This investigation utilized data from the Kailuan Study, an ongoing population-based longitudinal investigation initiated in 2006 with biennial follow-up assessments capturing updated demographic, laboratory, imaging, and lifestyle parameters as previously detailed (24). From the baseline population of 136,967 participants who underwent initial health examinations between 2006–2007, 2008–2009, and 2010–2011, we excluded individuals with: 1) missing baseline BRI data; 2) missing alcohol consumption records or ultrasound examination data; 3) History of viral hepatitis or liver cirrhosis; 4) failure to meet MASLD diagnostic criteria; 5) malignancy; and 6) extreme BRI values (>99% or < 1%). After exclusions, 28,898 eligible MASLD patients were included in the final cohort (Figure 1). This study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Institutional Review Board of Kailuan General Hospital (Approval No: 2006-05). All participants provided written informed consent.

Trained personnel obtained height, weight, and waist circumference (WC) using standard protocols. Height and WC were measured to the nearest 0.1 cm, with weight recorded to 0.1 kg using calibrated platform scales. Participants wore light indoor clothing without shoes during measurements. WC was assessed at the umbilicus level with non-stretchable tape at the end of normal expiration. Body Roundness Index (BRI) was calculated as follows (17).BRI=364.2-365.5×1-(WC/2π)2(0.5×height)2

The diagnosis of MASLD followed a two-stage protocol. First, all participants underwent abdominal ultrasonography for hepatic steatosis assessment. The details of the ultrasound procedure for diagnosing hepatic steatosis are as described previously (24). Hepatic steatosis is diagnosed by ultrasound in the presence of at least two of the following conditions: (1) diffuse increased echogenicity of the liver relative to the kidney, (2) attenuation of the ultrasound beam, and (3) poor visualization of intrahepatic structures, and classified into three categories: mild (diffuse increase in fine echogenicity in the liver parenchyma), moderate (diffuse increase in fine echogenicity with impaired visualization of intrahepatic vascular borders and diaphragm), and severe (diffuse increase in fine echogenicity with no visibility of intrahepatic vascular borders and diaphragm). In the Kailuan Study, standardized abdominal ultrasounds were performed by certified radiologists using high-resolution B-mode ultrasound systems (ACUSON X300; Siemens Healthineers, Germany) equipped with 3.5 MHz convex-array transducers.

Participants diagnosed with hepatic steatosis who denied excessive alcohol consumption (ethanol intake ≥210 g/week for men or ≥140 g/week for women) subsequently underwent metabolic evaluation. A diagnosis of MASLD requires the presence of at least one of five cardiometabolic risk factors (CMRFs) (1): MASLD was defined as the presence of SLD and one or more of the following cardiometabolic risk factors: (1) body mass index ≥23kg/m2 or waist circumference ≥90cm (for males) or≥80cm (for females); (2) fasting glucose ≥100mg/dl or type 2 diabetes or glucose-lowering drug use; (3) BP≥130/85 mm Hg or BP-lowering drug use; (4) triglyceride ≥150mg/dl or lipid-lowering drug use; (5) high-density lipoprotein cholesterol < 40mg/dl (for males) or < 50mg/dl (for females) or lipid-lowering drug use. Participants demonstrating hepatic steatosis with ≥1 cardiometabolic risk factor (CMRF) were classified as MASLD cases after exclusion of secondary causes of steatosis.

The primary outcome was all-cause mortality. The follow-up period was defined as the time from baseline until the occurrence of death or the end of the follow-up period (December 31, 2021). All-cause mortality was ascertained annually by medical professionals using death certificates obtained from provincial vital statistics offices.

Epidemiological data encompassing anthropometric measurements, lifestyle factors, personal medical history, medication use (including glucose-lowering, antihypertensive, and lipid-modifying agents), and a history of cardiovascular disease were collected through face-to-face interviews using standardized structured questionnaires. Educational attainment was categorized as below high school or high school and above. Smoking status was classified as current or never/former smoker. Seated systolic blood pressure (SBP) was measured twice at 5-min intervals on the left upper arm using calibrated mercury sphygmomanometers after at least 5 mins of rest, with the average value used for analysis. Anthropometric measurements included height, weight, waist circumference, and hip circumference. Fasting blood samples were collected after overnight fasting and analyzed at the central laboratory of Kailuan General Hospital using standardized protocols and automated analyzers (Hitachi 747). Laboratory measurements included neutrophil count (NEUT), high-density lipoprotein cholesterol (HDL-C), high-sensitivity C-reactive protein (Hs-CRP), fasting blood glucose (FBG), uric acid (UA), and creatinine (Cr). Hepatitis B surface antigen (HBsAg) status was assessed using chemiluminescence immunoassays. Covariates were selected a priori based on biological plausibility and existing literature as potential confounders of the association between BRI and all-cause mortality. These adjustments aimed to account for baseline cardiometabolic, inflammatory, lifestyle, and socioeconomic factors that may influence both body fat distribution and mortality risk, while minimizing overadjustment by avoiding variables that clearly lie on the causal pathway.

Data from standardized physical examinations were entered by trained personnel into an Oracle 10.2g database hosted at Kailuan General Hospital. Analyses were performed using SAS version 9.4. Normally distributed continuous variables are expressed as mean ± standard deviation, with between-group comparisons assessed by one-way ANOVA. Non-normally distributed data are reported as median (25th−75th percentiles) using Kruskal-Wallis H tests. Categorical variables are presented as counts (%) with chi-square tests for group comparisons.

