Mapping acute encounters in end-stage renal disease: a multi-scale network analysis of presenting reasons and diagnoses

作者：Hairong Wang, X Zhang · 发表于：Network Modeling Analysis in Health Informatics and Bioinformatics · 年份：2026 · DOI：10.1007/s13721-026-00848-7 · 研究领域：Emergency and Acute Care Studies、Healthcare Systems and Technology、Clinical Reasoning and Diagnostic Skills

Abstract Emergency department (ED) encounters among patients with end-stage renal disease (ESRD) are clinically complex and multisystemic. Conventional frequency summaries do not capture the co-occurrence structure linking presenting reasons to downstream diagnoses, limiting their utility for system-level triage and pathway design. We performed a retrospective, encounter-level multi-scale network analysis of NHAMCS-ED 2020–2022 visits with the chronic condition indicator ESRD = 1 ( n = 533). We constructed ICD-10-CM chapter and Top-50 diagnosis code co-occurrence networks and assessed symptom-to-diagnosis convergence using bipartite heatmaps. Primary edge weights were raw co-occurrence counts (n₁₁); robustness was assessed using Jaccard, cosine, ϕ, odds ratio, and relative risk, with sensitivity checks for explicit N18.6 coding and pre-pandemic 2018–2019 encounters. ESRD-indicated ED encounters showed a renal-centered architecture. The strongest chapter-level links were Genitourinary System–Symptoms/Signs not elsewhere classified (n₁₁ = 111) and Genitourinary System–Circulatory System (n₁₁ = 110). In the Top-50 network, N18.6 (end-stage renal disease) was the dominant hub, and the strongest edge was N18.6–Z99.2 (dependence on renal dialysis; n₁₁ = 64). Heatmaps showed convergence from broad symptoms to the renal axis, led by General Symptoms–N18.6 (n₁₁ = 62) and Respiratory Symptoms–N18.6 (n₁₁ = 55). Although all encounters were ESRD-indicated, N18.6 appeared in the five avai...