Temporal Variation in Facility Performance of Process Measures After Transient Ischemic Attack in the Veterans Health Administration: Predictors and Association With Ischemic Stroke Incidence

Levine D, Perkins A, Sico J, Myers L, Phipps M, Zhang Y, Ferguson J, Bravata DM. Temporal Variation in Facility Performance of Process Measures After Transient Ischemic Attack in the Veterans Health Administration: Predictors and Association With Ischemic Stroke Incidence. [Abstract]. Stroke. 2019 Feb 1; 50(Suppl 1):WP361.

Background: Little is known about facility factors that contribute to change in facility performance and how temporal variation in facility performance of transient ischemic attack (TIA) process measures contributes to variation in facility-level ischemic stroke incidence. Methods: Prospective study of 7760 veterans (mean age, 70; 95% male) with TIA presenting to 53 VA facilities in 2014-2018. We examined process measures (high/moderate potency statin and anti-thrombotics < / = 7 days; hypertension control [ < 140/90 mmHg] at 90 days) and facility-level characteristics (percentage of TIA patients admitted, number of TIA patients per year, percentage of TIA patients getting Neurology consultation, and VA facility complexity level). We used multi-level hierarchical linear mixed-effects models with random effects for intercept and slope to measure the association between time-varying facility characteristics, changes in facility performance of process measures over time, and 90-day stroke incidence adjusting for patient factors. Results: Facilities significantly increased statin use and anti-thrombotic use, but not hypertension control over the 4-year study period (Table). Facilities who admitted more TIA patients had significantly greater use of statins and anti-thrombotics. Facilities with higher facility complexity had significantly greater use of statins. Facilities with higher rates of Neurology consultation had significantly lower stroke incidence. Conclusion: Facility characteristics, namely percentage of TIA patients admitted, are independently associated with changes in performance of process measures for TIA. Neurology consultations were associated with lower 90-day stroke incidence over time.