JMIR Publications

ABSTRACT

Background:

Pressure ulcers can be a painful and costly complications of hospital care. Hospital acquired pressure ulcers (HAPU) may cause not only additional length of hospital stay and associated care costs but also undesirable patient outcomes. Despite these technical advances in healthcare, that may support the documentation and variables responsible for HAPUs, the rates of incidence of HAPUs are still unacceptably high. We hypothesize that the care team’s decisions relative to pressure ulcer risk assessment and prevention may be better supported by a data-driven, ICU-specific prediction model of HAPUs.

Objective:

The purposes of this study was to build a prediction model for HAPU development in the ICU by applying multiple logistic regression using cumulative electronic health record data of patients in critical care settings.

Methods:

We conducted a retrospective cohort study by using a cumulative electronic health records of ICU patients in order to develop a prediction model for better risk assessment. Bivariate analyses were performed to identify risk factors for ICU HAPU development. Multiple logistic regression was used to develop a prediction model with significant predictors from the results of the bivariate analyses.

Results:

The total number of patient encounters studied were 12,654. The number of patient encounters who had HAPU as a discharge diagnosis was 735 (5.8%). As a result of logistic regression, gender (Male), diabetes, isolation, length of ventilator days, and Braden score were significant. The overall accuracy of the model was 91.7% and the AUC was .737. Male patients were 1.46 times more, patients with diabetes were 1.49 times more, patients under isolation had 3.10 times more likely to have a HAPU than patients who were not.

Conclusions:

Hospital acquired pressure ulcers are potentially avoidable but early identification of patients at higher risk is imperative in the ICU. Using an extremely large, EHR derived data set allowed us to compare characteristics of patients who develop a HAPU during their ICU stay relative to those who did not and develop a prediction model from the empirical data. The model may support better prediction of HAPU development for avoidable reductions in pressure ulcer incidence in intensive care.