Association between time in range and incident early diabetic kidney disease in patients with type 2 diabetes: a retrospective cohort study

Abstract

Background

Diabetic kidney disease (DKD) is a leading cause of end-stage renal disease in type 2 diabetes mellitus (T2DM). Time in range (TIR), a continuous glucose monitoring-derived metric, has emerged as an important glycemic control indicator, but evidence linking TIR to incident early DKD in T2DM remains limited.

Methods

This retrospective cohort study included 498 Chinese T2DM patients with preserved kidney function at baseline who underwent continuous glucose monitoring between January 2020 and December 2022, with follow-up until December 2024. TIR and other glycemic metrics were assessed at baseline. The primary outcome was incident early DKD. Cox proportional hazards models evaluated associations between TIR and early DKD risk, with comprehensive adjustment for demographics, clinical characteristics, and medications. Subgroup and sensitivity analyses were performed.

Results

During a median 24-month follow-up, 62 (12.4%) patients developed early DKD. After full adjustment, TIR ≥ 70% was associated with lower early DKD risk versus TIR < 70% (HR = 0.54, 95%CI 0.30–0.96, p = 0.048). A significant dose–response relationship was observed across TIR quartiles (P for trend = 0.008), with each 10% TIR increase associated with a 16% risk reduction (HR = 0.84, 95%CI 0.73–0.98, p = 0.022). The protective association was more pronounced in patients not using sodium-glucose cotransporter 2 inhibitors (P for interaction = 0.036). Results remained consistent across sensitivity analyses.

Conclusions

Higher TIR is significantly associated with reduced risk of incident early DKD in T2DM patients with preserved kidney function, exhibiting a clear dose–response pattern. TIR may serve as a valuable complementary metric for assessing renal complication risk.

Introduction

Diabetic kidney disease (DKD) is one of the most common and severe microvascular complications of type 2 diabetes mellitus (T2DM), affecting 30–40% of patients and representing a leading cause of end-stage renal disease worldwide [1]. Early DKD, characterized by microalbuminuria (30–300 mg/g creatinine) or mildly reduced estimated glomerular filtration rate with evidence of kidney damage, significantly increases cardiovascular morbidity and mortality risk [2, 3]. Early detection and intervention are crucial, as evidence indicates that timely treatment can delay or prevent disease progression [4].

Traditional glycemic markers, such as glycated hemoglobin (HbA1c), have long been used to assess long-term glycemic control and predict microvascular complications [5]. However, HbA1c has several limitations: it cannot reflect short-term glycemic variability, is influenced by conditions such as anemia and hemoglobinopathies, and fails to capture glucose fluctuations that may contribute to microvascular injury [6, 7]. The advent of continuous glucose monitoring (CGM) has enabled detailed characterization of glycemic profiles [8]. Among CGM-derived metrics, time in range (TIR)—the proportion of time spent with glucose levels between 3.9 and 10.0 mmol/L (70–180 mg/dL)—has emerged as a robust indicator of glycemic control [9]. International consensus recommends a TIR target > 70% for most adults with diabetes, corresponding approximately to an HbA1c of 7.0% [10, 11]. TIR offers several advantages over HbA1c, including real-time glucose pattern assessment, capture of glycemic variability, and better correlation with patient-perceived diabetes management success [12]. Growing evidence supports the association between TIR and diabetes complications. Using data from the Diabetes Control and Complications Trial, Beck et al. demonstrated that each 10% reduction in TIR was associated with a 64% increased hazard rate for retinopathy progression and a 40% increased hazard rate for microalbuminuria development [13]. Recent research in type 1 diabetes showed that both TIR and time in tight range are inversely associated with microvascular complications, including diabetic nephropathy [14]. Nevertheless, evidence regarding the association between TIR and early DKD in T2DM remains limited and most existing studies have employed cross-sectional designs that cannot establish a temporal relationship.

