Predictive validity of the EVARUCI scale to evaluate risk for pressure injury in critical care patients

Objective: To compare the predictive capacity of the current risk assessment scale for pressure ulcers in intensive care (EVARUCI), translated into Brazilian Portuguese, using the Braden scale. Method: This cross-sectional study collected prospective data from adult patients in three intensive care units. The receiver operating characteristic (ROC) and precision-recall curve (PR curve) were used to analyse the predictive capacity for pressure injury (PI) using both predictive values and odds ratios (ORs). Results: The incidence of PIs in the study sample of 324 patients was 14.2%. The area under the ROC curve was 0.807 for EVARUCI and 0.798 for the Braden scale. At a cutoff point of 10 on the EVARUCI scale, sensitivity was 69.6%; specificity 78.4%; positive predictive value 34.8%; and OR 8.3. At a cutoff point of 11 on the Braden scale, sensitivity was 76.1%; specificity 75.9%; positive predictive value 34.3%; and OR 10. The area under the PR curve was 0.396 for the EVARUCI scale and 0.348 for the Braden scale, reflecting a smaller area for both. The F1 score value was 0.476 with 37.5% precision and 65.2% recall for the EVARUCI scale, and 0.473 with 34.3% precision and 76.1% recall for the Braden scale. Conclusion: The EVARUCI scale predictive capacity was similar to that of the Braden scale. However, the precision of both scales was low for the accurate prediction of patients at risk of developing PIs.