NIH Stroke Scale and Age Predict Early Post-Stroke Cognitive Impairment

Abstract

Introduction: Acute post-stroke cognitive deficits, which may precede formal post-stroke cognitive impairment (PSCI), lacks evidence-based interventions or guidelines. Up to one-third of PSCI patients progress to dementia within five years. Deficits often emerge within two weeks, underscoring the need for early recognition of risk factors to guide prevention and management. However, predictive value of demographic, clinical, and stroke-specific factors remains inconsistent. We evaluated cognitive outcomes (Montreal Cognitive Assessment) at discharge and identified predictors of low function. Methods: In this retrospective study at the University of Oklahoma Medical Center, we reviewed 964 stroke patients, with 168 meeting inclusion criteria. Early PSCI was defined as Montreal Cognitive Assessment <26 at hospital discharge post-stroke. We calculated prevalence, performed univariable logistic regression, and developed multivariable logistic regression models with and without interaction terms. Results: Patients with Montreal Cognitive Assessment <26 were older, had longer hospital stays, higher NIH Stroke Scale, and more often discharged to non-home settings. Prevalence of early PSCI was higher with age, longer stays, non-white race, higher NIH Stroke Scale, and non-home discharge. Univariable analyses revealed length of stay, discharge disposition, NIH Stroke Scale, and race as strongest associations. Final multivariable model (NIH Stroke Scale, age, length of stay, discharge disposition) demonstrated NIH Stroke Scale and age as significant predictors, with good discrimination (Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) = 0.79). Models including interaction terms performed similarly (ROC-AUC = 0.78). Conclusion: Consistent with prior work, NIH Stroke Scale and age emerged as the most robust predictors of early PSCI. Length of stay, discharge disposition, and race were associated with early PSCI in univariable analyses and warrant further evaluation in larger prospective studies.