Risk score for predicting recurrence in patients with ischemic stroke: the Fukuoka stroke risk score for Japanese.

BACKGROUND

Currently, there are only a few risk scores to predict long-term recurrence of ischemic stroke and no risk score has been developed for the Asian population. The aim of the present study was to develop a new risk score to predict 1-year recurrence in Japanese patients with ischemic stroke and compare it with the other ones.

METHODS

Data used in this analysis were extracted from the Fukuoka stroke registry (FSR) database, including 3,067 Japanese patients who experienced ischemic stroke. FSR is a multicenter, hospital-based, prospective registry of Japanese patients with acute stroke. Associations between potential clinical variables that could be involved in the increase of risk and recurrence of ischemic stroke within 1 year after initial stroke onset were evaluated using univariate and multivariate analyses. To calculate risk scores, weighted points proportionally based on standardized coefficient values were assigned to variables that were identified to be significant risks for recurrence by multivariate analysis. The Hosmer-Lemeshow goodness-of-fit test was used to test agreement between a predicted outcome using a model and an actual observed outcome. Kaplan-Meier analysis was used to estimate average cumulative recurrence rates within three risk groups. Intergroup comparisons in recurrence rate among the risk groups were performed using a log-rank test.

RESULTS

Univariate and multivariate analyses identified nine significant predictors for 1-year recurrence, to which we assigned the following points: age (65-74 years, 1 point; ≥75 years, 2 points), hypertension (1 point), diabetes mellitus (1 point), smoker (1 point), atrial fibrillation (1 point), cardiac diseases (1 point), chronic kidney disease (1 point), nonlacunar stroke (1 point), and previous ischemic stroke (2 points). The Hosmer-Lemeshow goodness-of-fit test demonstrated good agreement between the observed and predicted recurrence rate (χ(2) = 2.30, p = 0.97). The ROC curve for the risk score models showed acceptable discriminant power with a C-statistic of 0.636 (95% confidence interval: 0.573-0.698). Trends toward increased risk of recurrence with higher scores were observed. The 1-year recurrence rates were 2.93, 5.83 and 7.96% in low- (≤3 points), intermediate (4-5 points), and high-risk groups (≥6 points), respectively. Kaplan-Meier curves with log-rank test showed a significant difference in recurrence among the three risk groups (p < 0.0001).

CONCLUSIONS

A new risk score was developed and successfully validated. Our results suggest that this simple risk score enables clinicians to assess 1-year recurrence risk in Japanese patients with ischemic stroke.