Participants were stratified into quartiles (Q1-Q4) based on BRI values. All-cause mortality rates were calculated per 1,000 person-years. Cumulative all-cause mortality probabilities were estimated using Kaplan-Meier curves with between-quartile differences assessed by log-rank tests. Dose-response relationships between continuous BRI and all-cause mortality risk were modeled using restricted cubic splines (RCS) with 3 knots at the 10th, 50th, and 90th percentiles. Multivariable Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between BRI and all-cause mortality after sequential adjustment: Model 1 (unadjusted), Model 2 (age and gender adjusted), Model 3 (fully adjusted for clinical/lifestyle covariates). The proportional hazards assumption was formally assessed using Schoenfeld residuals, and no significant violations were detected. Prespecified subgroup analyses were conducted to examine potential effect modification across clinically relevant strata within the MASLD population. A sensitivity analysis was conducted to assess the robustness of the findings. Participants who died within the first 2 years of follow-up were excluded, while participants with cancer at baseline were retained. Missing values for covariates were handled using multiple imputation by chained equations, assuming data were missing at random. All statistical tests were two-sided, and a P value < 0.05 was considered statistically significant.

This study enrolled 28,898 individuals with metabolic dysfunction-associated steatotic liver disease (MASLD) with a mean age of 52.34 ± 12.20 years (SD), comprising 22,998 male and 5,900 female participants (Figure 1). Participants were stratified into BRI quartiles: Q1 (BRI ≤ 3.64; n = 7,202), Q2 (3.64 < BRI ≤ 4.28; n = 7,276), Q3 (4.28 < BRI ≤ 5.04; n = 7,184), and Q4 (BRI >5.04; n = 7,236). Clinically significant between-quartile differences (P < 0.05) were observed for age, gender, Hs-CRP, FBG, HDL-C, CR, SBP, UA, Neutrophil, hip circumference, smoking status, cardiovascular disease history, hypertension history, and educational attainment (Table 1).

During a median follow-up of 13.62 years (interquartile range: 12.85–15.16 years), a total of 3,895 deaths were documented. The study population included 22,998 men and 5,900 women. The all-cause mortality rates per 1,000 person-years increased progressively across BRI quartiles, from 6.58 in Q1 to 8.11 in Q2, 10.70 in Q3, and 14.33 in Q4. Correspondingly, the cumulative all-cause mortality rose from 9.04% in Q1 to 11.15%, 14.56%, and 19.17% in Q2–Q4, respectively. Kaplan–Meier analysis demonstrated significant differences in cumulative all-cause mortality across BRI quartiles (log-rank test, P < 0.001; Figure 2).

Restricted cubic spline analysis showed a positive linear association between continuous BRI levels and all-cause mortality in the MASLD population (P for overall association < 0.001; P for nonlinearity = 0.143; Figure 3).

In Cox proportional hazards analyses, higher BRI quartiles were consistently associated with increased risk of all-cause mortality (Table 2). In the unadjusted model, the hazard ratios (HRs) for Q2, Q3, and Q4 compared with Q1 were 1.23 (95% CI: 1.11–1.37), 1.64 (95% CI: 1.48–1.80), and 2.20 (95% CI: 2.01–2.42), respectively. After adjustment for age and sex, the corresponding HRs were 1.02 (95% CI: 0.92–1.13), 1.12 (95% CI: 1.02–1.24), and 1.24 (95% CI: 1.12–1.36). Further multivariable adjustment attenuated the associations but they remained statistically significant for higher BRI categories, with HRs of 1.05 (95% CI: 0.94–1.18), 1.14 (95% CI: 1.02–1.28), and 1.21 (95% CI: 1.06–1.37) for Q2–Q4, respectively (P for trend < 0.001). When BRI was analyzed as a continuous variable, each standard deviation increase in BRI was associated with a 13% higher risk of all-cause mortality (adjusted HR: 1.13; 95% CI: 1.07–1.19; P < 0.001).

Stratified analyses were performed to examine the association between BRI and all-cause mortality across different characteristics of the MASLD population. Cox proportional hazards models were constructed with BRI quartiles (Q1 as the reference) as the exposure and all-cause mortality as the outcome, adjusting for the same covariates as in Model 3. Stratification was conducted according to age, sex, hypertension status, and education level. Among participants younger than 65 years, higher BRI levels were associated with increased all-cause mortality risk. Compared with Q1, the adjusted hazard ratios (HRs) for Q2, Q3, and Q4 were 1.10 (95% CI: 0.96–1.27), 1.19 (95% CI: 1.03–1.38), and 1.38 (95% CI: 1.17–1.63), respectively. In other subgroups, including individuals aged ≥65 years, those with education beyond high school, and non-hypertensive participants, the associations between BRI quartiles and all-cause mortality were not statistically significant. Interaction terms between BRI and subgroup variables were tested in the Cox proportional hazards models. A significant multiplicative interaction was observed for sex, whereas no significant interactions were found for age, educational attainment, or hypertension (Table 3).

Sensitivity analyses assessed the robustness of primary findings. Analysis 1 excluded 179 participants who died within 2 years of enrollment, with Cox models adjusted for Model 3 covariates. Compared to Q1, adjusted HRs (95% CIs) for Q2–Q4 were 1.03 (0.92–1.15), 1.11 (0.99–1.24), and 1.21 (1.07–1.37) (P for trend < 0.001). Analysis 2 included 1,756 cancer patients, yielding HRs of 0.99 (0.90–1.09), 1.07 (0.97–1.18), and 1.19 (1.07–1.33) (P for trend < 0.001) for Q2–Q4 versus Q1 after full adjustment (Table 4).