P-1022. Modeling the Impact of Sampling Intensity on Observing C. difficile Transmission in Healthcare Settings

TL;DR

This study shows that increasing the frequency of testing in healthcare settings improves the detection of C. difficile transmission events, which are often missed with standard weekly testing.

Contribution

The study introduces a stochastic transmission model to quantify how sampling intensity affects the observability of C. difficile transmission.

Findings

Weekly sampling detects only 21.9% of cases and 6.8% of transmission pairs.

Daily testing improves detection to 65.0% of cases and 55.2% of transmission pairs.

Transmission events often span multiple weeks, indicating pathogen persistence in healthcare settings.

Abstract

Whole genome sequencing (WGS) is increasingly used to investigate healthcare-associated infections, yet prior studies have only been able to link a small percentage of C. difficile cases. It remains unclear whether this is due to low sampling intensity or because transmission events are occurring outside of monitored healthcare settings.

We developed a stochastic transmission model integrated with an observation model to simulate C. difficile spread and case detection under varying sampling intensities. The model was originally fitted to data from a six-month prospective cohort study with weekly active surveillance sampling. Retrospective analyses of the data were combined with prospective simulation-based analyses using synthetic testing datasets at three sampling intensities (standard, intermediate, and intensive). Transmission trees were constructed to examine transmission events’ temporal, spatial, and network-scale dynamics.

At the standard sampling level (weekly testing), an estimated 21.9% of cases and only 6.8% of transmission pairs were observed. Increasing sampling frequency to daily testing improved observability to 65.0% of cases and 55.2% of transmission pairs. Transmission events occurred across diverse time scales, with 95.5% of modeled transmission trees containing at least one pair separated by four or more weeks, highlighting the pathogen's persistence in healthcare environments. Cluster analyses revealed that over half of transmission trees contained at least one cluster linking 15 or more new colonizations.

Our findings demonstrate that substantial transmission occurs but that weekly sampling is insufficient to capture most C. difficile transmission events. Enhanced sampling intensity substantially improves the observability of cases and transmission pairs, underscoring the need for more comprehensive surveillance protocols. Additionally, modeling approaches can augment genomic epidemiology by illuminating unobserved transmission pathways, allowing for improved tree reconstruction.

All Authors: No reported disclosures