Network analysis of the roles of fomites, patients, and healthcare personnel to identify potential superspreaders of healthcare-associated infections (HAIs) within a hospital setting

Abstract

Healthcare-associated infections (HAIs) increase patient morbidity, mortality, and healthcare burdens, yet the complex interplay between patients, healthcare workers, and fomites in transmission dynamics remains poorly understood. Traditional static network models fail to capture temporal fluctuations in contact patterns or the combined dynamics of patients, healthcare personnel, and fomites.

Objective: To investigate the roles of patients, fomites, and healthcare personnel in HAI transmission dynamics within a hospital ward, identify superspreaders using temporal network analysis, and evaluate screening and isolation intervention effectiveness.

Methods: We constructed a directed, weighted, temporal network comprising 94 nodes (18 patients, 6 nurses, 4 doctors, 14 allied health staff, 8 porters, and 44 fomites) based on published contact probabilities and staffing ratios. Bootstrapped temporal exponential random graph models (bTERGM) with 100 replications identified mechanisms structuring contact patterns. A Susceptible-Exposed-Asymptomatic-Infected-Recovered (SEAIR) compartmental network model was developed to simulate Clostridioides difficile transmission across 100 network iterations and 100 SEAIR iterations per scenario (40,000 total simulations). We evaluated screening and isolation interventions at 25%, 50%, and 75% effectiveness, comparing full dynamic, static person-to-person, and homogeneous environmental models.

Results: The networks exhibited moderate density (0.092, 95% CI: 0.088-0.096) and modularity (0.249, 95% CI: 0.233-0.266) indicating moderate tendency of nodes to form triangular relationships and moderate connectivity typical of hospital contact patterns. bTERGM revealed sparse baseline connectivity (OR = 0.062), strong reciprocity (OR = 3.28), and homophily by node type (OR = 1.54) indicating that hospital networks are organized by professional roles and bidirectional relationships rather than random interactions. Nurses had the highest degree centrality (43.9 ± 2.1), followed by patients (37.3 ± 1.8) and doctors (26.0 ± 3.2) highlighting their different superspreader potential in healthcare settings. Nurses, patients and allied health worker groups were in the top 15 superspreaders in the SEAIR model simulations with six, six and three nodes respectively. The static model underestimated outbreak size by 29.0% (p = 2.62 × 10^-15). Screening interventions produced "dose-dependent" reductions: 21.2% at 25%, 44.2% at 50%, and 70.1% at 75% effectiveness, progressively flattening the transmission hierarchy as the top 10 superspreaders infection contribution declined from 80.3% to 57.5% events.

Conclusion: This study demonstrates that patients and connected healthcare workers are the main drivers of HAI transmission, with screening and isolation achieving substantial HAI reductions while simultaneously attenuating superspreader contributions. These findings provide an evidence-based framework for optimizing infection control strategies through targeted patient screening and recognition of structured healthcare worker-patient-fomite interaction patterns.