Given the increasing prevalence of T2DM globally and the critical importance of early DKD detection, this retrospective cohort study aimed to investigate whether baseline TIR is associated with the risk of incident early DKD in Chinese patients with T2DM. By following patients without DKD at baseline and observing subsequent DKD development, we sought to provide stronger evidence for the temporal relationship between TIR and early DKD. Establishing such an association may strengthen the rationale for incorporating TIR into clinical practice as a target for optimizing glycemic management and preventing early renal injury.

Methods

Study design and participants

This retrospective cohort study was conducted at the Affiliated Nantong Hospital of Shanghai University (The Sixth People’s Hospital of Nantong), China. We reviewed electronic medical records of T2DM patients who underwent CGM monitoring between January 2020 and December 2022, with follow-up until December 2024. This study was reported according to the STROBE guidelines [15]. Of 1035 patients screened, 498 were included (Fig. 1). Exclusions: baseline UACR ≥ 30 mg/g or eGFR < 60 (n = 241), inadequate CGM data (n = 136), insufficient follow-up (n = 98), recent regimen changes (n = 38), and other reasons (n = 24).

Fig. 1.

Study flow diagram. T2DM type 2 diabetes mellitus, CGM continuous glucose monitoring, DKD diabetic kidney disease, UACR urinary albumin-to-creatinine ratio, eGFR estimated glomerular filtration rate, TIR time in range, IQR interquartile range

Inclusion criteria were: (1) age ≥ 18 years; (2) confirmed T2DM diagnosis; (3) CGM monitoring ≥ 14 days with ≥ 70% sensor active time; (4) baseline eGFR ≥ 60 mL/min/1.73 m² and UACR < 30 mg/g; and (5) minimum 12-month follow-up with ≥ 2 kidney function assessments.

Exclusion criteria were: (1) type 1 or other specific diabetes types; (2) glucose-lowering regimen changes within 3 months before CGM; (3) acute diabetic complications at baseline; (4) known non-diabetic kidney diseases; (5) conditions affecting UACR measurement; (6) pregnancy; (7) malignancy with expected survival < 2 years; or (8) incomplete data.

This study was approved by the Institutional Ethics Committee (Approval No. NTLYLL2025049). Informed consent was waived due to the retrospective design.

Data collection

Baseline data included demographics, clinical history, blood pressure, diabetic retinopathy status (assessed by fundoscopy within 6 months), laboratory parameters (HbA1c, fasting glucose, lipid profile, serum creatinine, UACR), and medication use. The eGFR was calculated using the CKD-EPI 2021 equation [16].

CGM data and TIR assessment

All patients underwent CGM using FreeStyle Libre Pro (Abbott Diabetes Care, Alameda, CA, USA) for ≥ 14 days. CGM metrics were calculated according to international consensus: [11] TIR (3.9–10.0 mmol/L), time above range (TAR, > 10.0 mmol/L), time below range (TBR, < 3.9 mmol/L), coefficient of variation (CV), and mean glucose. Patients were categorized as TIR ≥ 70% (target achieved) or TIR < 70% based on the same international consensus recommendation [11], and further divided into quartiles based on the sample distribution of baseline TIR values for dose–response analysis: Q1 (< 50%, n = 124), Q2 (50–62%, n = 124), Q3 (62–74%, n = 124), and Q4 (≥ 74%, n = 126).

Outcome definition

The primary outcome was incident early DKD, defined as UACR ≥ 30 mg/g confirmed on two consecutive measurements ≥ 3 months apart, with eGFR ≥ 60 mL/min/1.73 m² at diagnosis, according to KDIGO 2024 guidelines [17].

Statistical analysis

Continuous variables were expressed as mean ± SD or median (IQR). Categorical variables were expressed as n (%). Baseline characteristics were compared using t test, Mann–Whitney U test, or chi-square test as appropriate.

