Dialysis Access Steal Syndrome Risk Model with BMI Interaction Analysis

Supplementary file 1 contains data and analysis for developing a risk prediction model for Dialysis Access Steal Syndrome (DASS). The study by Yan Liu retrospectively analyzed 324 end-stage renal disease patients from March 2023 to June 2025. It identifies key predictors like Fried Frailty Score and BMI, which acts as an effect modifier.

Use Cases

• Predict DASS risk using independent predictors like Fried Frailty Score, BMI, CKD duration, sarcopenia, distal arterial pressure, and surgical experience.
• Analyze non-linear relationships and critical thresholds for BMI (23.3 kg/m²), distal arterial pressure (66 mmHg), and CKD duration (8.2 years) using restricted cubic splines.
• Investigate effect modification by BMI, where predictors like Fried Frailty Score and sarcopenia are significantly associated with DASS only in the low/normal BMI group.
• Validate model performance metrics such as AUC (0.934), Brier score (0.085), and C-index (0.931) through bootstrapping and cross-validation.
• Assess risk stratification by comparing DASS incidence between high-risk (50.00%) and low-risk (3.09%) patient groups.

Strengths

• Model demonstrates high predictive performance with an AUC of 0.934 and a C-index of 0.931.
• Analysis is based on 324 patient records with a clear outcome measure (86 DASS cases, 26.5% incidence).
• Robust validation methods include 1000-sample bootstrapping and 10-fold cross-validation.
• Identifies specific, clinically interpretable thresholds for key continuous variables like BMI and distal arterial pressure.

Limitations

• Small sample size of 324 patients limits generalizability and power for subgroup analyses.
• Retrospective, single-study design may introduce selection bias; geographic and institutional coverage is unknown.
• Dataset is 5.5 MB, suggesting it may contain primarily analysis documentation rather than raw patient-level data.

Provenance

Source

figshare, authored by Yan Liu.

Collection Method

Retrospective clinical study analyzing patient records.

Time Range

March 2023 to June 2025.

Freshness

Data covers patients up to June 2025; file was last updated in April 2026.

Geography

null