Time in Tight Range Versus Time in Range for Predicting Diabetic Retinopathy in Adults with Type 1 Diabetes Using Multiple Daily Insulin Injections: A Retrospective Longitudinal Study with isCGM Data

Abstract

Background:

Time in tight range (TITR, 70–140 mg/dL) has emerged as a glycemic metric offering stricter assessment than conventional time in range (TIR, 70–180 mg/dL). Whether TITR provides additional prognostic value for diabetic retinopathy (DR) in adults with type 1 diabetes (T1D) remains unclear.

Methods:

We conducted a retrospective cohort study of 309 adults with T1D on multiple daily insulin injections using intermittently scanned continuous glucose monitoring (CGM) system. Ophthalmological assessments were performed at baseline and after 12 months (May 2024–May 2025). DR incidence (in those free of DR at baseline) and progression (in those with established DR) were defined according to Early Treatment Diabetic Retinopathy Study criteria. Longitudinal TITR and TIR were extracted every 14–28 days. Multivariable logistic regression adjusted for age, sex, diabetes duration, HbA1c, hypertension, Low-density lipoprotein (LDL) cholesterol, body mass index, and smoking was applied.

Results:

At baseline, 198 participants (64.1%) had no DR, 71 (23.0%) nonproliferative, and 40 (12.9%) proliferative DR. During follow-up, 10/198 (5.1%) developed DR and 26/111 (23.4%) with baseline DR progressed. Higher TITR was independently associated with lower risk of incident DR (adjusted OR per % increase: 0.965; 95% CI: 0.950–0.980), whereas TIR was not. Receiver operating characteristic analysis confirmed superior discrimination for TITR versus TIR (area under the curve 0.580 vs. 0.430; P < 0.001). In stratified analyses, TITR predicted incident DR only among participants with HbA1c below the cohort median (7.1%). Both TITR and TIR were associated with lower risk of DR progression in models including HbA1c, with similar discriminative performance. Diabetes duration, HbA1c, hypertension, and smoking were independently associated with DR outcomes alongside CGM metrics.

Conclusions:

TITR provides modestly superior predictive value over TIR for incident DR, particularly in individuals with near-target HbA1c, but both metrics perform similarly for predicting progression. CGM-derived metrics should be interpreted in the context of overall glycemic control and clinical risk factors.

Keywords

diabetic retinopathy time in range time in tight range type 1 diabetes

Introduction

Type 1 diabetes (T1D) is a chronic autoimmune disease associated with a substantial clinical and economic burden, primarily due to its long-term complications.^1,2 Among these, diabetic retinopathy (DR) stands out as one of the most prevalent and specific microvascular manifestations of T1D. It is characterized by progressive neurovascular degeneration driven by chronic hyperglycemia.³ Proliferative DR, which represents its most advanced form, can lead to irreversible vision loss and remains one of the leading causes of preventable blindness worldwide,^4,5 while also significantly contributing to the financial pressure on healthcare systems.⁶ Long-term data from the Wisconsin Epidemiologic Study of Diabetic Retinopathy, with up to 25 years of follow-up, have shown that the majority of individuals with T1D eventually develop some degree of DR, with a substantial proportion progressing to vision-threatening stages.⁷ Although the incidence and progression of DR have declined over recent decades, likely as a result of improved glycemic control and earlier disease detection, the overall burden remains considerable.⁸

The development and progression of DR in people with T1D is influenced by a wide range of clinical and demographic factors. Established risk factors⁹ include younger age at diabetes onset,¹⁰ longer disease duration^11,12 hypertension,^13,14 dyslipidaemia,^15,16 smoking,¹⁷ obesity,^14,16 and poor glycemic control.^18–20 In this context, the advent of continuous glucose monitoring (CGM) systems has enabled the use of novel glucose metrics that offer a more nuanced characterization of glycemic exposure, opening new avenues for understanding their relationship with DR. While time in range (TIR; 70–180 mg/dL) has become the predominant metric following the publication of the International Consensus on Time in Range²¹ and has even been proposed as a valid clinical trial endpoint,²² time in tight range (TITR; 70–140 mg/dL) has gained increasing attention in recent years.^23–25 It is now considered an emerging target in the era of technological and therapeutic advances in T1D. Although TITR and TIR are highly correlated and both predict glycated haemoglobin (HbA1c) levels effectively, TITR may offer additional advantages in specific clinical contexts²⁶ and has been proposed as a useful alternative for individuals requiring stricter glycemic control.²⁷

To date, only one retrospective case–control study has longitudinally assessed the association between TITR and incident DR in individuals with T1D, based on CGM data collected annually during scheduled clinical visits.²⁸ In that study, TITR demonstrated a predictive performance for incident DR comparable with that of TIR, HbA1c, and mean glucose. In addition, some cross-sectional studies have reported a significant association between TITR and the presence of DR;²⁹ however, its prognostic value for the incidence or progression of DR has not yet been evaluated using continuous CGM data over a defined follow-up period or within a cohort design in adults treated with multiple daily insulin injections.

Against this background, the aim of the present study was to compare the predictive value of TITR versus TIR regarding the incidence and progression of DR in adults with T1D treated with multiple daily insulin injections.

