Background: Early cognitive screening of stroke patients may identify unmet rehabilitative needs during stroke recovery but validated tools are lacking. We report our experience with the Six Item Screener (SIS) compared to the Montreal Cognitive Assessment (MOCA) and attempt to improve upon its shortcomings in stroke patients.
Hypothesis: Low sensitivity of the SIS for cognitive impairment (CI) in stroke patients can be improved by incorporating visuoexecutive dimensions.
Methods: Patients admitted with ischemic stroke (IS), transient ischemic attack (TIA), intracerebral (ICH) or subarachnoid hemorrhage (SAH) between December 2014 and June 2015 underwent inpatient screening for CI using SIS and MOCA, administered by speech-pathologists if they were alert and not aphasic. Predictive value and sensitivity/specificity cut-offs of SIS for CI (MOCA≤23) were determined. A screening tool, created by adding a clock-drawing task and dropping least important SIS components was developed.
Results: Of 110 (IS/TIA: 56, ICH: 17, SAH: 26) patients who had MOCA and SIS performed at the same visit, 79 patients had CI; other patient characteristics including stroke severity, are described (Fig 1A). The AUC of SIS for CI was 0.78 and comparable across stroke types (AUC for IS: 0.81 ICH: 0.79 SAH: 0.95). SIS≤4 had 46.6% sensitivity for CI while SIS≤5 had 72.6% sensitivity and 80.6% specificity for CI. Excluding ‘year’ and ‘month’ questions of the SIS had no effect on the performance of the screening test (AUC=0.78 without). A 7-item tool (CISS) that included the clock drawing task (3 points) and omitted “year/month” SIS questions had excellent predictive power for CI (AUC=0.89) and comparable across stroke types (AUC 0.89-0.93) (Fig 1B,C). CISS≤6 had 94.5% sensitivity and 49.4% specificity for CI.
Conclusions: The CISS improves upon the low sensitivity of the SIS for CI in stroke patients. A validation study using 3-month neuropsychological testing as the gold-standard is underway.