Predicting pain and its association with mortality in patients with stroke

Abstract

Abstract Background and objectives Poststroke pain (PSP) is a prevalent and severe consequence of stroke, encompassing central, neuropathic, and nonneuropathic pain. In this study, we aimed to investigate clinical factors associated with PSP three months after stroke and concurrently explore the association between PSP and one-year mortality. Methods This registry-based study comprised data from stroke patients admitted to three hospitals in Sweden between November 2014 and June 2019. The outcome was PSP three months after stroke. Twelve (out of 28) predictor variables were selected by three machine learning methods, and a multivariable binary logistic regression model was fitted for predicting PSP. The association between PSP and one-year poststroke mortality was examined using Cox proportional hazards models. Results Among 4,160 stroke patients participating in the three-month follow-up, 54.7% reported PSP. Antiplatelet use, diabetes, hemiparesis, sensory deficits, and need for assistance before stroke were significant predictors of PSP. Male sex, being born in Sweden, higher income, and regular prestroke physical activity predicted the absence of PSP. After adjustment for age, sex, region of birth, and stroke severity, patients experiencing PSP had a significantly higher one-year mortality rate than those without pain, and the most severe level of pain (constant pain) was associated with the highest cumulative mortality. Conclusion The study findings indicate treatable factors associated with PSP, which highlight areas of improvement in management strategies. Clinicians should recognize that PSP is associated with increased one-year mortality, emphasizing the importance of pain prevention and treatment for enhanced poststroke outcomes.

Keywords

• Subacute stroke
• Predictors
• Mortality
• Machine learning
• Neuropathic pain
• Patient Outcome Assessment