Materials and Methods

Study design

We conducted a single-center, retrospective cohort study including individuals with T1D treated with multiple daily insulin injections. Eligible participants were regular users of the FreeStyle Libre 2® intermittently scanned CGM (isCGM) system (Abbott), with data available on the LibreView platform since at least May 1, 2024, and who had undergone an ophthalmological evaluation within the preceding 2 years (from May 1, 2022). From that index date, up to 22 sequential CGM data downloads (one baseline and 21 successive 14-day intervals) were analyzed over a 12-month follow-up period ending on May 1, 2025. Data about DR diagnosis and/or progression were also collected in the setting of ophthalmological evaluation during the follow-up period, starting from the index date above mentioned. Inclusion criteria were a confirmed diagnosis of T1D and regular use of isCGM. Exclusion criteria included: diagnosis of type 2 diabetes, monogenic diabetes (e.g., MODY), use of insulin pump therapy, and absence of ophthalmological evaluation during the follow-up period. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines³⁰ and was approved by the Research Ethics Committee of Hospital Universitario de La Princesa (Madrid,Spain) (Study number: 5814 01/25). The research was conducted in accordance with the Declaration of Helsinki. To reduce device- and treatment-related heterogeneity, we included only adults on multiple daily insulin injections using the same isCGM, as cross-platform differences in CGM accuracy may bias glucose metrics,^31,32 and pump users may differ systematically from multiple daily insulin injectionsusers in ways that could confound retinal outcomes.

Data collection and variables

DR was classified according to the internationally accepted criteria of the Early Treatment Diabetic Retinopathy Study (ETDRS).³³ Mild DR was defined as the presence of at least one microaneurysm. Moderate DR included microaneurysms with hard exudates, cotton wool spots or blot hemorrhages. Severe DR was defined by the presence of more than 20 intraretinal hemorrhages per quadrant, venous beading in ≥2 quadrants or prominent intraretinal microvascular abnormalities in ≥1 quadrant, in the absence of proliferative features. Proliferative DR was diagnosed in the presence of neovascularization or vitreous/preretinal hemorrhage. Incident DR was defined as no DR at baseline progressing to any DR at follow-up ophthalmological assessment. Progression of DR was defined as an increase in severity grade between the baseline and follow-up examinations. Stability or improvement was defined as no change in grade or a lower DR grade at follow-up, respectively. Glucose metrics were extracted from the LibreView platform at 14-day intervals and included: TIR (70–180 mg/dL), TITR (70–140 mg/dL), time below range (TBR, <70 mg/dL), time above range (TAR, >180 mg/dL), number of daily scans, sensor usage, coefficient of variation (CV) and standard deviation (SD). Downloads with sensor usage <70% were excluded from analysis due to data being considered not evaluable.²¹ Additional sociodemographic, clinical and laboratory data were retrieved from the electronic medical record: age, sex, diabetes duration, body mass index (BMI), smoking status, baseline HbA1c, most recent HbA1c available and total daily insulin dose. HbA1c was measured by ion-exchange high-performance liquid chromatography (ADAMS A1c HA8180 V, ARKRAY®). Total cholesterol was assessed by enzymatic methods (Alinity C Cholesterol Reagent Kit, Abbott).

Statistical analysis

Statistical analyses were performed using STATA version 17.0. Outliers were assessed through visual inspection and statistical methods. Normality of distribution was evaluated using the Kolmogorov–Smirnov test and normal probability plots. Continuous variables with suspected extreme values were winsorized at the 1st and 99th percentiles. Normally distributed continuous variables are presented as mean ± standard deviation (SD), while non-normally distributed variables are reported as median (25th–75th percentiles). Categorical variables are described as absolute and relative frequencies.

The cohort was initially stratified into three groups: no retinopathy, nonproliferative retinopathy, and proliferative retinopathy or macular edema. Baseline group comparisons were conducted using ANOVA or the Kruskal–Wallis test, depending on distribution. Comparisons of clinical and CGM-derived metrics between individuals who developed or did not develop DR, as well as between those with progression versus stability of DR, were conducted using the Wilcoxon signed-rank test. Descriptive plots were generated for mean TIR and TITR according to incident DR in participants without baseline retinopathy and according to progression versus stability among those with DR at baseline. Receiver operating characteristic (ROC) curves were generated using predicted probabilities from logistic regression models, and the corresponding areas under the curve (AUC) were estimated with standard nonparametric methods. AUCs were compared using the DeLong test.

Finally, multivariable logistic regression models for panel data were fitted to assess the association between CGM-derived metrics and the incidence or progression of DR. Models were adjusted for age, sex, mean HbA1c (2024–2025), hypertension, Low-density lipoprotein (LDL) cholesterol, diabetes duration, and BMI. Because of the collinearity between mean HbA1c (2024–2025) and CGM-derived glucose metrics, two sets of models were constructed: one including HbA1c and one excluding it. In addition, we explored potential effect modification by glycemic control through HbA1c×TIR and HbA1c×TITR interaction terms, and by performing stratified analyses according to the median HbA1c value of the cohort.

Results

Study cohort characteristics

Of the 342 individuals initially assessed for eligibility, 25 were excluded due to the absence of CGM data during follow-up or lack of ophthalmological evaluation within the study period. Among those excluded, 4 (16%) had DR, while 21 (84%) showed no signs of DR. The final cohort consisted of 309 participants who completed follow-up until May 2025. At baseline (May 2024), 198 participants (64.1%) had no evidence of DR, 71 (23.0%) had nonproliferative DR, and 40 (12.9%) presented with proliferative DR or diabetic macular edema. The cohort flowchart is shown in Figure 1. A total of 6198 CGM downloads were obtained out of a theoretical maximum of 7182 (86.3% completeness). After excluding downloads with <70% active sensor time, missing uploads, or temporary discontinuation of sensor use, and removing participants who did not complete follow-up, 4208 evaluable downloads were retained (median 11 [IQR 6–16] per participant).

FIG. 1.

Flowchart. Baseline: May 2024. A total of 115 individuals had some degree of diabetic retinopathy; four of them were lost to follow-up. *Incidence was estimated among individuals without DR at baseline.

