Geographic Variations of Stroke Hospitalization across France: A Diachronic Cluster Analysis.

Abstract

Background This study evaluates the clustering of hospitalization rates for stroke and compares this clustering with two different time intervals 2009-2010 and 2012-2013, corresponding to the beginning of the French National Stroke Plan 2010–2014. In addition, these data will be compared with the deployment of stroke units as well as socioeconomic and healthcare characteristics at zip code level. Methods We used the PMSI data from 2009 to 2013, which lists all hospitalizations for stroke between 2009 and 2013, identified on the most detailed geographic scale allowed by this database. We identify statistically significant clusters with high or low rates in the zip code level using the Getis-Ord statistics. Each of the significant clusters is monitored over time and evaluated according to the nearest stroke unit distance and the socioeconomic profile. Results We identified clusters of low and high rate of stroke hospitalization (23.7% of all geographic codes). Most of these clusters are maintained over time (81%) but we also observed clusters in transition. Geographic codes with persistent high rates of stroke hospitalizations were mainly rural (78% versus 17%, P < .0001) and had a least favorable socioeconomic and healthcare profile. Conclusion Our study reveals that high-stroke hospitalization rates cluster remains the same during our study period. While access to the stroke unit has increased overall, it remains low for these clusters. The socioeconomic and healthcare profile of these clusters are poor but variations were observed. These results are valuable tools to implement more targeted strategies to improve stroke care accessibility and reduce geographic disparities.