The study population was derived from the ongoing prospective Kailuan Study, initiated in 2006 with follow-up visits at 2-year intervals. Baseline and follow-up examinations include measurements of blood UA, HDL-C, serum creatinine, and urinary protein. To assess the association between cumUHR and CKD, we selected individuals who participated in the baseline examination and completed two consecutive follow-up visits as our study population. The inclusion criteria were as follows: participation in the Kailuan Group health examinations in 2006, 2008, and 2010; complete data on UA, HDL-C, serum creatinine, and urinary protein for all three examinations; and written informed consent obtained before study enrollment. Exclusion criteria encompassed individuals with an estimated glomerular filtration rate (eGFR) <15 mL/min/1.73 m² or heavy proteinuria (urinary protein ≥3+) and those with missing follow-up serum creatinine or urinary protein data were excluded. The study was approved by the Ethics Committee of Kailuan General Hospital and adhered to the principles of the Declaration of Helsinki.

The methods used to collect epidemiological, anthropometric, and biochemical indicators in the Kailuan Study have been described elsewhere (11). Height, body weight, blood pressure, and related measurements are obtained by trained medical staff using standardized methods.

On the day of examination, 5 mL of venous blood are collected from each participant between 07:00 and 09:00 after fasting for more than 8 h for determination of serum UA, HDL-C, creatinine, and urinary protein. Biochemical parameters are measured using an automatic biochemical analyzer (Hitachi 7600; Tokyo, Japan) in the central laboratory at Kailuan General Hospital. Serum UA is measured using an enzymatic method, HDL-C using a direct test method (Mind Bioengineering Co., Ltd., Shanghai, China), and creatinine using an enzymatic method with an intra-batch coefficient of variation of <10%, an inter-batch relative range of <10%, and a linear range of 44–106 μmol/L. The specific methods used to detect other biochemical indicators in blood have been published previously (12).

Urinalysis was performed using a dry chemistry method and a urine sediment detection method (H12-MA urine analysis reagent strips and DIRUI N-600 urine routine detection analyzer sourced from Changchun Derui Medical Technology Co., Ltd.). Urinary protein was measured using the semi-quantitative test strip method [(-), <15 mg/dL; trace, 15–29 mg/dL; (1+), 30–300 mg/dL; (2+), 300–1000 mg/dL; 3+), >1000 mg/dL].

According to the latest clinical practice guidelines from KDIGO (Kidney Disease Improving Global Outcomes) (16), CKD is defined as eGFR <60 mL/min/1.73 m² and/or urine protein positive (17), low eGFR is defined as eGFR <60 mL/min/1.73 m², and proteinuria is defined as a semi-quantitative urinary protein value ≥1+. According to the KDIGO guidelines (18) and the recent report by the CKD Prognosis Consortium (19), CKD progression is defined as (1) a decline in eGFR category (≥90 [G1], 60–89 [G2], 45–59 [G3a], 30–44 [G3b], 15–29 [G4], and <15 [G5] mL/min/1.73 m²) accompanied by a decrease in eGFR of ≥25% from baseline), (2) a >30% decline in eGFR from baseline, or (3) an increase in proteinuria (i.e., an increase in semi-quantitative urine protein [–, ±, 1+, 2+, 3+]) (20).

Hypertension is defined as a systolic blood pressure ≥140 mmHg, a diastolic blood pressure ≥90 mmHg, or current use of antihypertensive medication. Diabetes is defined as fasting blood glucose ≥7.0 mmol/L or use of antidiabetic medication. Alcohol consumption is defined as an average daily intake of 100 mL of liquor (alcohol content ≥50%) for at least 1 year. Smoking is defined as smoking at least 1 cigarette per day on average in the past year. Physical exercise is defined as exercising ≥3 times per week, with each session lasting ≥30 minutes. Body mass index (BMI) is calculated as weight (kg) divided by the square of height (m).

