Sundermann, Alexander John
(2022)
Novel Approaches for Healthcare Outbreak Detection and Investigation.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
Methods for detecting outbreaks in healthcare settings have remained unchanged for many years. Often this involves the use of geo-temporal clustering which looks for an increase in the number of expected infections within a small timeframe in a confined location. This approach often misses transmission where it did occur and mis-identifies transmission where it did not occur. Additionally, other routes of potential transmission, such as shared providers or procedures, are often not considered. These data are readily available within the electronic health record (EHR). Traditional infection prevention methods often use whole genome sequencing (WGS) at the end of an outbreak to confirm or refute its presence, referred to as reactive sequencing.
The objective of this dissertation is to create and evaluate the Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT), which better detects and investigates outbreaks compared to traditional infection prevention methods. EDS-HAT combines WGS surveillance with machine learning (ML) of the EHR. The creation of EDS-HAT was performed in three steps. First, we performed a systematic review of institutions performing WGS surveillance and/or machine learning of EHR data to obtain a better understanding of EDS-HAT’s use and implications. We found that very few institutions were performing WGS surveillance or machine learning of EHR, yet both had profound impact on outbreak detection and investigation. Second, we developed and trained a proof-of-concept ML algorithm on past, well-described outbreaks that occurred at our institution. We found that the algorithm could accurately identify the correct transmission route on the second patient in all but one outbreak. Lastly, we performed two years of WGS surveillance to directly compare to traditional infection prevention practice. Based on those results, EDS-HAT, if run in real time, could potentially identity otherwise undetected outbreaks, prevent many infections, save money, and be substantially more effective than traditional infection prevention practice. Overall, our findings support the use of WGS and machine learning of the EHR to detect and investigate outbreaks. If implemented in real-time, EDS-HAT represents a potential paradigm shift in infection prevention to increase patient safety.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
10 May 2022 |
| Date Type: |
Publication |
| Defense Date: |
25 March 2022 |
| Approval Date: |
10 May 2022 |
| Submission Date: |
18 April 2022 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
107 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
School of Public Health > Epidemiology |
| Degree: |
DrPH - Doctor of Public Health |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
Outbreak Infectious Diseases Whole Genome Sequencing |
| Date Deposited: |
10 May 2022 18:02 |
| Last Modified: |
10 May 2022 18:02 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/42641 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}