PREreview del WGS-Enabled Surveillance Improves Detection Of Transmission Events Within A Large Tertiary Care Hospital Trust In London

Summary:

This study evaluates whether adding whole-genome sequencing (WGS) to routine infection prevention and control (IPC) surveillance improves the detection of carbapenem-producing Enterobacterales (CPE) transmission in a large hospital system in London. The authors combine patient movement data with genomic data from bacteria and plasmids across two outbreaks with different characteristics. They found that standard IPC methods detected only about 20% of transmission events identified by genomic data. WGS-based surveillance could detect transmission 25 to 47 days earlier and may reduce costs, with estimated savings of up to £3.6 million annually.

Strengths:

1. Captures transmission pathways that standard genomic methods miss

A major strength is the inclusion of plasmid analysis alongside standard genome comparisons. This enables detection of cross-species transmission driven by shared resistance plasmids, which would likely be missed using genome-based methods alone. This adds important nuance to how transmission is conceptualized and highlights a potential blind spot in standard genomic surveillance.

2. Use of two distinct outbreak datasets improves generalizability

The study includes two outbreaks that differ in duration, species composition, and resistance mechanisms. This allows the authors to show that the performance of IPC surveillance varies depending on the outbreak context, which is highly relevant for real-world public health settings.

Major Concerns:

1. Early detection estimates do not account for real-world sequencing delays

The reported 25 to 47 day earlier detection assumes near real-time availability of genomic data. In practice, sequencing, processing, and interpretation often take 7 to 14 days or longer in routine healthcare settings. This delay is not trivial, since transmission chains for organisms like CPE can expand rapidly within days through patient transfers and shared hospital environments. A delay of even one week could allow multiple additional exposures, reducing the practical benefit of earlier detection and limiting the ability to intervene in time. The authors could strengthen their conclusions by conducting a sensitivity analysis that models realistic turnaround times and evaluates how much of the detection advantage remains under these conditions.

2. Plasmid analysis approach may introduce uncertainty in key findings

The study uses only the largest contig from each plasmid assembly, which may not fully capture complex plasmid structures. This could affect estimates of cross-species transmission, which is one of the study’s main contributions. The authors could assess how this choice influences their results, for example, by comparing with long-read sequencing data on a subset of samples or by discussing the potential direction and magnitude of bias more explicitly.

3. Cost-effectiveness analysis may not reflect real implementation costs

The economic analysis assumes access to existing sequencing infrastructure and does not include costs such as staffing, training, and system setup. This may overestimate the return on investment, especially for hospitals without current sequencing capacity. Including a scenario that accounts for implementation costs in a setting without existing infrastructure would make the findings more applicable to a wider range of healthcare systems.

Overall Recommendation:

This study makes an important contribution by demonstrating how WGS, particularly plasmid analysis, can improve the detection of transmission events in hospital settings. However, key conclusions about early detection and cost-effectiveness would be stronger if they accounted for real-world implementation constraints. Addressing these issues would improve the study’s relevance for public health decision-making.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.