The Challenge of Predicting Pressure Ulcers in Critically Ill Patients. A Multicenter Cohort Study.

RATIONALE

Pressure ulcers are preventable events. Critically ill patients are particularly vulnerable. The Braden scale has been used to identify hospitalized patients at high risk for the development of pressure ulcers; however, this predictive tool has not been adequately validated for critically ill patients.

OBJECTIVES

We aimed to validate and improve the Braden scale for critically ill patients by adding clinical variables to the original scale.

METHODS

We conducted a cohort study in 12 intensive care units (ICUs) within a network of hospitals in Brazil during 2013. We excluded patients who stayed less than 48 hours, patients with one or more pressure ulcers at admission, and those who developed a pressure ulcer within the first 48 hours. We evaluated the Braden scale and clinical variables through a competing risk analysis. Discrimination and calibration were evaluated using the Concordance index (C-index) and a calibration plot, respectively. We used bootstrapping to assess internal validation.

MEASUREMENTS AND MAIN RESULTS

Our primary outcome was incident pressure ulcer within 30 days of ICU admission. We analyzed 9,605 patients and observed 157 pressure ulcers (rate of 3.33 pressure ulcers/1,000 patient-days). The majority of pressure ulcers were detected at stage I or II (28.7 and 66.2%, respectively). The Braden scale had good discrimination (C-index, 0.753; 95% confidence interval, 0.712-0.795), although its performance decreased for the most severely ill patients. We derived a modified predictive tool by adding eight clinical variables to the Braden scale: age, sex, diabetes mellitus, hematological malignancy, peripheral artery disease, hypotension at ICU admission, and need for mechanical ventilation or renal replacement therapy in the first 24 hours after ICU admission. The derived score had better discrimination and calibration than the original Braden scale. The best score cutoff was at least 6 points, with a sensitivity of 87% and a specificity of 71%.

CONCLUSIONS

The original Braden scale measured at ICU admission is a valuable tool for pressure ulcer prediction, although it is not accurate for severely ill patients. To overcome the limitations of the original scale, we derived a modified score with better performance, which may identify high-risk ICU patients and support target interventions. External validation of the proposed clinical prediction score is needed.