Kaplan–Meier curves with the log-rank test were used to compare cumulative incidence. Cox proportional hazards models estimated HRs with 95% CIs. Multicollinearity was assessed using variance inflation factors (VIF). Four models were constructed: Model 1 (unadjusted); Model 2 (adjusted for age, sex, BMI, diabetes duration, smoking); Model 3 (Model 2 + systolic blood pressure, LDL-C, diabetic retinopathy); Model 4 (Model 3 + baseline eGFR, baseline UACR, ACEI/ARB use, SGLT2i use). TIR was analyzed as binary (≥ 70% vs < 70%), quartiles, and continuous (per 10% increase). The proportional hazards assumption was tested using Schoenfeld residuals.

Subgroup analyses assessed effect modification by age, diabetes duration, SGLT2i use, ACEI/ARB use, and GLP-1RA use, with interaction terms tested by likelihood ratio test. Sensitivity analyses included: (1) excluding patients with regimen changes during follow-up; (2) excluding baseline SGLT2i users; (3) excluding baseline UACR 20–29 mg/g; (4) stricter outcome definition (3 consecutive confirmations); (5) Fine-Gray competing risk model with death as a competing event; and (6) excluding insulin users. Missing data (< 5%) were handled using multiple imputation. Subgroup and sensitivity analyses were prespecified and considered exploratory for assessing effect modification and robustness of findings. No adjustments for multiple comparisons were made, as these were not independent primary hypotheses.

Statistical analyses were performed using R version 4.3.1. A two-sided P < 0.05 was considered significant.

Results

Study population

Baseline characteristics are shown in Table 1. Mean age was 55.2 ± 10.3 years, 54.0% were male, median diabetes duration was 6.0 years, mean HbA1c was 7.9 ± 1.4%, and mean TIR was 60.9 ± 18.6%. Diabetic retinopathy was present in 89 (17.9%) patients. Compared with the TIR < 70% group, the TIR ≥ 70% group was younger, had shorter diabetes duration, lower HbA1c, and higher SGLT2i use (all p < 0.05). No significant differences in baseline eGFR, UACR, or retinopathy prevalence.

Table 1.

Baseline characteristics by time in range status

Variable	Total (n = 498)	TIR ≥ 70% (n = 183)	TIR < 70% (n = 315)	p value
Age, years	55.18 ± 10.32	53.62 ± 9.84	56.08 ± 10.52	0.012
Male, n (%)	269 (54.0)	101 (55.2)	168 (53.3)	0.689
BMI, kg/m2	25.42 ± 3.62	25.18 ± 3.48	25.56 ± 3.70	0.268
Diabetes duration, years	6.0 (3.0–10.0)	5.0 (2.0–8.0)	7.0 (4.0–12.0)	< 0.001
Hypertension, n (%)	212 (42.6)	74 (40.4)	138 (43.8)	0.464
Current smoker, n (%)	142 (28.5)	49 (26.8)	93 (29.5)	0.509
Diabetic retinopathy, n (%)	89 (17.9)	28 (15.3)	61 (19.4)	0.253
Systolic BP, mmHg	132.48 ± 15.56	131.36 ± 14.82	133.14 ± 15.96	0.226
HbA1c, %	7.86 ± 1.42	7.12 ± 0.94	8.29 ± 1.48	< 0.001
Fasting glucose, mmol/L	8.22 ± 2.12	7.32 ± 1.62	8.74 ± 2.22	< 0.001
LDL-C, mmol/L	2.84 ± 0.86	2.80 ± 0.84	2.86 ± 0.88	0.456
eGFR, mL/min/1.73 m2	98.36 ± 14.28	98.92 ± 13.96	98.04 ± 14.48	0.508
UACR, mg/g	12.42 (8.24–18.68)	11.96 (7.92–17.36)	12.72 (8.48–19.24)	0.112
TIR, %	60.85 ± 18.62	77.94 ± 6.42	50.92 ± 14.38	< 0.001
TAR, %	33.02 ± 17.68	18.42 ± 6.84	41.50 ± 15.82	< 0.001
TBR, %	6.13 ± 4.82	3.64 ± 3.22	7.58 ± 5.18	< 0.001
CV, %	34.28 ± 8.58	29.92 ± 6.24	36.82 ± 8.84	< 0.001
Metformin, n (%)	419 (84.1)	153 (83.6)	266 (84.4)	0.806
SGLT2i, n (%)	134 (26.9)	63 (34.4)	71 (22.5)	0.004
GLP-1RA, n (%)	75 (15.1)	31 (16.9)	44 (14.0)	0.369
Insulin, n (%)	148 (29.7)	40 (21.9)	108 (34.3)	0.003
ACEI/ARB, n (%)	158 (31.7)	57 (31.1)	101 (32.1)	0.833
Statins, n (%)	228 (45.8)	86 (47.0)	142 (45.1)	0.680
MRA, n (%)	18 (3.6)	6 (3.3)	12 (3.8)	0.764

