Abstract
PURPOSE: To develop and validate a method of predicting whether patients will develop a heel pressure ulcer during their hospital stay.
DESIGN: This retrospective case-control study used 2 separate data sets, one for an initial analysis followed by a second data set for validation analysis.
SUBJECTS AND SETTING: From 2009 to 2011, medical records of discharged patients with a DRG code for heel pressure ulcers in our urban, tertiary medical center were retrospectively reviewed. Using age as the matching criterion, we then reviewed cases of patients without heel pressure ulcers. The initial analysis comprised 37 patients with hospital-acquired heel pressure ulcers and 300 without. The validation analysis included 12 patients with heel pressure ulcers and 68 without.
METHOD: In order to develop this method of identifying patients with heel pressure ulcers, logistic regression modeling was used to select a set of patient characteristics and hospital conditions that, independently and in combination, predicted heel pressure ulcers. Logistic modeling produced adjusted and unadjusted odds ratios for each of the significant predictor variables. The validation analysis was employed to test the predictive accuracy of the final model.
RESULTS: Initial analysis revealed 4 significant and independent predictors for heel pressure ulcer formation during hospitalization: diabetes mellitus, vascular disease, immobility, and an admission Braden Scale score of 18 or less. These findings were also supported in the validation analysis.
CONCLUSION: Beyond a risk assessment scale, staff should consider other factors that can predispose a patient to heel pressure ulcer development during their hospital stay, such as comorbid conditions (diabetes mellitus and vascular disease) and immobility.