Development and Evaluation of the Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection

Abstract

Catheter-associated urinary tract infection is one of the most common healthcare-acquired infections. It is important to institute preventive measures such as surveillance of the appropriate use of indwelling urinary catheters and timely removal by identifying patients at high risk for catheter-associated urinary tract infection. The purpose of this study was to develop an Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection and evaluate its predictive validity. This study involved secondary data analysis based on a case-control study and used the data extracted from electronic health records. The Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection was developed using a risk-scoring algorithm that was based on a logistic regression model and integrated into the electronic health records. The following eight risk factors for urinary tract infection were included in the logistic regression model: length of stay, admission to the Intensive Care Unit, dependent physical activity, highest neutrophil level (%), lowest blood sodium level of less than 136 mEq/L, lowest blood albumin level of less than 3.5 g/dL, highest blood urea nitrogen level of greater than 20 mg/dL, and indwelling urinary catheter application period (days). The risk groups classified by the Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection were automatically displayed on the patient summary screen of the electronic health record. The predictive validity of the Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection gradually increased up to the fifth and sixth assessment data after patients' admission; then, it leveled. It is possible to allocate nurses' time and effort for catheter-associated urinary tract infection risk assessment to surveillance of the use, removal, and management of indwelling urinary catheters and education and training by using the Automated Risk Assessment System for Catheter-Associated Urinary Tract Infection in clinical settings.

Evaluation of manual and electronic healthcare-associated infections surveillance: a multi-center study with 21 tertiary general hospitals in China

“…However, there was a large underestimation of the risk of HAIs. ESS were wildly utilized to understand the nosocomial infections development (17)(18)(19)(20)(21)(22). To date, our study firstly adopted large sample, multi-center studies to overall assess the impact of SIPS in the diagnosis of HAIs.…”

Prediction of risk of acquiring urinary tract infection during hospital stay based on machine-learning: A retrospective cohort study

“…More in line with our study, Hur et al aimed to identify high-risk patients for catheter-associated UTI at an acute care hospital and to develop an automated risk assessment system for effective preventive measures against catheter-associated UTI [12]. Their study comprised 2150 patients of whom one fifth had a diagnosed catheter-associated UTI and the remaining patients were chosen as a non-UTI group, 1505 patients (70%) constituted the training set.…”

“…Two studies have dealt specifically with the use of machine learning in relation to predicting the risk of urinary tract infections [12,25]. Taylor et al using a retrospective cohort study design examined algorithms for prediction of the outcome of a positive urine culture result (>10 4 colony forming units/mL) in a group of emergency department patients having clinical symptoms attributable to UTI and with urine samples taken for culture.…”

“…The adoption of Electronic Medical Records (EMRs) in many hospitals and advances in data science creates an opportunity to develop big data-driven machine-learning algorithms [8,9]. Trained and evaluated on patient cases derived from automated registration and surveillance of hospitalacquired infections with the purpose of predicting individuals admitted to hospital with a greater risk of developing serious infections such as surgical site infections, bacteremia, and HAI-UTI during admission [10][11][12][13]. A real-time early warning system for monitoring inpatient disease risk should reflect underlying clinical causality and thereby enable caregivers to provide clinical intervention or treatment, e.g., in case of inpatient mortality risk [14].…”

Derivation and validation of a nomogram for predicting nonventilator hospital-acquired pneumonia among older hospitalized patients

“…A longer "time at risk" usually means a longer period of bed rest and hospitalization. Previous studies have shown that hospital stay duration is an independent predictor of healthcare-acquired infections [33,34]. Therefore, we hypothesize that "time at risk" may help predict NV-HAP occurrence by reflecting the length of hospital stay of the patients.…”