OBJECTIVES: To study the components of two risk assessment scales for decubitus ulcer risk, Waterlow and Braden, and of the Chailey score for the same purpose. DESIGN: Experimental study of patients at risk of developing decubitus ulcers. SETTING: The West Midlands and Yorkshire. SUBJECTS: One hundred and fifty wheelchair users from the West Midlands and 9022 patients from a District General Hospital in York, the latter consisting of all admissions to the hospital in a four-month period. INTERVENTIONS: Braden, Chailey scores (wheelchair users) and Waterlow scores (all subjects) measured. MAIN OUTCOME MEASURES: Development of a pressure sore, receiver operating characteristic (ROC) curves. RESULTS: Waterlow outperformed Braden for classification of wheelchair patients with respect to decubitus ulcer. The Chailey score performed randomly in this group. The sensitivity and specificity as seen in ROC curves was different for Waterlow scores for wheelchair users and general patients, the latter being much better classified. Only three items out of 11 in the Waterlow score appeared to have any classification ability in the wheelchair group. CONCLUSIONS: Risk indicators used for general patients are probably poorly suited for wheelchair users. There is a need for large-scale predictive studies of wheelchair users and other groups to allow regression analysis of the subscales of risk indicators. From the provisional data of this study it appears that splitting patients by gender and into full- and part-time wheelchair users classifies almost as well the much more complicated risk assessment tools currently available.
OBJECTIVES: To study the components of two risk assessment scales for decubitus ulcer risk, Waterlow and Braden, and of the Chailey score for the same purpose. DESIGN: Experimental study of patients at risk of developing decubitus ulcers. SETTING: The West Midlands and Yorkshire. SUBJECTS: One hundred and fifty wheelchair users from the West Midlands and 9022 patients from a District General Hospital in York, the latter consisting of all admissions to the hospital in a four-month period. INTERVENTIONS: Braden, Chailey scores (wheelchair users) and Waterlow scores (all subjects) measured. MAIN OUTCOME MEASURES: Development of a pressure sore, receiver operating characteristic (ROC) curves. RESULTS: Waterlow outperformed Braden for classification of wheelchair patients with respect to decubitus ulcer. The Chailey score performed randomly in this group. The sensitivity and specificity as seen in ROC curves was different for Waterlow scores for wheelchair users and general patients, the latter being much better classified. Only three items out of 11 in the Waterlow score appeared to have any classification ability in the wheelchair group. CONCLUSIONS: Risk indicators used for general patients are probably poorly suited for wheelchair users. There is a need for large-scale predictive studies of wheelchair users and other groups to allow regression analysis of the subscales of risk indicators. From the provisional data of this study it appears that splitting patients by gender and into full- and part-time wheelchair users classifies almost as well the much more complicated risk assessment tools currently available.