Thirteen‐year trends in risk scores predictive values for subsequent stroke in patients with acute ischemic event

INTRODUCTION: A high residual risk of subsequent stroke suggested that the predictive ability of Stroke Prognosis Instrument‐II (SPI‐II) and Essen Stroke Risk Score (ESRS) may have changed over the years. AIM: To explore the predictive values of the SPI‐II and ESRS for 1‐year subsequent stroke risk in a pooled analysis of three consecutive national cohorts in China over 13 years. RESULTS: In the China National Stroke Registries (CNSRs), 10.7% (5297/50,374) of the patients had a subsequent stroke within 1 year; area under the curve (AUC) of SPI‐II and ESRS was .60 (95% confidence interval [CI]: .59–.61) and .58 (95% CI: .57–.59), respectively. For SPI‐II, the AUC was .60 (95% CI: .59–.62) in CNSR‐I, .60 (95% CI: .59–.62) in CNSR‐II, and .58 (95% CI: .56–.59) in CNSR‐III over the past 13 years. The declining trend was also found in ESRS scale (CNSR‐I: .60 [95% CI: .59–.61]; CNSR‐II: .60 [95% CI: .59–.62]; and CNSR‐III: .56 [95% CI: .55–.58]). CONCLUSIONS: The predictive power of the traditional risk scores SPI‐II and ESRS was limited and gradually decreased over the past 13 years, thus the scales may not be useful for current clinical practice. Further derivation of risk scales with additional imaging features and biomarkers may be warranted.