Data: mean ± SD, median (IQR), or n (%). TIR = time in range; BMI = body mass index; BP = blood pressure; eGFR = estimated glomerular filtration rate; UACR = urinary albumin-to-creatinine ratio; TAR = time above range; TBR = time below range; CV = coefficient of variation; SGLT2i = sodium-glucose cotransporter 2 inhibitor; GLP-1RA = glucagon-like peptide-1 receptor agonist; ACEI/ARB = angiotensin-converting enzyme inhibitor/angiotensin receptor blocker; MRA = mineralocorticoid receptor antagonist

Incidence of early DKD

Median follow-up was 24 months (IQR: 18.0–30.0). During follow-up, 62 (12.4%) patients developed early DKD. 11 patients died and 19 were lost to follow-up.

The TIR < 70% group had higher early DKD incidence than TIR ≥ 70% group (14.9% vs 8.2%, p = 0.028; incidence rates 7.46 vs 4.10 per 100 person-years). Kaplan–Meier analysis confirmed significantly higher cumulative incidence in TIR < 70% group (log-rank p = 0.012; Fig. 2).

Fig. 2.

Kaplan–Meier curves for incident early DKD by TIR status. Cumulative incidence of early DKD was significantly lower in the TIR ≥ 70% group (blue) versus the TIR < 70% group (red). Log-rank χ² = 6.28, p = 0.012. DKD diabetic kidney disease, TIR time in range

Association between TIR and early DKD

Cox regression results are shown in Table 2. In the unadjusted model, TIR ≥ 70% was associated with lower early DKD risk (HR = 0.52, 95%CI 0.29–0.93, p = 0.028). After full adjustment (Model 4), the association showed borderline significance (HR = 0.54, 95% CI: 0.30–0.96, p = 0.048), with consistent direction across all models.

Table 2.

Cox regression for association between TIR and incident early DKD

TIR Category	Events/N	Rate (/100PY)	Model 1	Model 2	Model 3	Model 4
< 70%	47/315	7.46	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
≥ 70%	15/183	4.10	0.52 (0.29–0.93)*	0.54 (0.30–0.97)*	0.55 (0.31–0.98)*	0.54 (0.30–0.96)*
Quartiles
Q1 (< 50%)	25/124	10.08	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)	1.00 (Ref)
Q2 (50–62%)	17/124	6.85	0.66 (0.36–1.20)	0.67 (0.36–1.23)	0.68 (0.37–1.24)	0.68 (0.37–1.25)
Q3 (62–74%)	12/124	4.84	0.47 (0.24–0.92)*	0.48 (0.24–0.95)*	0.49 (0.25–0.96)*	0.50 (0.25–0.98)*
Q4 (≥ 74%)	8/126	3.17	0.35 (0.16–0.75)**	0.37 (0.17–0.80)**	0.38 (0.17–0.82)**	0.39 (0.18–0.84)*
p for trend			0.002	0.004	0.006	0.008
Continuous
Per 10% increase	–	–	0.82 (0.71–0.94)**	0.83 (0.72–0.96)*	0.84 (0.72–0.97)*	0.84 (0.73–0.98)*

