Co-occurrence network characteristics and key comorbidity node identification based on health examination indicators among rural older adults aged 65 and above

BACKGROUND: To analyze the co-occurrence network characteristics of abnormal health examination indicators among rural older adults aged 65 years and above, identify key nodes and co-occurrence modules, and explore gender- and age-specific differences in network topology, thereby providing hypothesis-generating evidence for integrated health monitoring in aging populations. METHODS: Data were obtained from 4,195 rural adults aged ≥ 65 years who underwent health examinations in a secondary hospital between March and May 2024. Seventeen variables were extracted, including demographic characteristics and fifteen binary, including sex, age, blood pressure, blood glucose, blood lipids, and liver and kidney function, were extracted. The associations between indicators were quantified using the φ coefficient. The Louvain algorithm was applied for modular partitioning, and centrality measures (degree, betweenness, and closeness) were used to identify key nodes. Network characteristics were further compared by gender and age strata. RESULTS: Co-occurrence network analysis of 15 health indicators among 4,195 rural older adults revealed that metabolic abnormalities (dyslipidemia 46.94%, hypertension 39.05%, diabetes 17.64%) and organ structural damage (abnormal ultrasound 14.90%, gallstones 5.37%) were most prevalent, along with significant hematological abnormalities (urinary abnormalities 37.90%). The overall network exhibited high connectivity (density = 0.695; clustering coefficient = 0.812) and was divided into three major modules: a metabolic-related abnormalities module (diabetes, hypertension, fatty liver), an organ structure and function abnormalities module (abnormal ultrasound findings, renal cysts), and an infection and mental health-related abnormalities module(tuberculosis, mental disorders). Stratified analysis showed that females had significantly higher network density (0.933) and clustering coefficient (0.936) than males (P < 0.05). In the ≥ 80-year group, network density increased to 0.971 (23% higher than the < 80-year group), while the modularity coefficient (Q) decreased to 0.225, suggesting more complex co-occurrence patterns of abnormal health indicators in advanced age.Centrality analysis identified fatty liver (Degree = 12) as a bridging node connecting metabolic and organ structural indicators, while diabetes and hypertension ranked highest in betweenness centrality (0.32 and 0.28, respectively), serving as structurally central nodes within the metabolic module. CONCLUSIONS: Abnormal health indicators among rural older adults exhibit a metabolism-centered and systemically interconnected co-occurrence pattern. Fatty liver, diabetes, and hypertension are structurally central nodes in the network, highlighting their prominence rather than implying causality. Females show stronger indicator interconnections, and advanced age is associated with more complex co-occurrence patterns. These findings provide network-based, hypothesis-generating evidence to support integrated health monitoring in rural older adults.