This prospective population-based cohort study included participants enrolled in the UKB, a large cohort study that recruited almost half a million participants aged 40–69 years in the UK between 2006 and 2010 (5% response rate) through postal invitation. The UKB was approved by the North West Multicenter Research Ethics Service (NHS National Research Ethics Service, 16/NW/0274). Written informed consent was obtained from all participants for data collection, analysis, and linkage. Each participant completed a touchscreen questionnaire and a nurse-led interview and underwent physical measurements in one of the 22 assessment centers across England, Scotland, and Wales. Blood and urine samples were collected for biological marker assessments. This study was conducted under approved application number 532248. The inclusion criteria required participants to meet the definition of CKD at baseline (n = 28,283). The exclusion criteria included meeting the definition of ESKD at baseline (n = 391), having missing data on the WWI (n = 226) or other obesity-related indicators (n = 35), and lacking complete information on covariates (n = 5,108). After applying these criterias, a total of 22,523 participants were retained for the final analysis (Figure 1). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

Chronic kidney disease was identified based on baseline biological sample testing. In accordance with the “Kidney Disease: Improving Global Outcomes (KDIGO) 2021 Guidelines,” CKD is defined by a urinary albumin-to-creatinine ratio (UACR) exceeding 30 mg/g and/or an estimated glomerular filtration rate (eGFR) below 60 mL/min/1.73 m² (4). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2021 creatinine-based equation for standardized creatinine (27).

This study utilized baseline anthropometric data from the UKB, which were collected by trained personnel under standardized protocols with regularly calibrated equipment to ensure high measurement accuracy. Body weight (kg) was assessed using the Tanita BC-418 MA body composition analyzer after the participants removed their shoes and heavy outer clothing and standing height (m) was measured without shoes using a Seca 202 (28). WC was measured to the nearest 0.1 cm at the level of the umbilicus using a Wessex non-stretchable spring tape, with the participant in a resting standing position. BMI was calculated as body weight divided by the square of standing height. WWI (cm/√kg) was calculated as WC divided by the square root of body weight and rounded to two decimal places.

The outcome events of this study included all-cause mortality, CVD mortality and ESKD incidence. Mortality data were ascertained through data linkage with national death registries (NHS Information Center for participants in England and Wales and NHS Central Registry for participants in Scotland). CVD mortality was determined based on the primary cause of death, identified using ICD-10 codes (I20-I26, I60-I69). ESKD was defined as CKD stage G5 or the requirement for renal replacement therapy, including long-term dialysis or kidney transplantation, identified through hospital admission ICD-10 codes and OPCS-4 codes or ICD-10 codes recorded in any position of a death certificate (Supplementary Table 1).

This study included multiple potential covariates, including age, sex, education level (college/above, high school/equivalent, less than high school or other), townsend deprivation index (TDI), annual household income (< £18 000, £18 000 to £51 999, ≥ £52 000 or other), smoking status (non-smoker, ex-smoker, current smoker or other), alcohol consumption (never, light, excessive or other), healthy diet score (poor, medium or ideal), physical activity (goal or below goal), white blood cell (WBC) count, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglyceride, eGFR, UACR, hypertension (yes or no), diabetes mellitus (yes or no) and hyperuricemia (yes or no), which were collected through baseline questionnaires, verbal interviews, and laboratory test indicators. The TDI was calculated using national census data and assigned postal codes of residence and categorized by quartiles, with a lower score indicating a higher area level of socioeconomic status (29). Physical activity was defined according to the American Heart Association (AHA) recommendations, with goal physical activity defined as engaging in ≥150 min/week of moderate-intensity activity, ≥ 5 min/week of vigorous-intensity activity or an equivalent combination (30). Healthy diet score were determined based on a touchscreen food frequency questionnaire (FFQ), including the dietary consumption of poultry, beef, lamb/mutton, processed meat, oily fish, non-oily fish, fresh fruit, dried fruit, raw vegetables and cooked vegetables and categorized into poor, medium or ideal groups following a previous UKB study (31). Hypertension was identified by self-reported physician diagnosis, use of antihypertensive medication, or elevated baseline blood pressure (systolic ≥ 140 mmHg or diastolic ≥ 90 mmHg). Diabetes mellitus was identified by self-reported diagnosis, use of glucose-lowering medication, or increased fasting glucose (≥7.0 mmol/L), 2-h plasma glucose (≥11.1 mmol/L), or HbA1c (≥6.5%). Hyperuricemia was identified by increased serum uric acid levels (>7.0 mg/dL in men and > 6.0 mg/dL in women). Standardized protocols for the collection and processing of baseline blood and urine samples have been previously described (32). Detailed definitions and classification criteria for covariates are provided in the Supplementary Methods.

Continuous variables with a non-normal distribution were expressed as median (interquartile range, IQR), whereas those with a normal distribution were expressed as mean (standard deviation, SD). Categorical variables were presented as counts (percentages). Group comparisons were performed using the Kruskal–Wallis test or one-way ANOVA for continuous variables, and the χ² test for categorical variables. The association of WWI, BMI and WC with outcomes was analyzed using the Cox proportional hazards model and expressed as hazard ratios (HRs) and 95% confidence intervals (CIs). In the analysis, WWI, BMI and WC were included in the model as continuous variables and quartiles, respectively. Model 1 was adjusted for age and sex. Model 2 further included annual household income, smoking status, alcohol consumption, education, TDI, healthy diet score, physical activity, hyperuricemia, hypertension, diabetes mellitus, HDL-C, LDL-C, triglycerides and WBC count. Model 3 additionally adjusted for eGFR and UACR. Survival time was defined as the duration from the baseline to the outcome event or censoring date (31 December 2021). Kaplan–Meier survival curves for the different groups were compared using log-rank tests. Restricted cubic spline (RCS) regression was used to examine the non-linear association between WWI and outcomes, and a two-stage Cox proportional hazards regression model was constructed based on the threshold point. The effectiveness of WWI, BMI, and WC in predicting outcomes was evaluated using smoothed receiver operating characteristic (ROC) curves. Pairwise comparisons of the areas under the ROC curves (AUCs) were performed using a non-parametric bootstrap procedure with 2,000 iterations to assess statistical differences between indices. Bonferroni correction was applied to adjust for multiple comparisons, and corresponding p-values were reported to determine the significance of AUC differences.