At baseline, the mean age was 51.4 ± 17.3 years, and 46.6% were women. Fifty-seven individuals (18.5%) were active smokers and 90 (29.3%) had a diagnosis of hypertension. The mean diabetes duration was 24.5 ± 13.6 years. The most recent mean HbA1c was 7.3% ± 1.0% (56 ± 11 mmol/mol [45–67]). Median TIR (70–180 mg/dL) was 68% (IQR: 55–77), and TITR (70–140 mg/dL) was 45% (IQR: 33–55). Compared with participants without DR, those with retinopathy were older, had a higher prevalence of hypertension, a longer diabetes duration, and higher levels of HbA1c, mean glucose, glucose management indicator (GMI), and TAR >250 mg/dL. No significant differences were observed in TIR or TITR. Baseline characteristics are detailed in Table 1.

Table 1.

Baseline Characteristics of the Cohort

Variable	Obsn = 309	No diabetic retinopathyn = 198	Nonproliferative retinopathyn = 71	Proliferative retinopathy/macular edemaN = 40	P value
Age	51.4 ± 17.3	48.4 ± 18.0	56.4 ± 14.9	57.2 ± 14.1	<0.001
Sex women	144 (46.6%)	96 (48.5%)	32 (45.1%)	15 (40.0%)	0.592
BMI (Kg/m²)^*	25.4 ± 4.1	25.2 ± 3.9	25.5 ± 4.2	26.1 ± 5.1	0.448
Smokers	57 (18.5%)	37 (18.7%)	16 (22.5%)	4 (10.0%)	0.260
Hypertension	90 (29.3%)	38 (19.2%)	33 (46.5%)	19 (50.0%)	<0.001
LDL (mg/dL)	97 ± 32	99 ± 29	94 ± 37	96 ± 38	0.495
Duration of diabetes (years)	24.5 ± 13.6	19.3 ± 11.3	31.3 ± 12.1	37.9 ± 11.9	<0.001
HbA1c %, mmol/mol	7.3 ± 1.0; 56 ± 11	7.2 ± 1.0; 55 ± 11	7.6 ± 1.1; 60 ± 12	7.4 ± 0.8; 57 ± 9 mmol/mol	0.017
Insulin (IU/Kg/day)*	0.64 ± 0.36	0.66 ± 0.39	0.60 ± 0.30	0.64 ± 0.26	0.505
TIR	68 (55–77)	69 (58–78)	63 (51–76)	67 (51–78)	0.128
TITR (70–140 mg/dL)	45 (33–55)	45 (33–58)	39 (27–51)	45 (35–54)	0.089
TBR <70 mg/dL	3 (1 −6)	3 (1 −6)	3 (1 −4)	3 (1 −6)	0.185
TAR >180 mg/dL	28 (18 −42)	28 (17 −40)	32 (21 −48)	26 (17–38)	0.061
TAR >250 mg/dL	5 (2–12)	5 (2–11)	7 (3–16)	6 (1–13)	0.041
CV (%)	35 (31–39)	35 (31–39)	35 (30–39)	35 (31–43)	0.469

n = 289.

TAR, time above range; GMI, glucose management indicator; TIR, time in range (70–180 mg/dL); TITR, time in tight range (70–140 mg/dL); CV, coefficient of variation. U-Mann–Whitney test used. TAR >180 includes TAR >250.

Incidence of DR

Among the 198 participants without DR at baseline, 10 (5.1%) developed DR between May 2024 and May 2025. Those who developed DR had a longer diabetes duration (26.4 ± 11.0 vs. 18.9 ± 11.3 years; P = 0.034) and a higher median HbA1c during follow-up (7.4% [IQR: 6.7–7.9] (57 [IQR: 50–63] mmol/mol) vs. 7.0% [IQR: 6.5–7.5] (53 [IQR: 48–58] mmol/mol); P = 0.014). They also exhibited significantly higher values of mean glucose, GMI, and glucose coefficient of variation. In contrast, participants who did not develop DR had significantly higher TITR values (P < 0.05). No significant differences were observed in TIR between groups (P = 0.877). Full results are presented in Table 2.

Table 2.

Diabetic Retinopathy Incidence

Variable	Obsn = 198	No diabetic retinopathyn = 188	Incident retinopathyn = 10	P value
Age	48.4 ± 18.0	48.1 ± 18.0	53.9 ± 18.4	0.243
Sex women	96 (48.5%)	90 (47.9%)	6 (60.0%)	0.455
BMI (Kg/m²)^*	25.2 ± 3.9	25.2 ± 4.0	24.2 ± 1.9	0.426
Smokers	37 (18.7%)	35 (18.6%)	2 (20.0%)	0.913
Hypertension	38 (19.2%)	34 (18.1%)	4 (40.0%)	0.086
LDL (mg/dL)	99 ± 29	100 ± 30	93 ± 30	0.779
Duration of diabetes (years)	19.3 ± 11.3	18.9 ± 11.3	26.4 ± 11.0	0.034
HbA1c (%, mmol/mol) (2025)	7.0 (6.6–7.5) 53 (49–58)	7.0 (6.6–7.5) 53 (49–58)	7.4 (6.9–8.1) 57 (52–65	0.264
HbA1c (%, mmol/mol) (previous)	7.1 (6.5–7.6) 54 (48–60)	7.1 (6.5–7.6) 54 (48–60)	7.8 (6.7–8.2) 62 (50–66)	0.086
HbA1c (%, mmol/mol) (average)	7.1 (6.6–7.7) 54 (49–61)	7.0 (6.5–7.5) 53 (48–58)	7.4 (6.7–7.9) 57 (50–63)	0.0144
Insulin (IU/Kg/day)^*	0.66 ± 0.39	0.66 ± 0.40	0.56 ± 0.15	0.614
Glucose average (mg/dL)	150 (135–170)	150 (135–169)	159 (144–178)	<0.001
GMI (%)	6.9 (6.6–7.3)	6.9 (6.6–7.3)	7.2 (6.6–7.3)	<0.001
TIR ^*	68 (55–77)	69 (56–80)	68 (63–75)	0.8765
TITR (70–140 mg/dL)^*	41 (30–52)	43 (32–54)	41 (32–49)	0.009
TBR <70 mg/dL^*	3 (1–6)	3 (1–6)	3 (1–10)	0.104
TAR >180 mg/dL^*	29 (18–44)	28 (16–42)	28 (21–40)	0.3116
TAR >250 mg/dL^*	5 (2–11)	4 (2–9)	5 (3–8)	0.0588
CV (%)^*	35 (31–40)	35 (31–40)	35 (33–43)	0.0129

