Statistical outbreak detection by joining medical records and pathogen similarity

Area of Science:

• Infectious disease epidemiology
• Statistical modeling
• Genomic epidemiology

Background:

• Hospital-associated infections (HAIs) pose a significant threat to patient safety.
• Effective outbreak detection and characterization are crucial for infection control.
• Current methods may lack the sensitivity or specificity to identify all outbreaks and their origins.

Purpose of the Study:

• To develop a statistical inference model for detecting and characterizing hospital-associated infection outbreaks.
• To integrate patient exposure data from electronic medical records with pathogen genomic similarity.
• To simultaneously identify probable outbreaks and their root causes.

Main Methods:

• Developed a statistical inference model combining patient exposure data and whole-genome sequencing (WGS) data of pathogens.
• Utilized electronic medical records to determine patient exposures.
• Employed WGS to assess pathogen similarity.
• Demonstrated targeted sequencing strategies to optimize WGS utility.
• Incorporated a method for learning model parameters from known outbreak data.

Main Results:

• The model successfully detects and characterizes hospital-associated infection outbreaks.
• Integrating patient exposures and pathogen genomics improves outbreak identification and root-cause analysis.
• Targeted WGS strategies enhance outbreak detection even with incomplete sequencing data.
• Model parameters can be effectively learned from reference data of known outbreaks.
• Model performance was validated using semi-synthetic experiments.

Conclusions:

• The presented statistical model offers a robust approach for identifying and understanding hospital-associated infection outbreaks.
• The integration of diverse data sources (electronic health records and WGS) is key to improving outbreak surveillance.
• The model's ability to guide targeted sequencing provides a cost-effective strategy for enhancing detection capabilities.
• This framework has the potential to significantly improve infection prevention and control measures in healthcare settings.