Occurrence, predictive models and potential health risk assessment of viable but non-culturable (VBNC) pathogens in drinking water

Abstract

Viable but non-culturable (VBNC) pathogens are prevalent in drinking water systems and can resuscitate under favorable conditions, thereby posing significant public health risks. This study investigated the occurrence of VBNC Escherichia coli and Pseudomonas aeruginosa in source water, tap water, and potable water in eastern China, using propidium monoazide-quantitative PCR and culture-based methods. Multiple linear regression (MLR) and artificial neural network (ANN) models were developed based on conventional water quality indicators to predict VBNC pathogen concentrations. The results indicated that drinking water treatment plants effectively reduced VBNC pathogens by 1–3 log units, however, concentrations ranging from 10⁰ to 10² CFU/100 mL remained in tap and potable water, with detection rates between 83.33% and 100%. Furthermore, potable water contained a higher concentration of VBNC pathogens than tap water, suggesting a potential risk of microbial leakage from water dispensers. The constructed ANN models outperformed than MLR models, with R values greater than 0.8, indicating a strong correlation between measured values and model predictions for VBNC pathogens. ANN models also demonstrated superior accuracy than MLR models in predicting VBNC pathogens across different type of drinking water, achieving accuracies of 88.89% for Escherichia coli and 77.78% for Pseudomonas aeruginosa. The QMRA revealed that annual infection risks and disease burdens from VBNC pathogens in potable water were greater than those in tap water, with both exceeding acceptable safety thresholds. This study emphasizes that the risks posed by VBNC pathogens deserve attention and model predictions provide critical evidence for health risk identification.