U-Mann–Whitney test used. TAR >180 includes TAR >250. ^*n = 289.

Panel data.

Longitudinal analysis of TIR and TITR revealed inconsistent TIR trajectories over time, while TITR showed a more consistent pattern, with persistently lower values in participants who developed DR. This suggests greater discriminative capacity of TITR regarding the development of DR (see Fig. 2).

FIG. 2.

Monthly time in range (TIR) and time in tight range (TITR) by diabetic retinopathy (DR) incidence status and discriminative performance for incident diabetic retinopathy. (A) Median TIR and TITR are plotted monthly over 12 months, stratified by incident DR status. Participants who developed DR during follow-up (red lines) consistently showed lower TITR values compared with those who remained free of DR (blue lines), whereas differences in TIR were less evident. TITR more clearly distinguished individuals who developed DR from those who did not, supporting its potential as a predictive glycemic marker in type 1 diabetes. (B) Receiver operating characteristic (ROC) curves demonstrate the superior discriminative ability of TITR compared with TIR for predicting incident DR. AUC, area under the curve.

Discriminative performance analysis

ROC curves were constructed to compare the discriminative performance of TITR and TIR in predicting incident DR. The AUC for TITR was 0.580, significantly higher than that of TIR (AUC = 0.430; P < 0.001) (see Fig. 2). Among other evaluated metrics, TAR >250 mg/dL showed the best performance (AUC = 0.5428; 95% CI: 0.5035–0.5822), followed by mean glucose (AUC = 0.5314) and TAR >180 mg/dL (AUC = 0.5148). GMI showed the lowest discriminative ability (AUC = 0.4625). Global comparisons between metrics were statistically significant (P < 0.001), indicating heterogeneity in their predictive performance (see Supplementary Table S1).

Multivariable analysis for DR incidence

We first evaluated the independent association between TITR and incident DR using multivariable logistic regression adjusted for age, sex, diabetes duration, hypertension, smoking, and LDL cholesterol. Higher TITR was significantly associated with a reduced risk of DR (adjusted OR: 0.986; 95% CI: 0.975–0.997; P = 0.016). In contrast, TIR was not significantly associated with incident DR (adjusted OR: 1.003; 95% CI: 0.999–1.018; P = 0.591). Other significant predictors in the TITR model included age, diabetes duration, hypertension, and smoking. Comparable associations were observed in the TIR model, except that TIR itself was not a significant predictor (Supplementary Table S2).

As a sensitivity analysis, both models were repeated with mean HbA1c (2024–2025) included as a covariate. TITR remained independently associated with a lower risk of incident DR (adjusted OR: 0.965; 95% CI: 0.950–0.980; P < 0.001), confirming the robustness of the association. However, in this setting, HbA1c appeared paradoxically protective (adjusted OR: 0.466; 95% CI: 0.318–0.680; P < 0.001), a finding that likely reflects collinearity between HbA1c and CGM-derived metrics rather than a true biological effect (Supplementary Table S2). We next examined whether the association of TITR and TIR with DR was modified by long-term glycemic control. Both metrics demonstrated significant interactions with HbA1c (TITR×HbA1c, P < 0.001; TIR×HbA1c, P = 0.003), indicating that their predictive value varied by HbA1c level. In stratified analyses, TITR was strongly associated with reduced DR risk among participants with HbA1c below the cohort median (7.1%), but not among those with higher HbA1c. Conversely, TIR showed an inverse association with DR only in the lower HbA1c stratum, while in the higher stratum it was not protective and even trended toward increased risk. These findings suggest that CGM-derived metrics, particularly TITR, provide additional prognostic information when HbA1c is near target, whereas their discriminative value is attenuated at higher HbA1c levels (Supplementary Table S2).

Progression of DR

Among the 111 participants with DR at baseline, 85 (76.6%) remained stable during follow-up, while 26 (23.4%) experienced disease progression. Those with progression had a significantly lower baseline BMI (24.0 ± 3.5 vs. 26.2 ± 4.7 kg/m²; P = 0.026) and lower TITR values [median: 36%; IQR: 27–48] compared with those with stable disease [median: 40%; IQR: 28–50]; P = 0.021. Participants with progression also showed higher values of mean glucose, GMI, and time above range (>180 and >250 mg/dL). TIR was lower in the progression group, although this did not reach statistical significance. Complete data are shown in Supplementary Table S3. Longitudinal analysis revealed consistently higher TIR and TITR values in participants without progression over the follow-up period, with stable trends over time (Fig. 3).

FIG. 3.

