# The impact of integrated genomic surveillance on non-typhoidal Salmonella infection in Australia: an ecological study

## TL;DR

This study shows that using whole genome sequencing in Australia reduced non-typhoidal Salmonella cases by 11.6% to 17.5%, saving millions annually.

## Contribution

The study provides real-world, large-scale evidence of WGS effectiveness in NTS surveillance, filling a gap in existing literature.

## Key findings

- WGS implementation was linked to a 11.6%–17.5% reduction in NTS cases.
- The use of WGS averted 7200–10,900 NTS cases annually, saving US$11.3 m–US$17.0 m per year.
- Advanced statistical methods confirmed the robustness of the observed effects.

## Abstract

Whole Genome Sequencing (WGS) is a powerful technology for monitoring and detecting outbreaks of infectious pathogens, including non-typhoidal Salmonella (NTS). Despite its higher cost than traditional typing methods, WGS offers numerous advantages, including higher resolution and potentially quicker turnaround time. However, evidence regarding its effectiveness in NTS surveillance has predominantly stemmed from micro-simulations or small-scale data. Notably, a recent systematic review identified a lack of real-world, large-scale evidence on the impact of WGS application in NTS surveillance. Our study fills this gap by estimating the effects of WGS on NTS surveillance in Australia using national notifiable disease datasets.

The main dataset was the National Notifiable Diseases Surveillance System (NNDSS) for NTS from 2009 to 2024. The treatment variable was defined as a binary variable representing the period when WGS was implemented in each jurisdiction of Australia. To minimise the effects of unobserved confounders, we employed a two-stage difference-in-difference (2sDiD) approach. This method estimated the parameters of state and period fixed-effects and then adjusted the observed outcomes from those fixed-effects in the first stage. The average treatment effect was obtained in the second stage by regressing the adjusted outcome against the treatment in the second stage. We also conducted a sensitivity analysis using a multi-period DiD model with a double machine-learning estimator.

Compared to the pre-WGS periods, the introduction of WGS was associated with an average of 11.6% reduction in NTS cases when a static specification was applied. Results of a dynamic specification were slightly higher, with a 12.7% reduction in NTS cases after WGS. The estimated effects increased to 17.5% when a multi-period DiD model with a double machine learning estimator was applied.

Our study shows that WGS was associated with a significant reduction (11.6%–17.5%) of NTS cases in Australia. Using the cost and break-even point of NTS from previous Australian studies, our findings suggest that WGS is associated with 7200–10,900 cases of NTS averted, saving US$11.3 m–US$17.0 m per year.

Australian National Health and Medical Research Council, Medical Research Futures Fund (FSPGN00049), and Investigator Grant (GNT1196103) to BPH.