Development and Comparison of Predictive Models for Pressure Injuries in Surgical Patients: A Retrospective Case-Control Study

Abstract

Purpose: The purpose of this study was to develop and compare 3 predictive models for pressure injury (PI) occurrence in surgical patients.

Design: Retrospective case-control study.

Subjects and setting: Data on PI risk assessment and preanesthesia evaluation records from 400 patients (80 patients who developed PIs after surgery and 320 patients who did not) in a South Korean acute care setting who underwent surgery between January 2015 and May 2016 were extracted from the electronic health record.

Methods: Three models were developed with items from the Braden Scale (model 1), the Scott Triggers tool (model 2), and the Scott Triggers tool in addition to type of anesthesia, laboratory test results, and comorbid conditions (model 3) using logistic regression to analyze items (factors) in each model. Predictive performance indices, which included sensitivity, specificity, positive predictive value, negative predictive value, area under the receiver operating characteristics curve, and Akaike information criterion, were compared among the 3 models.

Results: Findings showed there were no significant factors in model 1, the estimated surgery time and serum albumin level were significant in model 2, and the estimated surgery time, serum albumin level, and brain disease were significant in model 3. The model performance evaluation revealed that model 2 was the best fitting model, demonstrating the highest sensitivity (84.4%), highest negative predictive value (94.6%), and lowest Akaike information criterion (302.03).

Conclusions: The Scott Triggers tool in model 2, which consists of simple items that are easy to extract from preanesthesia evaluation records, was the best fitting model. We recommend the Scott Triggers tool be used to predict the development of PIs in surgical patients in acute care settings.