(A) Monthly evolution of TIR and TITR by DR progression status and discriminative performance for retinopathy progression. Median TIR and TITR are shown monthly in participants with pre-existing diabetic retinopathy, stratified by progression status. Individuals with progression (red lines) exhibited consistently lower values for both metrics compared with those without progression (blue lines), with similar temporal patterns. These results suggest that TIR and TITR provide comparable information for identifying individuals at risk of worsening retinopathy in type 1 diabetes. (B) ROC curves further demonstrate that TIR (70–180 mg/dL) and TITR (70–140 mg/dL) had nearly identical discriminative performance for DR progression (AUC = 0.534 vs. 0.541; P = 0.56), with no significant difference between them.

Predictive analysis of DR progression

ROC curve analysis showed that both TIR and TITR had similar discriminative performance in identifying DR progression (AUC = 0.534 and 0.541, respectively; P = 0.56, Fig. 3). GMI and TAR >250 mg/dL had the highest AUC values (0.5765 and 0.5756, respectively), followed by mean glucose (AUC = 0.5634). TAR >180 mg/dL had lower performance (AUC = 0.540). Overall differences among metrics were statistically significant (P < 0.001), indicating heterogeneity in their capacity to discriminate DR progression (see Supplementary Table S4).

Multivariable analysis for DR progression

Multivariable models adjusted for age, sex, diabetes duration, smoking status, hypertension, LDL cholesterol, and BMI showed that both TITR and TIR were associated with a lower risk of DR progression. In the model including TITR, higher TITR was associated with reduced progression risk (adjusted OR: 0.993; 95% CI: 0.981–1.002; P = 0.112), although the association was not statistically significant. In contrast, TIR remained significantly associated with lower progression risk (adjusted OR: 0.989; 95% CI: 0.981–0.998; P = 0.022). Subsequent models including mean HbA1c (2024–2025) confirmed that both TITR (adjusted OR: 0.982; 95% CI: 0.966–0.996; P < 0.001) and TIR (adjusted OR: 0.977; 95% CI: 0.965–0.990; P < 0.001) remained independently associated with DR progression. Full results are provided in Supplementary Table S5.

Discussion

This is the first study to directly compare the longitudinal utility of TITR (70–140 mg/dL) versus conventional TIR (70–180 mg/dL) for predicting the incidence and progression of DR in adults with T1D treated with multiple daily insulin injections. Our findings suggest that TITR may offer slightly greater predictive value than TIR for incident DR, independently of classical clinical variables such as HbA1c, hypertension, smoking status, or diabetes duration. In contrast, both metrics demonstrated comparable performance in predicting DR progression in individuals with preestablished disease.

In our cohort, CGM-derived metrics showed modest standalone discrimination and were most informative when interpreted alongside established risk factors. For incident DR, TITR outperformed TIR (AUC 0.580 vs. 0.430) and remained independently associated with lower risk after multivariable adjustment (adjusted OR per 1% increase 0.965; ≈30% lower odds per +10% TITR), whereas TIR was not independently predictive. For DR progression, TITR and TIR displayed similar and modest discrimination (AUC 0.541 and 0.534, respectively) and both were independently associated with lower risk in HbA1c-adjusted models (TITR OR per 1% 0.982; ≈16% lower odds per +10%; TIR OR per 1% 0.977; ≈21% per +10%). Other markers of glycemic burden (e.g., average glucose, GMI, and TAR >250 mg/dL) showed better discrimination for progression than TITR/TIR. Notably, classical determinants including longer diabetes duration, higher HbA1c, hypertension, and smoking contributed substantially to risk in all models, suggesting that CGM metrics should complement rather than replace established clinical predictors when stratifying the risk of incident and progressive DR.

TITR has gained increasing attention in recent literature, particularly in the context of advanced hybrid closed-loop systems, which aim to optimize glycemic control near euglycemia. Prior studies have demonstrated a strong correlation between TITR and TIR,^26,29 although this concordance may diminish in settings of low glycemic variability.^34,35 A recent study proposed that achieving a TITR >50% may represent a clinically meaningful target, being associated with a GMI <6.5%.²⁴ The longitudinal impact of TITR on DR was previously explored in a retrospective study collecting glucose metrics during annual clinical visits over 7 years, which found concordance between several glycemic metrics and DR incidence.²⁸ However, that study did not explicitly compare the predictive value of different metrics—likely due to its design and reliance on isolated annual data points. Our study expands upon these findings by employing continuous isCGM-derived data in a real-world clinical context, with an overall acceptable glycemic control (median HbA1c ∼7.0%) and a homogeneous cohort in terms of treatment regimen and CGM device.

