AIS patients admitted between July 2008 and May 2014 with elevated baseline FVIII levels and repeat FVIII levels drawn for more than 7 days postdischarge were included. The patients were dichotomized by repeat FVIII level for univariate analysis at 150% and 200% activity thresholds. An adjusted model was developed to predict the likelihood of persistently elevated FVIII levels.
Among 1616 AIS cases, 98 patients with elevated baseline FVIII had repeat FVIII levels. Persistent FVIII elevation was found in more than 75% of patients. At the 150% threshold, the prediction score ranged from 0 to 7 and included black race, female sex, prior stroke, hyperlipidemia, smoking, baseline FVIII > 200%, and baseline von Willebrand factor (vWF) level greater than 200%. At the 200% threshold, the prediction score ranged from 0-5 and included female sex, prior stroke, diabetes mellitus, baseline FVIII level greater 200%, and baseline vWF level greater than 200%. For each 1-point increase in score, the odds of persistent FVIII at both the 150% threshold (odds ratio [OR] = 10.4, 95% confidence interval [CI] 1.63-66.9, P = .0134) and 200% threshold (OR = 10.2, 95% CI 1.82-57.5, P = .0083) increased 10 times.
Because an elevated FVIII level confers increased stroke risk, our model for anticipating a persistently elevated FVIII level may identify patients at high risk for recurrent stroke. FVIII may be a target for secondary stroke prevention.