Wastewater Surveillance for Carbapenem-Resistant Organism Outbreak Prediction in a Chinese Tertiary Hospital

Carbapenem-resistant organisms (CRO) pose a critical threat to global public health, and hospitals serve as major epicenters for their emergence and spread. Traditional clinical infection surveillance often detects CRO outbreaks only after infections have already occurred, missing the window for early intervention. Wastewater-based epidemiology has demonstrated its early warning potential during the COVID-19 pandemic and is increasingly recognized as a promising tool for antimicrobial resistance surveillance.

Our preliminary 22-day pilot study at Peking Union Medical College Hospital revealed two distinct antimicrobial resistance gene (ARG) dynamics patterns across different hospital areas: the Internal Medicine Ward exhibited a "chronic resistance background" with persistently high abundance of carbapenemase genes (IMP/GES types >5,000 ppm), while the Emergency/Fever Clinic showed "acute pulse outbreaks" characterized by transient 50- to 200-fold surges of mcr-3 and QnrVC genes. These findings suggest that hospital functional areas have fundamentally different ARG profiles with unique temporal signatures.

This prospective observational cohort study aims to establish a wastewater-based early warning system for hospital-acquired CRO outbreaks by conducting longitudinal monitoring across four key functional areas: Outpatient Building, Internal Medicine Ward, Surgical Ward, and Emergency/Fever Clinic at Peking Union Medical College Hospital over a 6- to 9-month period. Twenty-four-hour flow-proportional composite wastewater samples will be collected daily using automatic samplers. Laboratory analyses include ARG large-panel testing (300+ subtypes, daily), metagenomic sequencing (weekly, plus pulse-triggered intensified sampling), and viable bacterial culture with whole-genome sequencing of key isolates.

Concurrently, we will collect de-identified clinical CRO isolates and antibiotic consumption data (Defined Daily Doses) from the corresponding hospital buildings. Multidimensional association analyses will be performed using cross-correlation function analysis, Granger causality tests, and cgMLST-based genomic comparisons to determine the lead time of wastewater ARG signals ahead of clinical CRO diagnoses and to provide direct evidence of clonal homology between wastewater and clinical isolates. An early warning model will be constructed using dynamic thresholds (moving average + 2SD/3SD) and machine learning algorithms.

This study integrates environmental, clinical, and pharmaceutical data following the One Health framework. By establishing a replicable building-level wastewater resistome surveillance protocol, this research is expected to provide hospitals with a proactive tool for early CRO outbreak detection, enabling timely infection prevention and control measures. All clinical data will be de-identified, and the study has been designed to pose no greater than minimal risk to patients, with a waiver of informed consent sought in accordance with relevant ethical regulations.