“…This was particularly important for the language measures since total FHV scores did not correlate with these outcomes (see Table 3 ). While age and education correlated with only some outcome measures, both variables were also included in each of our models based on the assumption that these demographic variables contribute to the variance in confrontational naming ( Neils et al, 1995 , Zec et al, 2007 , Tsang and Lee, 2003 , Henderson et al, 1998 , Welch et al, 1996 ), content production ( Yorkston and Beukelman, 1980 , Cooper, 1990 , Ardila and Rosselli, 1996 , Pistono et al, 2017 , Mackenzie, 2000 ), and stroke severity—in terms of the effect of premorbid neurological function on stroke outcomes ( Umarova et al, 2021 , Umarova et al, 2019 , Habeck et al, 2017 ). The MLRM for NIHSS is reported in Table 4 (Model B), in Table 5 for picture naming, and in Table 6 for content production on Cookie Theft picture description.…”