The baseline for follow-up of study participants was completion of the 2010 health examination. For participants without CKD, the endpoint events included new-onset CKD, low eGFR, and proteinuria, with the follow-up endpoint being the last health examination (December 31, 2019). In the case of participants with CKD, the endpoint events were defined as a decline in eGFR category, a >30% decline in eGFR, and an increase in proteinuria, and the follow-up endpoint was the last health examination (December 31, 2019). In instances where a participant experienced two or more endpoint events, the earliest time and event were utilized as the outcome.

The baseline data for this study were collected from the 2010 health examination. Normally distributed continuous variables are expressed as the mean ± standard deviation, and group comparisons were conducted using one-way analysis of variance. For non-normally distributed continuous variables, the median (interquartile range) was utilized, and non-parametric (Kruskal-Wallis) tests were employed for between-group comparisons. Categorical variables are expressed as percentages, and group comparisons were assessed using chi-squared tests. Participants were stratified into cumUHR quartiles, and a Cox proportional hazards regression model was used to analyze the association between cumUHR levels and incidence and progression of CKD. Models were adjusted progressively for age, sex, educational level, smoking, alcohol consumption, physical exercise, BMI, low-density lipoprotein cholesterol, high-sensitivity C-reactive protein, baseline eGFR (in 2010), hypertension, diabetes, antihypertensive, antidiabetic medications, and the UHR level in 2006. The cumulative incidence rate of endpoint events in different cumUHR groups was calculated using the Kaplan-Meier method, and group comparisons were performed using the log-rank test. The CKD prediction model by Nelson et al. (21) served as the CKD base model, and a CKD progression prediction model was developed based on traditional risk factors, including age, sex, and systolic blood pressure as the CKD progression base model (22), UHR₂₀₀₆ and cumUHR were respectively added to calculate the C-index (C-statistic), net reclassification improvement (NRI), and integrated discrimination improvement (IDI) for comparing the predictive abilities of UHR2006 and cumUHR for the incidence and progression of CKD.

To further confirm the robustness of our study, we used a Cox model stratified by time intervals and fitted Cox models with time-varying coefficients. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). All p-values were two-tailed and considered statistically significant at <0.05.

A total of 57,927 participants underwent physical examinations in 2006, 2008, and 2010, and they were included in the study. After excluding participants with missing UA, HDL-C, urine protein, or serum creatinine data from the three physical examinations (n=542), those with an eGFR <15 mL/min/1.73 m² and/or heavy proteinuria (n=45), and those with missing follow-up urine protein and serum creatinine data (n=2128), 49,913 participants (mean age of 52.57 ± 11.77 years, 77.0% males) were available for analysis ( Figure 1 ). Participants were divided into four quartiles based on cumUHR level: first quartile, 746.34 ± 115.21; second quartile, 1004.72 ± 63.55; third quartile, 1273.68 ± 95.52; and fourth quartile, 1870.00 ± 593.72. Increasing cumUHR was associated with gradual increases in participant age, BMI, high-sensitivity C-reactive protein, systolic blood pressure, fasting blood glucose, and UA levels and a decrease in the HDL-C level. There were also gradual increases in the proportions of men, smokers, consumers of alcohol, and those who exercised regularly, as well as in the proportions of participants who tested positive for urine protein, had hypertension or diabetes, or were taking antihypertensive or antidiabetic medication. The differences between the groups were statistically significant (P<0.05, Table 1 ). Of the 49,913 participants, 13,024 (26.09%) had CKD, and 36,889 (73.91%) did not. The baseline characteristics of the non-CKD and CKD groups according to cumUHR quartile are shown in Supplementary Tables 1 , 2 , respectively.

Over a mean follow-up of 8.0 ± 1.7 years, the non-CKD group experienced 4843 cases of new-onset CKD, 2504 instances of low eGFR, and 2617 cases of proteinuria. In the CKD group, with a mean follow-up of 8.0 ± 1.6 years, there were 1952 cases of a decline in eGFR category, 1465 cases of >30% decline in eGFR, and 2100 cases of increased proteinuria. The corresponding cumulative incidence rates across various cumUHR groups are presented in Figure 2 (log-rank test P < 0.01).