Data: HR (95% CI). Model 1: unadjusted. Model 2: adjusted for age, sex, BMI, diabetes duration, smoking. Model 3: Model 2 + systolic BP, LDL-C, diabetic retinopathy. Model 4: Model 3 + baseline eGFR, baseline UACR, ACEI/ARB, SGLT2i. *p < 0.05, **p < 0.01. Per 10% TIR increase (Model 4): HR = 0.84 (0.73–0.98), p = 0.022

A significant dose–response was observed across quartiles. Using Q1 (< 50%) as reference, adjusted HRs were: Q2 0.68 (p = 0.216), Q3 0.50 (p = 0.045), Q4 0.39 (p = 0.016); p for trend = 0.008. As a continuous variable, each 10% TIR increase was associated with a 16% risk reduction (HR = 0.84, 95%CI 0.73–0.98, p = 0.022).

The proportional hazards assumption was satisfied (global test P = 0.386). All variance inflation factors were < 3, indicating no multicollinearity.

Subgroup analyses

The association was consistent across most subgroups (Table 3). A significant interaction was observed with SGLT2i use (p = 0.036): in non-SGLT2i users, TIR ≥ 70% was associated with a 48% risk reduction (HR = 0.52, 95%CI 0.27–0.99); in SGLT2i users, the association was attenuated (HR = 0.82, 95%CI 0.28–2.41). The association was consistent regardless of GLP-1RA use (p for interaction = 0.284), although the analysis in GLP-1RA users was limited by small sample size (n = 75). No significant interactions for age, diabetes duration, or ACEI/ARB use.

Table 3.

Subgroup analyses

Subgroup	TIR ≥ 70% Events/N	TIR < 70% Events/N	HR (95% CI)	p interaction
Age				0.286
< 60 years	7/116	28/182	0.51 (0.22–1.18)
≥ 60 years	8/67	19/133	0.66 (0.29–1.50)
DM duration				0.208
< 10 years	9/132	27/200	0.53 (0.25–1.12)
≥ 10 years	6/51	20/115	0.70 (0.28–1.75)
SGLT2i use				0.036
No	10/120	40/244	0.52 (0.27–0.99)*
Yes	5/63	7/71	0.82 (0.28–2.41)
ACEI/ARB use				0.456
No	10/126	32/214	0.55 (0.27–1.11)
Yes	5/57	15/101	0.61 (0.22–1.68)
GLP-1RA use				0.284
No	13/152	41/271	0.53 (0.29–0.97)
Yes	2/31	6/44	0.62 (0.18–2.15)

All HRs are adjusted for Model 4 covariates (excluding the stratification variable). *P < 0.05

TIR time in range, HR hazard ratio, CI confidence interval, DM diabetes mellitus, SGLT2i sodium-glucose cotransporter 2 inhibitor, ACEI/ARB angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, GLP-1RA glucagon-like peptide-1 receptor agonist

Sensitivity analyses

Results remained robust across all sensitivity analyses (Table 4). After excluding patients with regimen changes during follow-up (n = 465), the association remained significant (HR = 0.51, p = 0.029). Excluding baseline SGLT2i users (n = 364) strengthened the protective association (HR = 0.48, p = 0.031), consistent with subgroup findings. Excluding insulin users (n = 350) yielded consistent results (HR = 0.52, p = 0.042), indicating that the association was not driven by disease severity related to insulin use. When excluding patients with baseline UACR 20–29 mg/g (n = 418), the association showed a similar direction but attenuated significance (HR = 0.58, p = 0.072). Using a stricter outcome definition requiring three consecutive confirmations (54 events) yielded an HR = 0.52 (p = 0.034). A complete case analysis without imputation produced nearly identical results (HR = 0.53, p = 0.035). The Fine-Gray competing risk model accounting for death confirmed consistent findings (subdistribution HR = 0.55, p = 0.041).

