Implementing an Automated Pneumonia Surveillance System

Abstract

Background: Although definitions from the CDC were developed to increase the reliability of surveillance data, reduce the burden of surveillance in healthcare facilities, and enhance the utility of surveillance data for improving patient safety, the algorithm is still laborious for manual use. We implemented an automated surveillance system that combines 2 CDC pneumonia surveillance definitions to identify pneumonia infection in inpatients. Methods: The program was implemented at an academic health center with >40,000 inpatient admission per year. We used Window Task Scheduler with a batch file daily to run a validated pneumonia surveillance algorithm program written with SAS version 9.4 software (SAS Institute, Cary, NC) and a natural language processing tool that queries variables (Table 1) and text found in the electronic medical records (EMR) to identify pneumonia cases (Fig. 1). We uploaded all computer-identified positive cases into a Microsoft Access database daily to be reviewed by a hospital epidemiologist. Every week, we also validated 5 computer-identified negative cases from the prior 2 weeks to ensure accuracy of the computer algorithm. We defined negative cases as pneumonia present on admission or chest x-ray indicative of pneumonia but without CDC-defined surveillance symptoms. We also wrote a program to automatically send e-mails to key stakeholders and to prepare summary reports. Results: Since November 2019, we have successfully implemented the automated computer algorithm or program to notify, via e-mail, infection prevention staff and respiratory therapy providers of CDC-defined pneumonia cases on a daily basis. This automated program has reduced the number of manual hours spent reviewing each admission case for pneumonia. A summary report is created each week and month for distribution to hospital staff and the Department of Health, respectively. Conclusions: The implementation of an automated pneumonia surveillance system proves to be a timelier, more cost-effective approach compared to manual pneumonia surveillance. By allowing an automated algorithm to review pneumonia, timely reports can be sent to infection prevention control staff, respiratory therapy providers, and unit staff about individual cases. Hospitals should leverage current technology to automate surveillance definitions because automated programs allow near real-time identification and critical review for infection and prevention activities.

Funding: None

Disclosures: None