In the non-CKD group, utilizing cumUHR quartile as the independent variable and CKD, low eGFR, and proteinuria as dependent variables, with the first quartile as the reference group, the multivariate-adjusted HRs for new-onset CKD in the second to fourth quartiles were 1.12 (95% CI 1.03–1.22), 1.22 (95% CI 1.12–1.33), and 1.48 (95% CI 1.36–1.62), respectively (P_trend<0.01); the corresponding HRs for developing low eGFR were 1.16 (95% CI 1.12–1.43), 1.27 (95% CI 1.12–1.43), and 1.64 (95% CI 1.46–1.85), while those for developing proteinuria were 1.06 (95% CI 0.94–1.19), 1.19 (95% CI 1.06–1.33), and 1.32 (95% CI 1.18–1.49) (P_trend<0.01). For each 1 standard deviation increase in cumUHR, the HRs for developing CKD, low eGFR, and proteinuria were 1.09 (95% CI 1.07–1.11), 1.09 (95% CI 1.07–1.11), and 1.08 (95% CI 1.06–1.11), respectively ( Table 2 ).

In the CKD group, using cumUHR quartile as the independent variable and decline in eGFR category, >30% decline in eGFR, and increase in proteinuria as the dependent variables, with the first quartile of cumUHR as the reference group, the multivariate-adjusted HRs for decline in eGFR category in the second to fourth quartiles were 1.06 (95% CI 0.93–1.02), 1.07 (95% CI 0.93–1.22), and 1.34 (95% CI 1.16–1.53) (P_trend<0.01); the respective HRs for >30% decline in eGFR were 1.12 (0.96–1.30), 1.17 (0.99–1.37), and 1.43 (1.22–1.68) (P_trend<0.01); the HRs in the second to fourth quartiles for an increase in proteinuria were 0.96 (95% CI 0.85–1.10), 1.15 (95% CI 1.01–1.32), and 1.45 (95% CI 1.27–1.65) (P_trend<0.01). For each 1 standard deviation increase in cumUHR, the HRs for decline in eGFR category, >30% decline in eGFR, and increase in proteinuria were 1.07 (95% CI 1.04–1.10), 1.04 (95% CI 1.02–1.07), and 1.06 (95% CI 1.04–1.08), respectively ( Table 3 ).

To determine whether the effect of cumUHR on new-onset CKD and progression of CKD could be influenced by a single UHR measurement, we further adjusted for the 2006 UHR level, and the results were consistent with these reported above ( Tables 2 , 3 ).

We observed significant interactions between cumUHR and sex (P-interaction = 0.047) with risk of CKD. The association between cumUHR and CKD was more pronounced in female population ( Table S3 ).

To reduce the impact of CKD on cumUHR, we further stratified the CKD population into subgroups according to whether eGFR was ≥45 or <45 mL/min/1.73 m²; after adjusting for confounding factors, the HRs for decline in eGFR category, >30% decline in eGFR, and increase in proteinuria were 1.34 (95% CI 1.17–1.54), 1.44 (95% CI 1.22–1.69), and 1.45 (95% CI 1.27–1.66) in the group with eGFR ≥45 mL/min/1.73 m² group (P<0.01). Conversely, there were no associations in the group with eGFR <45 mL/min/1.73 m² (P<0.01, Figure 3 ).

We separately added UHR₂₀₀₆ and cumUHR to the CKD base prediction model and the CKD progression base prediction model, and found that the cumUHR model had the highest predictive value. Compared with the UHR₂₀₀₆ model, the C-index of cumUHR model increased by 0.09%, 0.1% and 0.13% for new-onset CKD, low eGFR, and proteinuria, respectively, in the non-CKD group. The C-index of cumUHR model increased by 0.21%, 0.27% and 0.46% for decline in eGFR category, >30% decline in eGFR, and increase in proteinuria in the CKD group. In addition, compared with the base models, both NRI and IDI of cumUHR model were higher than those of the UHR2006 model ( Tables 4 , 5 ).

The time-varying cox model stratified by time intervals and fitted with resampling methods also showed consistent results with the conventional Cox proportional model ( Tables S4 , S5 ).