In individuals without DR at baseline, TITR appeared to more accurately capture subtle fluctuations in glycemic exposure within the context of chronic good control, where extreme glucose values are less frequent. This greater sensitivity to reflect minor but sustained elevations in glucose may explain its superior utility in predicting DR onset. Supporting this hypothesis, studies in individuals with type 2 diabetes using short-term CGM and prospective ophthalmological follow-up have shown that DR development is more closely linked to mild but sustained hyperglycemic exposure—patterns that TITR may reflect better than TIR does.³⁶ Conversely, among individuals with pre-existing DR, TIR demonstrated comparable predictive utility with TITR. In this subgroup, characterized by higher median HbA1c (∼7.4%), glycemic burden appeared to be more heavily influenced by overt hyperglycemia. Indeed, TAR >250 mg/dL and GMI were shown to be the metrics with the strongest predictive value for DR progression. This pattern aligns with prior cross-sectional findings in individuals with HbA1c <7%, where TAR >250 mg/dL was the metric most strongly associated with the presence of DR.³⁷ The differing predictive value of TITR and TIR according to DR stage may reflect the nonlinear nature of their association with glycemic control. Beck et al. demonstrated that while TITR and TIR are highly concordant, this relationship is modulated by additional parameters such as the coefficient of variation and time in hypoglycemia.³⁸ A longitudinal study in nearly 2000 young individuals with T1D also found that glycemic variability influences the relationship between TITR and TIR.²⁶ In that study, TIR showed stronger predictive power for HbA1c, which may explain its weaker association with incident DR in our cohort. Specifically, because TIR is highly collinear with HbA1c, its independent predictive value diminishes when HbA1c is excluded from the model. In contrast, TITR may better reflect broader aspects of glucose stability, particularly postprandial and nocturnal control, and appears less affected by collinearity with HbA1c-potentially accounting for its consistent association with DR incidence in both adjusted and unadjusted models. This interpretation is supported by our stratified analyses, which revealed that TITR was strongly associated with a reduced risk of DR incidence only among participants with HbA1c below the cohort median of 7.1%, whereas the association disappeared at higher HbA1c levels. By contrast, TIR also predicted lower DR risk in the lower HbA1c stratum, but showed a positive association with DR incidence when HbA1c was above 7.1%. These findings suggest that TITR may be more informative in individuals achieving near-target glycemic control, while the apparent association of TIR with increased DR risk at higher HbA1c levels could reflect collinearity and statistical artifacts rather than a true biological relationship.

Taken together, our results support the combined use of TITR and HbA1c as prognostic tools in individuals without DR. Despite a uniform follow-up with consistent treatment regimens and isCGM devices, the overall predictive capacity of CGM-derived metrics for DR was moderate. In this context, modifiable cardiovascular risk factors such as smoking¹⁷ and hypertension^13,14 emerged as relevant determinants of DR incidence, while classical factors such as age, sex and dyslipidemia were associated with a greater risk of progression in individuals with pre-existing DR.⁸ An unexpected finding was the inverse association between BMI and DR risk, a relationship previously described in type 2 diabetes,³⁹ but less explored in T1D. Kim et al. reported a non-significant trend toward greater DR progression among individuals with lower BMI⁴⁰ and a recent large multicenter study suggested that the association between BMI and DR progression could vary by geographic region.⁴¹ Further prospective studies are warranted to confirm this hypothesis in T1D populations.

This study has important limitations. Its retrospective, single-center design limits external validity and precludes causal inference. We did not evaluate optimal cut-off points for TITR or TIR, as this would require internal or split-sample cross-validation in accordance with TRIPOD guidelines,⁴² followed by prospective external validation. Moreover, lifestyle factors such as physical activity and diet-which may influence DR risk-were not collected. Socioeconomic status, increasingly recognized as a determinant of diabetes complications,^43–45 was also not analyzed. Data on ophthalmological interventions and DR regression or improvement were not collected. Furthermore, although isCGM metrics were extracted in a standardized manner, HbA1c values were not always temporally aligned with CGM downloads. Finally, the small number of incident events may limit the robustness of the statistical analyses, and the exclusive inclusion of individuals treated with multiple daily insulin injections and isCGM restricts the generalizability of our findings.

In conclusion, TITR appears to offer superior utility over conventional TIR for predicting the incidence of DR, independently of other clinical variables including HbA1c. However, both metrics demonstrate similar performance in predicting disease progression. The relative value of each glycemic metric may depend on the stage of DR and the baseline level of glycemic control. Therefore, CGM-derived metrics should be interpreted within the context of the patient’s broader clinical profile rather than be considered as standalone predictors.

Authors’ Contributions

F.S.-V. conceived the study, coordinated the project, researched data, conducted the literature review, and wrote the first draft of the article. E.G.-A. and A.S.-R. collected and curated the data, contributed to the discussion, and reviewed/edited the article. S.G.C., S.A., and M.S.T.-S. contributed to data analysis, figure preparation, and article editing. V.N.-M. and J.A.A.-M. assisted with clinical interpretation of the results and critically revised the article. E.P.L.-P. contributed to the ophthalmological assessments and interpretation of retinopathy findings. M.M. supervised the study, supported funding acquisition, and reviewed/edited the article. All authors reviewed and approved the final version of the article.

Footnotes

Acknowledgment

The authors acknowledge the Spanish Society of Diabetes (Sociedad Española de Diabetes, SED) for its decisive support of this project through the “XV AYUDAS SED A PROYECTOS DE INVESTIGACIÓN EN DIABETES DIRIGIDOS POR JÓVENES INVESTIGADORES.”

Author Disclosure Statement

The authors declare that they have no competing interests.

Funding Information

This work was funded by Proyectos de Investigacion en Salud PI19/00584, PI22/01404 and PMP22/00021 (funded by Instituto de Salud Carlos III), iTIRONET- P2022/BMD7379 (funded by Comunidad de Madrid), and co-financed by FEDER funds to M.M.

Availability of Data and Materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics Approval and Consent to Participate

The Research Ethics Committee of Hospital Universitario de La Princesa (Madrid, Spain), (Study number: 5814 01/25 approved this study and waived informed consent from patients. The research was conducted according to the Declaration of Helsinki.

