Background: Effective discharge planning is increasingly recognised as a critical component of hospital-based Rehabilitation. The BRASS index is a risk screening tool for identification, shortly after hospital admission, of patients who are at risk of post-discharge problems.
Aim: To evaluate the internal construct validity and reliability of the Blaylock Risk Assessment Screening Score (BRASS) within the rehabilitation setting.
Design: Observational prospective study.
Setting: Rehabilitation ward of an Italian district hospital.
Population: One hundred and four consecutively admitted patients.
Methods: Using classical psychometric methods and Rasch analysis (RA), the internal construct validity and reliability of the BRASS were examined. Also, external and predictive validity of the Rasch-modified BRASS (RMB) score were determined.
Results: Reliability of the original BRASS was low (Cronbach's alpha=0.595) and factor analyses showed that it was clearly multidimensional. A RA, based on a reduced 7-BRASS item set (RMB), satisfied model's expectations. Reliability was 0.777. The RMB scores strongly correlated with the original BRASS (rho=0.952; P<0.000) and with FIM™ admission scores (rho=-0.853; P<0.000). A RMB score of 12 was associated with an increased risk of nursing home admission (RR=2.1, 95%CI=1.7-2.5), whereas a score of 17 was associated to a higher risk of length of stay >28 days (RR=7.6, 95%CI=1.8-31.9).
Conclusion: This study demonstrated that the original BRASS was multidimensional and unreliable. However, the RMB holds adequate internal construct validity and is sufficiently reliable as a predictor of discharge problems for group, but not individual use.
Clinical rehabilitation impact: The application of tools and methods (such as the BRASS Index) developed under the biomedical paradigm in a Physical and Rehabilitation Medicine setting may have limitations. Further research is needed to develop, within the rehabilitation setting, a valid measuring tool of risk of post-discharge problems at the individual level.