Aim of our study was to explore the determinants of aphasia recovery in first 3 months following first-ever acute stroke utilizing decision tree analysis.

Recovery of post-stroke aphasia depends on multiple factors. Relative importance of these predictors needs to be explored further.

Screened cases of first-ever acute stroke were recruited. Bengali version of western aphasia battery was used for language examination. Each patient underwent initial language assessment during first week after stroke. Language tests were repeated between 90 and 100 days post-stroke in patients available for follow-up. Severity was assessed by calculating aphasia quotient. Lesion assessment was performed by Magnetic resonance imaging (3T) for ischemic stroke (if not contraindicated) and computed tomography for hemorrhagic stroke. Demographic factors (age, gender, bilingualism; number of years of formal education), lesion-related factors (type of stroke, lesion volume, cortical versus sub-cortical location; site of lesion) and initial severity & type of aphasia were considered independent variables while aphasia recovery (no change versus change to a milder type or complete recovery) was the dependant variable. Chi square automatic interaction detection (CHAID) was performed to evaluate predictor importance and build a decision tree for aphasia recovery.

Among 515 recruited cases, aphasia was observed in 208, of whom 163 patients could be followed up. On univariate analysis, factors associated with change to milder type or recovery were- hemorrhagic stroke (p=0.000); pure sub-cortical stroke (p=0.001) and initial non-severe classification (p=0.000). According to initial aphasia type, most change (p=0.000) was observed with global aphasia while most recovery (p=0.000) was observed with milder aphasia types. In CHAID analysis, most important predictor favoring type change or recovery was initial non-severe classification; which had a predictive value of 67.2%.

This is the first time decision tree analysis has been employed to determine predictor importance for post-stroke aphasia recovery.