Supplemental Material

References

Crossen

, Xing

, Hoch

. Changing costs of type 1 diabetes care among US children and adolescents. Pediatr Diabetes, 2020; 21(4):644–648; doi: 10.1111/pedi.12996

Karl

, Winkler

, Ziegler

A-G

, et al. Costs of public health screening of children for presymptomatic type 1 diabetes in bavaria, Germany. Diabetes Care, 2022; 45(4):837–844; doi: 10.2337/dc21-1648

Solomon

, Chew

, Duh

, et al. Diabetic retinopathy: A position statement by the american diabetes association. Diabetes Care, 2017; 40(3):412–418; doi: 10.2337/dc16-2641

American Diabetes Association Professional Practice Committee. 12. retinopathy, neuropathy, and foot care: Standards of care in diabetes—2025. Diabetes Care, 2025; 48(Suppl 1):S252–S265; doi: 10.2337/dc25-S012

Ting

DSW

, Cheung

GCM

, Wong

. Diabetic retinopathy: Global prevalence, major risk factors, screening practices and public health challenges: A review. Clin Exp Ophthalmol, 2016; 44(4):260–277; doi: 10.1111/ceo.12696

Orji

, Rani

, Narayanan

, et al. The economic burden of diabetic retinopathy care at a tertiary eye care center in South India. Indian J Ophthalmol, 2021; 69(3):666–670.

Klein

, Knudtson

, Lee

, et al. The wisconsin epidemiologic study of diabetic retinopathy XXII: The twenty-five-year progression of retinopathy in persons with type 1 diabetes. Ophthalmology, 2008; 115(11):1859–1868; doi: 10.1016/j.ophtha.2008.08.023

Perais

, Agarwal

, Evans

, et al. Prognostic factors for the development and progression of proliferative diabetic retinopathy in people with diabetic retinopathy. Cochrane Database Syst Rev, 2023; 2(2):CD013775; doi: 10.1002/14651858.CD013775.pub2

Simó

, Hernández

. What else can we do to prevent diabetic retinopathy? Diabetologia, 2023; 66(9):1614–1621; doi: 10.1007/s00125-023-05940-5

10.

Yokoyama

, Uchigata

, Otani

, et al. Development of proliferative retinopathy in Japanese patients with IDDM: Tokyo Women’s Medical College Epidemiologic Study. Diabetes Res Clin Pract, 1994; 24(2):113–119; doi: 10.1016/0168-8227(94)90028-0

11.

Janghorbani

, Jones

, Allison

. Incidence of and risk factors for proliferative retinopathy and its association with blindness among diabetes clinic attenders. Ophthalmic Epidemiol, 2000; 7(4):225–241; doi: 10.1076/opep.7.4.225.4171

12.

Pambianco

, Costacou

, Ellis

, et al. The 30-year natural history of type 1 diabetes complications: The pittsburgh epidemiology of diabetes complications study experience. Diabetes, 2006; 55(5):1463–1469; doi: 10.2337/db05-1423

13.

Cruickshanks

, Moss

, Klein

, et al. Physical activity and the risk of progression of retinopathy or the development of proliferative retinopathy. Ophthalmology, 1995; 102(8):1177–1182; doi: 10.1016/S0161-6420(95)30893-7

14.

Grauslund

, Green

, Sjølie

. Prevalence and 25 year incidence of proliferative retinopathy among Danish type 1 diabetic patients. Diabetologia, 2009; 52(9):1829–1835; doi: 10.1007/s00125-009-1450-4

15.

Harris Nwanyanwu

, Talwar

, Gardner

, et al. Predicting development of proliferative diabetic retinopathy. Diabetes Care, 2013; 36(6):1562–1568; doi: 10.2337/dc12-0790

16.

Porta

, Sjoelie

A-K

, Chaturvedi

, et al.; EURODIAB Prospective Complications Study Group. Risk factors for progression to proliferative diabetic retinopathy in the EURODIAB Prospective Complications Study. Diabetologia, 2001; 44(12):2203–2209; doi: 10.1007/s001250100030

17.

Styles

, Dodds

, Watkins

, et al. Development of proliferative retinopathy in patients with long-standing insulin-dependent diabetes mellitus. Eye (Lond), 2000; 14(Pt 6):851–854; doi: 10.1038/eye.2000.235

18.

Nordwall

, Abrahamsson

, Dhir

, et al. Impact of HbA1c, Followed From Onset of Type 1 diabetes, on the development of severe retinopathy and nephropathy: The VISS Study (Vascular Diabetic Complications in Southeast Sweden). Diabetes Care, 2015; 38(2):308–315; doi: 10.2337/dc14-1203

19.

Arfken

, Reno

, Santiago

, et al. Development of proliferative diabetic retinopathy in african-americans and whites with type 1 diabetes. Diabetes Care, 1998; 21(5):792–795; doi: 10.2337/diacare.21.5.792

20.

Skrivarhaug

, Fosmark

, Stene

, et al. Low cumulative incidence of proliferative retinopathy in childhood-onset type 1 diabetes: A 24-year follow-up study. Diabetologia, 2006; 49(10):2281–2290; doi: 10.1007/s00125-006-0364-7

21.

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; doi: 10.2337/dci19-0028

22.

Beck

, Bergenstal

, Riddlesworth

, et al. Validation of time in range as an outcome measure for diabetes clinical trials. Diabetes Care, 2019; 42(3):400–405; doi: 10.2337/dc18-1444

23.

Shah

, DuBose

, Li

, et al. Continuous glucose monitoring profiles in healthy nondiabetic participants: A multicenter prospective study. J Clin Endocrinol Metab, 2019; 104(10):4356–4364; doi: 10.1210/jc.2018-02763

24.

Castañeda

, Arrieta

, van den Heuvel

, et al. Time in tight glucose range in type 1 diabetes: Predictive factors and achievable targets in real-world users of the minimed 780G System. Diabetes Care, 2024; 47(5):790–797; doi: 10.2337/dc23-1581

25.

Battelino

, Alexander

, Amiel

, et al. Continuous glucose monitoring and metrics for clinical trials: An international consensus statement. Lancet Diabetes Endocrinol, 2023; 11(1):42–57; doi: 10.1016/S2213-8587(22)00319-9

26.

