Predicting postoperative surgical site infection with administrative data: a random forests algorithm

Abstract

Background Since primary data collection can be time-consuming and expensive, surgical site infections (SSIs) could ideally be monitored using routinely collected administrative data. We derived and internally validated efficient algorithms to identify SSIs within 30 days after surgery with health administrative data, using Machine Learning algorithms. Methods All patients enrolled in the National Surgical Quality Improvement Program from the Ottawa Hospital were linked to administrative datasets in Ontario, Canada. Machine Learning approaches, including a Random Forests algorithm and the high-performance logistic regression, were used to derive parsimonious models to predict SSI status. Finally, a risk score methodology was used to transform the final models into the risk score system. The SSI risk models were validated in the validation datasets. Results Of 14,351 patients, 795 (5.5%) had an SSI. First, separate predictive models were built for three distinct administrative datasets. The final model, including hospitalization diagnostic, physician diagnostic and procedure codes, demonstrated excellent discrimination (C statistics, 0.91, 95% CI, 0.90–0.92) and calibration (Hosmer-Lemeshow χ2 statistics, 4.531, p = 0.402). Conclusion We demonstrated that health administrative data can be effectively used to identify SSIs. Machine learning algorithms have shown a high degree of accuracy in predicting postoperative SSIs and can integrate and utilize a large amount of administrative data. External validation of this model is required before it can be routinely used to identify SSIs.

Harnessing of Artificial Intelligence for the Diagnosis and Prevention of Hospital-Acquired Infections: A Systematic Review

“…In particular, the authors have developed random forest algorithm to perform a preliminary screening of variables followed by the high-performance logistic regression approach to select top 30 most important predictors for SSIs with point system or risk scores. Using datasets including physician procedure claims, hospital (ICD-10) codes and physician (ICD-9) diagnostic codes, the authors were able to demonstrate a high performance of the random forest algorithm for the prediction of SSIs with a high degree of accuracy [ 147 ]. In more recent research published in 2023, Wu et al described the development of an ML model for the detection of SSIs following total hip and knee arthroplasty in which nine XGBoost models were developed and validated to identify incisional SSIs, organ space SSIs and complex SSIs using administrative data and electronic medical records free text data.…”

Derivation and validation of a clinical risk score to predict death among patients awaiting cardiac surgery in Ontario, Canada: a population-based study

“…We used a hybrid approach of the random forest method for initial variable selection, followed by stepwise logistic regression for clinical interpretability and parsimony. 34 , 35 Details of the random forest method have been described elsewhere. 36 – 38 In short, we used a bootstrap sample of the data to build each of the classification trees.…”

Ending surgical site infection by negative pressure wound therapy (NPWT): A case report

“…Hundreds of millions of people may go through surgery every year worldwide. In the USA, 23 million patients go through surgery every year [1] , [2] . Surgical site infection (SSI), which usually occurs 30 days after surgery, is one of the most common complications of the surgery, despite advances in surgical techniques, understanding the pathogenesis of surgical wound infections, and the widespread use of prophylactic antibiotics [2] .…”

“…In the USA, 23 million patients go through surgery every year [1] , [2] . Surgical site infection (SSI), which usually occurs 30 days after surgery, is one of the most common complications of the surgery, despite advances in surgical techniques, understanding the pathogenesis of surgical wound infections, and the widespread use of prophylactic antibiotics [2] . SSI occurs in approximately 1–3 out of every 100 patients who undergo surgery and often results from bacteria, including streptococci , staphylococci , or other microorganisms [3] .…”