Introduction
Rapid assessment of pre-stroke function is essential to determine eligibility for time-critical, hyper-acute reperfusion therapies. The modified Rankin Scale (mRS) is the gold standard post-stroke outcome measure but is less reliable for measuring pre-stroke function. This study will investigate the psychometric properties of a new measure, the DBS24, which assesses independence in complex daily activities.
Methods
Prospective patients in this single-centre observational study provided opt-out consent. A waiver was approved for retrospective data (Ethical approval SWSLHD HREC, 2022/ETH00483). Data collection included demographic/medical data and pre/post-stroke DBS24 and mRS. Data analyses (item response theory and classical test theory) were conducted using R and Winsteps.
Results:
Data included 334 DBS24 measures collected from 173 (51.8%) participants with pre-stroke DBS24, and remainder post-stroke. Participants (median age=72,IQR=63-82) were mostly male (57.5%), experienced LVO stroke (90.9%), and born overseas (57.5%).
Cronbach’s alpha (α=0.88) and Rasch item reliability (0.94) supported internal consistency. The Rasch item-person map indicated a ceiling effect, a higher mean for persons than for items, and close item difficulty for ‘Banking’ and ‘Shopping’ (47.0,46.5).
The Rasch infit/outfit statistics summarising items (Mean square [MNSQ]: 0.95/1.17; Standardised [ZSTD]: 0.0/0.1) and for ‘Banking’, ‘Shopping’, and ‘Home for 24 hours’ met Rasch model expectations, providing evidence of uni-dimensionality. Only the Driving/domestics Infit MNSQ was acceptable (1.3, others>2). The Rasch Principal Components Analysis may indicate a second dimension (Eigenvalue=4.000, unexplained variance=65.2%).
Concurrent validity was demonstrated by correlation with pre-stroke (0.78,p<0.001) and three months (0.79,p<0.001) mRS.
Conclusion:
This study provides evidence for preliminary validation of the DBS24.