Background and Objective:
Surveillance of surgical site infections (SSIs) is an essential component of infection prevention and control activities. The standard method is manual review of medical records, although time-consuming and requiring relevant resources. Technological progress allows the use of standardized, semi-automated systems. The purpose of this study is to evaluate validity and feasibility of a semi-automated surveillance for the detection of SSIs in an acute hospital setting and to quantify the workload reduction.

Methods:
To assess SSIs probability, an algorithm based on surgical, microbiological and discharge datasets was designed to identify low-risk procedures (LRP) and high-risk procedures (HRP). Its validity was tested on 1,138 surgical procedures performed in the University Hospital of Ancona in pre-COVID period, between October 1st and December 31st, 2019 and already manually reviewed. We compared semi-automated algorithm performance to the manual routine surveillance by assessing sensitivity, specificity, positive and negative predictive values with 95% confidence intervals (CI). Since the algorithm requires that LRP are not manually reviewed, workload reduction was calculated as the percentage of LRP on the total number of procedures. Stata 15 was used for data analysis.

Results:
Among 81 HRP, 7 had already been identified as SSIs applying the manual method. Among the 1,057 LRP, only one had been identified as SSI. The sensitivity of the semi-automated method was 87.5% (CI 47.3-99.7); specificity was 93.5% (CI 91.8-94.8). Positive and negative predictive values were respectively 8.64%(CI 3.5-17) and 99.9% (CI 99.5-99.9). Workload reduction was 92.9% (CI 91.4-94.4).

Conclusion:
These results are comparable to those of other studies evaluating the performance of algorithms for semi-automated surveillance. Given the workload reduction, our semi-automated surveillance turned out to be feasible in our setting. Validity of our method could be strengthened by applying it to other hospitals and broadly to the surveillance of other healthcare associated infections.