Table 4.

Sensitivity analyses

Analysis	N	Events	HR (95% CI)	p value
Main analysis	498	62	0.54 (0.30–0.96)	0.048
Excluding regimen changes	465	58	0.51 (0.28–0.93)	0.029
Excluding SGLT2i users	364	50	0.48 (0.25–0.93)	0.031
Excluding insulin users	350	42	0.52 (0.27–0.98)	0.042
Excluding UACR 20–29 mg/g	418	48	0.58 (0.32–1.06)	0.072
Stricter definition (3 ×)	498	54	0.52 (0.28–0.95)	0.034
Complete case analysis (no imputation)	475	59	0.53 (0.29–0.95)	0.035
Competing risk model	498	62	0.55 (0.30–0.97)	0.041

All models were adjusted for Model 4 covariates

HR hazard ratio, CI confidence interval, SGLT2i sodium-glucose cotransporter 2 inhibitor, UACR urinary albumin-to-creatinine ratio

Discussion

In this retrospective cohort study of 498 Chinese patients with T2DM and normal baseline kidney function, we observed that higher TIR was associated with lower risk of incident early DKD over a median follow-up of 24 months. The primary binary comparison between TIR ≥ 70% and < 70% showed borderline statistical significance in the fully adjusted model (HR = 0.54, 95% CI: 0.30–0.96, P = 0.048), with a relatively wide confidence interval reflecting imprecise effect estimation. However, multiple converging lines of evidence strengthen this association: TIR analyzed as a continuous variable demonstrated a significant 16% risk reduction per 10% increment (HR = 0.84, 95% CI: 0.73–0.98, P = 0.022), and a robust dose–response gradient was evident across quartiles (P for trend = 0.008). The consistency and dose-dependency of these findings suggest a clinically meaningful relationship between TIR and early renal outcomes, though confirmation in larger prospective cohorts is warranted given the observational design and modest sample size.

Our findings align with emerging evidence linking CGM-derived metrics to microvascular complications in diabetes. A recent systematic review examining TIR as a predictor of microvascular complications in type 2 diabetes found consistent associations between lower TIR and increased risk of nephropathy, retinopathy, and neuropathy across multiple studies [18]. Kovatchev et al.[19] recently applied machine learning techniques to reconstruct virtual CGM data from the landmark Diabetes Control and Complications Trial, demonstrating that TIR was significantly associated with the development of retinopathy, nephropathy, and neuropathy, with predictive value comparable to or exceeding that of HbA1c. In the context of diabetic kidney disease specifically, Yoo et al.[20] reported that lower TIR was associated with albuminuria progression over 3 years in Korean patients with type 2 diabetes, while Mayeda et al.[21] found associations between TIR and peripheral neuropathy in patients with diabetes and chronic kidney disease. Our retrospective cohort design with incident outcomes and a requirement for confirmation of albuminuria on two separate occasions extends these observations and strengthens the evidence for a temporal relationship between TIR and early renal injury.

The observed early DKD incidence of 12.4% over approximately 2 years (6.22 per 100 person-years) merits contextualization within existing epidemiological data. Systematic reviews report an annual albuminuria incidence of approximately 8% in type 2 diabetes populations, with low eGFR incidence of 2–4% per year regardless of diabetes type [22, 23]. The UK Prospective Diabetes Study documented that 38% of newly diagnosed type 2 diabetes patients developed albuminuria over a 15-year median follow-up, with 28% developing renal impairment [24]. A US claims-database study (Zhou et al., 2017) reported cumulative incidences of “diabetic nephropathy” (defined as newly recorded albuminuria, low eGFR, or initiation of renal-replacement therapy) of 5.3% at 1 year, 12.3% at 3 years and 14.7% at 5 years among adults with type 2 diabetes and hypertension [25]. The higher incidence rate observed in our study may reflect selection bias inherent in a tertiary-referral population, where patients undergoing CGM usually represent a high-risk subgroup with more advanced disease and poorer glycemic control than the broader diabetes population.