To ensure the robustness of our findings, three different sensitivity analysis were performed. The first excluded participants with major CVDs (myocardial infarction, coronary heart disease, or stroke) or cancer at baseline; the second excluded participants with less than 2 years of follow-up; and the third employed Fine–Gray competing risk models to account for potential competing events, where death was considered a competing risk for ESKD incidence and non-CVD death for CVD mortality. Additional subgroup analysis and interaction tests evaluated the association between WWI and outcomes across sex, age and disease status.

All analysis were performed using R software version 4.4.1 [R Core Team (33). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL¹ ]. Statistical significance was defined as a two-tailed p-value of <0.05.

This study included 22,523 participants who met the eligibility criteria. The median (IQR) age of the participants was 61 (54–66) years, and 9,289 (52.9%) were female. The median follow-up duration was 12.6 (11.7–13.4) years. During 269,858.2 person-years of follow-up, 3,874 mortality occurred, including 786 cardiovascular mortality, and 823 ESKD incidence. The median WWI was 10.5. Participants were grouped according to WWI quartiles: quartile 1 (6.59–9.96), quartile 2 (9.96–10.50), quartile 3 (10.50–11.10) and quartile 4 (11.10–14.91). Participants with a higher WWI were older, more likely to be male, had a higher TDI, lower annual household income, below goal physical activity, unhealthy diet and a higher prevalence of diabetes, hypertension and hyperuricemia (Table 1).

The KM survival analysis showed significant differences in all-cause (Supplementary Figure 1A), CVD (Supplementary Figure 1B) mortality and ESKD incidence (Supplementary Figure 1C) between the different WWI quartiles. A higher WWI was associated with higher all-cause and CVD mortality during follow-up (p < 0.0001 for all log-rank tests).

Based on COX regression analysis, continues WWI was associated with outcomes in all models. The results in model3 showed significant positive associations with all-cause mortality (per unit increase; HR: 1.30, 95% CI: 1.24–1.37), CVD mortality (1.26, 1.12–1.41) and ESKD incidence (1.27, 1.14–1.43) (Table 2). Additionally, compared with those in quartile 1, participants in quartile 4 had a significantly higher risk of all-cause mortality (HR: 1.44, 95% CI: 1.28–1.63), cardiovascular mortality (1.42, 1.06–1.90) and ESKD incidence (1.41, 1.09–1.82). In model 3, participants in quartile 4 of BMI and WC were not significantly associated with all-cause and CVD mortality compared to quartile 1. Highest levels of BMI was found to be protective against the ESKD incidence (Supplementary Tables 2, 3).

Restricted cubic spline analysis showed that WWI was non-linearly associated with all-cause mortality (p for non-linear < 0.001) and ESKD incidence (p for non-linear < 0.001), while it had a linear association with CVD mortality (p for non-linear = 0.175) (Figure 2). Notably, all RCS curves exhibited a consistent inflection pattern, with breakpoints located approximately at the median WWI value (10.5); therefore, the median was selected as the threshold for consenquent analysis. Cox regression analysis showed a positive correlation between WWI and all-cause mortality (HR: 1.46, 95% CI: 1.34–1.58), CVD mortality (1.34, 1.13–1.58) and ESKD incidence (1.55, 1.29–1.86) to the right of the threshold (Table 3), whereas to the left, there was no significant correlation. The RCS analysis of BMI (Supplementary Figure 2) and WC (Supplementary Figure 3) demonstrated relatively consistent patterns. A significant non-linear U-shaped association was observed for both all-cause mortality and ESKD incidence, with significance evident on either side of the breakpoint. In contrast, the association with CVD mortality exhibited a linear trend.

Subgroup analysis and interaction tests stratified by age, sex, and disease status showed a significant interaction between WWI and CVD mortality according to sex, hypertension, and hyperuricemia (p < 0.05) (Supplementary Table 4). However, the direction of this association remained consistent across the subgroups. The association between WWI and ESKD incidence remained consistent across age, sex, and other disease subgroups, except in the hyperuricemia group (p < 0.05). In sensitivity analysis, the associations of WWI with all-cause and CVD mortality were largely unchanged. Regarding ESKD incidence, the results were consistent across analysis except for the second sensitivity analysis, in which a slight deviation was observed (Supplementary Tables 5–7).

Figure 3 demonstrates the predictive performance of WWI, WC and BMI for all-cause (Figure 2A) and CVD (Figure 2B) mortality and ESKD incidence (Figure 2C). The area under the curves (AUCs) for predicting all-cause mortality were 0.636, 0.598, and 0.541, respectively. The CVD mortality rates were 0.661, 0.642, and 0.585, ESKD incidences were 0.612, 0.604 and 0.558, respectively. Compared with other traditional obesity indicators, WWI demonstrated a significantly higher AUC for predicting all-cause mortality (p < 0.0001 for all AUC comparisons). It also outperformed BMI in predicting cardiovascular mortality and ESKD, while showing comparable performance to WC (Supplementary Table 8).