PREreview of Sewer monitoring for antimicrobial resistance genes and organisms at healthcare facilities

By Mam Roheya Jack and Kian Sahabi

Review of the Study on Sewer Monitoring for Antimicrobial Resistance Genes at Healthcare Facilities

DOI: https://doi.org/10.1101/2025.03.16.25324079

This paper provides a comprehensive analysis of sewer monitoring as a promising tool for tracking antimicrobial resistance (AMR) in healthcare settings, focusing on multidrug-resistant organisms (MDROs) and associated genes, particularly Candida auris and carbapenemase-producing organisms (CPOs). The authors conducted an experimental study evaluating various sampling, concentration, and PCR detection methods to identify key MDROs in wastewater from a long-term acute care hospital in Chicago. They evaluated multiple methods for wastewater collection (passive, composite, and grab), concentration (nanoparticles, filtration, and centrifugation), and PCR quantification (real-time quantitative PCR vs. digital PCR) for C. auris and 5 carbapenemase genes (blaKPC, blaNDM, blaVIM, blaIMP, and blaOXA-48-like)

STRENGTHS:

• While wastewater surveillance for Candida auris is a timely and relevant public health measure, prior studies have already demonstrated the feasibility of this  C. auris detection in wastewater. The strength and novelty in this paper comes from the comparison of different methods of wastewater sampling and detection, which could render this technology more feasible to implement in healthcare settings.

• The researchers used a robust, longitudinal design, sampling wastewater biweekly for six months. They examined different sample collection techniques (passive, composite, and grab), treatment (nanoparticles, filtration, and centrifugation), and detection (real-time and digital PCR), showing the study's rigor. Moreover, they also emphasize the reproducibility of their technique: "Experiments were done in triplicate with gene copy, variance, and number of detections between triplicates used to determine method efficacy".

• The research results on the comparison between qPCR and dPCR show the technical challenges and advancements in molecular detection. The authors highlight the improved sensitivity of dPCR: "dPCR was more sensitive while qPCR was unable to reliably detect blaNDM and blaOXA-48-like".

• The study found a recurring presence of significant resistance genes, including C. auris and carbapenemase genes, in healthcare facility wastewater. They also explored the clinical relevance of these findings.

• One of the major limitations of the study was highlighted: "Although the purpose of WWS development is to detect patient shedding in healthcare facilities, longitudinal patient sampling (e.g., repeated point prevalence surveys) was not performed in this study". The authors' acknowledgment of study limitations, including that additional research would be required to prove clinical significance, strengthens the paper. They recognize that while early culture data on Candida auris and carbapenemase-producing organisms in wastewater have been reported, longitudinal studies are needed for assessing clinical relevance and setting baseline data. Being receptive and responsive to what such future research opportunities might look like adds credibility and value to the study so that clear guidance on how to increase application of wastewater surveillance for public health monitoring can be provided.

GENERAL LIMITATION:

• The absence of patient-specific longitudinal data makes it challenging to directly correlate wastewater detections with clinical outcomes. The authors appropriately acknowledge this limitation: "Although the goal of establishing WWS is to detect patient shedding in healthcare facilities, this study did not incorporate longitudinal sampling of patients." While this is a reasonable constraint, it would strengthen the manuscript if the authors briefly discussed how future studies might address this gap.

MAJOR/MINOR REVISIONS:

• Although the authors admit: "This study was limited by a relatively small sample size of 50 regular collections and 10 each of the transport and storage collections" making the study more credible, the relative sample size limits the potential impact on generalizability of the results.  I recommend  considering conducting post hoc power analyses or reporting confidence intervals around key estimates to provide insight into the precision of their findings. 

• The article mentions the use of "multilevel models to account for repeated measurements within collection date and within triplicates" in certain analyses, and "linear models" in others. Use of multilevel models (hierarchical models) is appropriate for repeated measures and nested data (e.g., multiple measurements taken on the same collection date or triplicate samples). However, the paper falls short in providing adequate details on assumptions of the models in question, e.g., residual normality, random effects, and whether potential confounders were accounted for and how. In the main text the authors should address these assumptions explicitly and note how they were validated and which tests were conducted for model fit.

• Even though statistical significance (p-values) is required, it is also extremely important to report effect sizes, especially for complex models with multiple predictors. The authors should state in the main text  the size of differences between methods (e.g., PCR methods, collection methods) in addition to the p-values.

• The analysis excludes non-detects in certain instances, treating them as "false zeros." While it is common practice with environmental data, this assumes that non-detections are exclusively due to technical limitations and not to true absences. This could underestimate the true prevalence of pathogens. An alternative approach such as zero-inflated models could have been employed to treat non-detects more analytically.

• Since multiple methods were evaluated (e.g., different collection techniques, concentration methods, and PCR quantification methods), there is a potential risk of multicollinearity in the statistical models. Multicollinearity can distort relationships between variables and make it difficult to interpret the effect of each individual method. The authors should  test for multicollinearity (e.g., using Variance Inflation Factor, or VIF) to ensure that the statistical models accurately capture the relationships between the methods.

CONCLUSION: 

While the research is not novel in terms of pathogen detection, it makes a significant contribution by comparing multiple methods for wastewater sampling, concentration, and PCR detection, which enhances the practical application of this technology in healthcare. The authors’ transparency in acknowledging the limitations of their study, such as the small sample size and the absence of longitudinal patient data, strengthens the paper’s credibility and provides clear directions for future research. With MINOR REVISIONS to address some statistical considerations and further expansion of the clinical relevance of their findings, this paper could serve as a solid foundation for advancing wastewater surveillance as a public health tool.

Competing interests

The authors declare that they have no competing interests.