Burckhardt

M-A

, Auzanneau

, Rosenbauer

, et al. What is the relationship between time in range, time in tight range, and hba1c in youth and young adults with type 1 diabetes? results from the german/austrian/luxembourgian/swiss diabetes prospective follow-up registry. J Diabetes Sci Technol, 2024:19322968241288870; doi: 10.1177/19322968241288870

27.

Zhang

, Wang

, Lu

, et al. Time in tight range: A key metric for optimal glucose control in the era of advanced diabetes technologies and therapeutics. Diabetes Obes Metab, 2025; 27(2):450–456; doi: 10.1111/dom.16033

28.

Shah

, Kanapka

, Akturk

, et al. Time in range is associated with incident diabetic retinopathy in adults with type 1 diabetes: A longitudinal study. Diabetes Technol Ther, 2024; 26(4):246–251; doi: 10.1089/dia.2023.0486

29.

De Meulemeester

, Charleer

, Visser

, et al. The association of chronic complications with time in tight range and time in range in people with type 1 diabetes: A retrospective cross-sectional real-world study. Diabetologia, 2024; 67(8):1527–1535; doi: 10.1007/s00125-024-06171-y

30.

von Elm

, Altman

, Egger

, et al.; STROBE Initiative. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Ann Intern Med, 2007; 147(8):573–577; doi: 10.7326/0003-4819-147-8-200710160-00010

31.

Jafri

, Balliro

, El-Khatib

, et al. A Three-Way accuracy comparison of the dexcom g5, abbott freestyle libre pro, and senseonics eversense continuous glucose monitoring devices in a home-use study of subjects with type 1 diabetes. Diabetes Technol Ther, 2020; 22(11):846–852; doi: 10.1089/dia.2019.0449

32.

Hanson

, Kipnes

, Tran

. Comparison of point accuracy between two widely used continuous glucose monitoring systems. J Diabetes Sci Technol, 2024; 18(3):598–607; doi: 10.1177/19322968231225676

33.

Anonymous. Grading diabetic retinopathy from stereoscopic color fundus photographs—an extension of the modified airlie house classification: ETDRS report Number 10. Ophthalmology, 1991; 98(5):786–806; doi: 10.1016/S0161-6420(13)38012-9

34.

Dunn

, Ajjan

, Bergenstal

, et al. Is it time to move beyond TIR to TITR? Real-world data from Over 20,000 users of continuous glucose monitoring in patients with type 1 and type 2 diabetes. Diabetes Technol Ther, 2024; 26(3):203–210; doi: 10.1089/dia.2023.0565

35.

Passanisi

, Piona

, Salzano

, et al. Aiming for the best glycemic control beyond time in range: Time in tight range as a new continuous glucose monitoring metric in children and adolescents with type 1 diabetes using different treatment modalities. Diabetes Technol Ther, 2024; 26(3):161–166; doi: 10.1089/dia.2023.0373

36.

Wang

, Lu

, Yu

, et al. Association between time in tight range and incident diabetic retinopathy in adults with type 2 diabetes. Diabetes Obes Metab, 2025; 27(3):1415–1422; doi: 10.1111/dom.16143

37.

Sebastian-Valles

, Martínez-Alfonso

, Arranz Martin

, et al. Time above range and no coefficient of variation is associated with diabetic retinopathy in individuals with type 1 diabetes and glycated hemoglobin within target. Acta Diabetol, 2025; 62(2):205–214; doi: 10.1007/s00592-024-02347-5

38.

Beck

, Raghinaru

, Calhoun

, et al. A comparison of continuous glucose monitoring-measured time-in-range 70–180 mg/dL versus time-in-tight-range 70–140 mg/dL. Diabetes Technol Ther, 2024; 26(3):151–155; doi: 10.1089/dia.2023.0380

39.

, Hou

, Zhang

, et al. Association between body mass index and diabetic retinopathy in Chinese patients with type 2 diabetes. Acta Diabetol, 2015; 52(4):701–708; doi: 10.1007/s00592-014-0711-y

40.

Kim

H-K

, Kim

C-H

, Kim

S-W

, et al. Development and progression of diabetic retinopathy in koreans with NIDDM. Diabetes Care, 1998; 21(1):134–138; doi: 10.2337/diacare.21.1.134

41.

Keen

, Lee

, Russell

, et al. The appearance of retinopathy and progression to proliferative retinopathy: The WHO multinational study of vascular disease in diabetes. Diabetologia, 2001; 44(Suppl 2):S22–S30; doi: 10.1007/PL00002935

42.

Collins

, Reitsma

, Altman

, et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): The TRIPOD statement. Ann Intern Med, 2015; 162(1):55–63; doi: 10.7326/M14-0697

43.

Alvarez-Ramos

, Jimenez-Carmona

, Alemany-Marquez

, et al. Socioeconomic deprivation and development of diabetic retinopathy in patients with type 1 diabetes mellitus. BMJ Open Diabetes Res Care, 2020; 8(2):e001387; doi: 10.1136/bmjdrc-2020-001387

44.

Mathur

, Bhaskaran

, Edwards

, et al. Population trends in the 10-year incidence and prevalence of diabetic retinopathy in the UK: A cohort study in the clinical practice research Datalink 2004–2014. BMJ Open, 2017; 7(2):e014444; doi: 10.1136/bmjopen-2016-014444

45.

Klein

, Klein

BEK

, Jensen

, et al. The relation of socioeconomic factors to the incidence of proliferative diabetic retinopathy and loss of vision. Ophthalmology, 1994; 101(1):68–76; doi: 10.1016/S0161-6420(94)31354-6

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.

0.00 MB

0.49 MB