关键词: 数据库匹配; 卒中队列; 死亡

Abstract:

Objective To explore the data matching profile between stroke cohort in large multicenter study and the administrative data, and to provide new ideas and methods for data quality control in similar clinical studies. Methods The data of 12 112 inpatients with ischemic stroke who completed the 90-day followup from multicenter clinical trial from 2015 to 2017 were selected. The selected data were matched with all the deaths in the China National Disease Surveillance Points System from 2015 to 2017. The differences in regional and hospital distribution and patients characteristics were compared and analyzed between the group of unmatched patients with no death information in stroke cohort study while who had died as shown in China National Disease Surveillance Points System and the group of matched patients with the same death information in the two data. Results There were no statistically significant differences in regional and hospital distribution and patients characteristics between the two groups. However, according to the data trend, the unmatched rate of death information in the patients with NIHSS ≤3 on admission was higher than that in the patients with NIHSS >3 on admission (11.36% vs 6.56%), and the unmatched rate of death information in the patients living alone was higher than that in the patients living with family (13.16%vs 7.65%). Conclusions It is more difficult to obtain accurate death outcome information in patients with relatively mild neurologic impairment and patients living alone after discharge. The database matching method can make up for the lack of follow-up information in large multicenter cohort studies, which can provide accurate death outcome information of patients for stroke cohort studies.

Key words: Database matching; Stroke cohort study; Death