The significant interaction between TIR and SGLT2i use (P for interaction = 0.036) represents an intriguing finding that warrants careful interpretation. The protective association of higher TIR was substantially attenuated in patients receiving SGLT2i therapy (HR = 0.82) compared with non-users (HR = 0.52). SGLT2 inhibitors confer renoprotection through mechanisms largely independent of systemic glycemic control, including restoration of tubuloglomerular feedback, reduction of intraglomerular hypertension, improvement in tubular oxygenation, and attenuation of inflammation and fibrosis [26, 27]. The CREDENCE and DAPA-CKD trials established the nephroprotective benefits of SGLT2i even in patients with modest glycemic improvements [28, 29]. In this context, our observation may reflect overlapping or redundant protective mechanisms—patients already receiving SGLT2i may derive proportionally less additional benefit from TIR optimization because a substantial component of their renal risk has already been mitigated. However, this interpretation requires validation in larger studies, and we emphasize that this finding should not diminish the importance of glycemic optimization in SGLT2i-treated patients. Current guidelines recommend both SGLT2i therapy and comprehensive glycemic management as complementary strategies for DKD prevention [30].

Our findings should be interpreted within the context of the evolving therapeutic landscape for DKD. Current evidence supports a multi-pillar approach to DKD management, encompassing renin-angiotensin system (RAS) inhibitors, SGLT2 inhibitors, GLP-1 receptor agonists, and nonsteroidal mineralocorticoid receptor antagonists (MRAs) such as finerenone [31, 32]. These agents provide renoprotection through complementary mechanisms that are largely independent of glycemic control [31]. Our observation that TIR retains a significant association with early DKD risk after adjustment for ACEI/ARB and SGLT2i use suggests that the TIR-DKD relationship is not merely a surrogate for the renoprotective effects of these medications. Notably, GLP-1RA use did not significantly modify the TIR-DKD association in our subgroup analysis (P for interaction = 0.284), although this analysis was limited by the small number of GLP-1RA users (n = 75). The relatively low utilization of GLP-1RAs (15.1%) and MRAs (3.6%) in our cohort, along with the absence of finerenone use, limits our ability to fully evaluate the interplay between TIR and these newer therapeutic agents. Future studies with higher adoption rates of these four-pillar therapies are needed to clarify whether TIR provides incremental predictive value in comprehensively treated patients [32].

Several pathophysiological mechanisms may underlie the observed TIR-DKD relationship. Lower TIR inherently reflects greater cumulative hyperglycemic exposure, which activates multiple injury pathways in the diabetic kidney including advanced glycation end-product accumulation, protein kinase C activation, oxidative stress generation, and polyol pathway flux [33]. Additionally, patients with lower TIR typically exhibit greater glycemic variability, which emerging evidence suggests may independently contribute to microvascular injury through repeated oxidative stress cycles and endothelial dysfunction [34, 35]. A recent study by Zinman et al. demonstrated that glucose variability metrics, independent of mean glucose, were associated with cardiovascular and renal outcomes in high-risk T2DM patients [36]. Whether the TIR-DKD association in our study is driven primarily by reduced hyperglycemic exposure, attenuated glycemic variability, or both cannot be determined from our data, as these parameters are inherently correlated.

