougel, Clara Bouché, Françoise Giroud, Sandrine Lablanche and Sophie Borot in Journal of Diabetes Science and Technology

Footnotes

Abbreviations

AGP, ambulatory glucose profile; AUC, area under the curve; CGM, continuous glucose monitoring; CI, confidence interval; CIRDIA, inter-regional center for automatized insulin in diabetes; ESGT, efficacy and safety glucose targets; GMI, glucose management indicator; GRI, glycemia risk index; HbA1c, glycated hemoglobin; HCL, hybrid closed loop; M₀, month 0 = baseline; M₁₂, month 12; PwT1D, persons with type 1 diabetes; ROC, receiver operating characteristic; SFD, French-speaking Diabetes Society; TAR, time above range; TBR, time below range; TIR, time in range.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: None directly related to this paper. Outside of this work, SP has received consulting and/or speaking fees from Abbott, Air Liquide, Asdia, Dexcom, Insulet, Isis Diabete, Lilly, Medtronic, Novo-Nordisk, Orkyn, Roche Diabetes Care, Sanofi, VitalAire; BC has received consulting and/or speaking fees from Abbott, ASDIA, Dexcom, Dinno Santé, Glooko, Lilly, NHC, Novo-Nordisk, Orkyn, Sanofi, VitalAire; JD has received speaking fees from Novo-Nordisk; ELS has received consulting and/or speaking fees from Abbott, Dexcom, Insulet, Lifescan, Lilly, Novo-Nordisk, Sanofi; EP has received speaking fees from Astra Zeneca and Lilly; FM has received speaking fees from Lilly, CB has received consulting and speaking fees from Dexcom and Lilly; SL has received consulting and/or speaking fees from Abbott, Insulet, Medtronic and Dexcom; and SB has received consulting and/or speaking fees from Dexcom, Insulet, Lilly, Medtrum. FAC and FG have no competing interest.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Outside of this work, the CIRDIA has been supported by a research grant from the Clement-Drevon Foundation, Dijon, France, a research grant from Santelys, Loos, France and by donations from the French Diabetic Association (Burgundy-Franche Comte and AFD21-25-70), the French Mutualite 39 and the CIRDIA received an award from the FENAREDIAM.

ORCID iDs

Sylvie Picard

Françoise Giroud

Sandrine Lablanche

Sophie Borot

Supplemental Material

Supplemental material for this article is available online.

References

Klonoff

Wang

Rodbard

, et al. A glycemia risk index (GRI) of hypoglycemia and hyperglycemia for continuous glucose monitoring validated by clinician ratings. J Diabetes Sci Technol. 2023;17(5):1226-1242.

Oriot

Hermans

Vandelaer

Philips

Prévost

. Glycemic Risk Index (GRI) variability in type 1 diabetes adults with HbA1c ≤ 7%: insights for clinical evaluation and intervention. Diabetes Metab. 2024;50(2):101520.

Battelino

Danne

Bergenstal

, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care. 2019;42(8):1593-1603.

Picard

Courbebaisse

Dupont

, et al. Glycemia risk index (GRI) and international glucose targets before and 6 months after initiation of hybrid closed loop system in the CIRDIA, a French multisite out-of-hospital center. Diabetes Metab. 2025;51(2):101617.

Bismuth

Joubert

Renard

, et al. Practical implementation of automated insulin delivery systems in 2025: a French position statement update. Diabetes Metab. 2025;51(3):101637.

https://www.diabetestechnology.org/gri/. Accessed April 20, 2025.

Nørgaard

Ranjan

Laugesen

, et al. Glucose monitoring metrics in individuals with type 1 diabetes using different treatment modalities: a real-world observational study. Diabetes Care. 2023;46:1958-1964.

Choudhary

Arrieta

van den Heuvel

Castañeda

Smaniotto

Cohen

. Celebrating the data from 100,000 real-world users of the MiniMed™ 780G system in Europe, Middle East, and Africa collected over 3 years: from data to clinical evidence. Diabetes Technol Ther. 2024;26(suppl 3):32-37.

Resmini

Zarra

Dotti

, et al. Impact on glycemia risk index and other metrics in type 1 adult patients switching to Advanced Hybrid Closed-Loop systems: a one-year real-life experience. Eur J Med Res. 2024;29:365.

10.

Rizzi

Tartaglione

Lucaccini Paoli

, et al. Evaluation of time in tight range and the glycaemia risk index in adults with type 1 diabetes using an advanced hybrid closed loop system: a 1-year real-world assessment. Diabetes Obes Metab. 2024;26(9):4078-4086.

11.

Rodbard

. Metrics to evaluate quality of glycemic control: comparison of time in target, hypoglycemic, and hyperglycemic ranges with “risk indices..” Diabetes Technol Ther. 2018;20(5):325-334.

12.

Oktavian

Kencono Wungu

Mudjanarko

Amin

. A comparison of ultra-rapid and rapid insulin in automated insulin delivery for type 1 diabetes: a systematic review and meta-analysis of randomized controlled trials. Diabetes Obes Metab. 2025;27(5):2658-2669.

13.

Bombaci

Hropot

Bratina

, et al. Daily carbohydrate entries and glycemic outcomes following automated insulin delivery system initiation in youth with type 1 diabetes: a real-world multinational longitudinal study. Diabetes Care. 2025;48:1434-1437.

14.

Reznik

Bonnemaison

Fagherazzi

, et al. The use of an automated insulin delivery system is associated with a reduction in diabetes distress and improvement in quality of life in people with type 1 diabetes. Diabetes Obes Metab. 2024;26(5):1962-1966.

15.

Oriot

Prévost

Philips

Klipper Dit Kurz

Hermans

. Glycemia risk index (GRI): a metric designed to facilitate the interpretation of continuous glucose monitoring data: a narrative review. J Endocrinol Invest. 2025;48(9):1995-2000.

16.

Eichenlaub

Waldenmaier

Wehrstedt

, et al. Performance of three continuous glucose monitoring systems in adults with type 1 diabetes J Diabetes Sci Technol. Published online February 4, 2025. doi:10.1177/19322968251315459.

17.

Shah

Sakamoto

Pyle

. Optimal sampling duration for continuous glucose monitoring for the estimation of glycemia risk index. Diabetes Technol Ther. 2023;25(2):140-142.