Algorithm demonstrates efficacy for semiautomated surveillance of SSIs
Click Here to Manage Email Alerts
Gathering information on manual surveillance and clinical practices related to surgical site infections, or SSIs, can lead to the development of algorithms for semiautomated surveillance of these infections, according to findings from a retrospective cohort study.
“Feedback of infection rates from surveillance data to clinicians and other stakeholders is a cornerstone of hospital-acquired infection (HAI) prevention programs. Participation in surveillance programs contributes to the reduction of HAI incidence. . . ,” Stephanie M. van Rooden, PhD, of the Julius Center for Health Sciences and Primary Care at the University Medical Center Utrecht in the Netherlands, and colleagues wrote. “With the large-scale adoption of electronic health records, automation of the surveillance process is increasingly feasible . . . A framework to develop algorithms for semiautomated surveillance, applicable to routine care data and not requiring complex data-driven modeling, may facilitate the development of reliable algorithms implementable on a larger scale.”
The researchers evaluated the performance of a framework that could be used to develop semiautomated surveillance by comparing the algorithms they developed with routine manual surveillance for the identification of deep incisional and/or organ-space SSIs. The study included three hospitals, one each in the Netherlands, France and Spain, with extensive experience in manual SSI surveillance following total hip arthroplasty (THA) and total knee arthroplasty (TKA), cardiac procedures and other surgical procedures. Those procedures were selected because of how frequently they are performed and because they are often evaluated by surveillance programs, according to van Rooden and colleagues.
Information from questionnaires about manual surveillance and clinical practices was used to create semiautomated surveillance algorithms. Those algorithms, which were standardized for multiple hospitals as well as center-specific applications, categorized the procedures by each center’s probability (high or low) of an SSI. Researchers analyzed the algorithm performance and compared results to traditional surveillance.
Altogether, 4,770 orthopedic procedures, 5,047 cardiac procedures and 3,906 colon procedures were evaluated. Standardized algorithm sensitivity across hospitals ranged from 82% to 100% for orthopedic surgery, from 67% to 100% for cardiac surgery and from 84% to 100% for colon surgery, with a decrease in workload that ranged from 72% to 98%. Center-specific algorithms demonstrated a greater reduction in workload but lower sensitivity.
The pre-emptive algorithm development used in this study requires only a few data sources that can be obtained from EHRs and does not involve complex modeling, calibration or missing data, which suggests that it may be accessible to a wide variety of hospitals, according to van Rooden and colleagues. Several factors may limit the generalizability of the framework, including differences in the availability of high-quality data and registration practices, but studies that evaluate the implementation of the framework could lead to more detailed practical guidance.
“Algorithms with good performance can be developed without the need for specific modeling by each hospital and based on limited data sources only,” the authors wrote. “Further validation could provide insight into the feasibility of broader applications of this method, both in other hospitals and for other targeted HAIs.” - by Erin T. Welsh
Disclosure: The authors report no relevant financial disclosures.
Click Here to Manage Email Alerts