Our findings have potential implications for clinical practice. TIR could be incorporated into routine follow-up assessments as a complementary metric to HbA1c for identifying T2DM patients at elevated risk of early DKD, with the international consensus target of TIR > 70%[11] serving as a practical therapeutic goal for overall glycemic management that, based on our findings, may also have implications for renal risk reduction. As CGM devices become increasingly available and affordable, periodic CGM assessment (e.g., 14-day evaluations at regular intervals) could be implemented in primary care and endocrinology clinic settings to monitor glycemic patterns beyond what HbA1c captures. Furthermore, integrating TIR data into electronic health records and clinical decision support systems may facilitate early identification and timely intervention for patients at risk of developing DKD.

Our study has several limitations that should be acknowledged. The single-center retrospective design limits generalizability, and the population undergoing CGM may not be representative of the broader T2DM population. TIR was assessed only at baseline, and we could not account for subsequent changes in glycemic control that may have influenced DKD development. However, prior studies have demonstrated that a 14-day CGM assessment provides reproducible TIR estimates with good intra-individual consistency [37]. Additionally, patients with glucose-lowering regimen changes within 3 months before CGM were excluded, ensuring a relatively stable glycemic state at assessment. Nevertheless, future prospective studies incorporating serial CGM evaluations are warranted to provide a more comprehensive profile of glycemic trends and validate our findings. Despite comprehensive covariate adjustment including diabetic retinopathy status, residual confounding by unmeasured factors such as dietary habits, physical activity, and genetic susceptibility cannot be excluded. In particular, insulin users were more prevalent in the TIR < 70% group (34.3% vs 21.9%), raising concern about confounding by disease severity. However, sensitivity analysis excluding insulin users yielded consistent results, suggesting the association is unlikely to be substantially driven by disease severity related to insulin use. Although the primary binary comparison reached statistical significance (P = 0.048), the relatively wide confidence interval (0.30–0.96) suggests imprecise effect estimation, and larger studies are needed to obtain more precise estimates. Moreover, the relatively short follow-up duration precluded assessment of hard renal endpoints such as a 50% or greater decline in eGFR, doubling of serum creatinine, or initiation of dialysis. Given that our study population had preserved kidney function at baseline (mean eGFR 98.4 mL/min/1.73 m²), no patients experienced these hard endpoints during the 24-month follow-up, which is consistent with the natural history of early-stage disease. Future long-term prospective studies are needed to evaluate the association between TIR and these hard renal outcomes. Although our subgroup and sensitivity analyses were prespecified, the multiple comparisons involved increase the possibility of chance findings, particularly for the borderline significant interaction with SGLT2i use, and these results should be considered exploratory and require validation in independent cohorts. The observational nature of our study also precludes causal inference; whether interventions specifically targeting TIR improvement can reduce DKD incidence requires evaluation in randomized controlled trials.

Conclusions

This retrospective cohort study demonstrates that higher TIR is significantly associated with reduced risk of incident early DKD in T2DM patients with preserved kidney function. The association exhibits a clear dose–response pattern, with each 10% increase in TIR corresponding to approximately 16% lower risk. These findings suggest that TIR may serve as a valuable complementary metric to HbA1c for assessing glycemic control and identifying patients at elevated risk of early renal complications. Future prospective studies with serial CGM assessments are warranted to confirm these observations and evaluate whether interventions targeting TIR improvement can effectively reduce diabetic kidney disease incidence.

Author contributions

Ling Chen made substantial contributions to the conception and design of the study, analysis and interpretation of data, and drafting or revising the manuscript. Lingyan Zhang performed the statistical analysis and original drafting. Niuniu Zhou provided critical revisions to the manuscript. All authors approved the final version of the manuscript.

Funding

There is no funding.

Data availability

All data generated or analyzed in this study are included in the present manuscript.

Declarations

Conflict of interest

The authors declare no conflict of interest related to this study.

Ethics approval

This study was approved by the institutional ethics committee of the Affiliated Nantong Hospital of Shanghai University (NTLYLL2025049). The study was conducted in accordance with the Declaration of Helsinki.

Informed consent

The requirement for informed consent was waived due to the retrospective nature of the study.