Supplementary results for stroke, timing of atrial fibrillation diagnosis, and risk of death

Objective:  To evaluate the prognosis of ischemic stroke patients according to the timing of an atrial fibrillation (AF) diagnosis, we created an inception cohort of incident stroke events and compared the risk of death between stroke patients with a) sinus rhythm; b) known atrial fibrillation (KAF); and c) AF diagnosed after stroke (AFDAS). 

Methods:  We utilized the Penn AF Free study to create an inception cohort of patients with incident stroke.  Mortality events were identified after linkage with the National Death Index through June 30, 2017.  We also evaluated initiation of anticoagulants and antiplatelets across the study duration.  Cox proportional hazards models evaluated associations between stroke subtypes and death.

Results:  We identified 1,489 individuals who developed an incident ischemic stroke event:  985 did not develop AF at any point during the study period, 215 had KAF before stroke, 160 had AF detected ≤6 months after stroke and 129 had AF detected >6 months after stroke.  After a median follow-up of 4.9 [IQR 1.9, 6.8] years, 686 deaths occurred.  The annualized mortality rate was 8.8% in the stroke, no AF group; 12.2% in the KAF group, 15.8% in the AFDAS ≤6 months; and 12.7% in the AFDAS > 6 months. Patients in the AFDAS ≤6 months had the highest independent risk of all-cause mortality even after multivariable adjustment for demographics, clinical risk factors and the use of antithrombotic therapies [HR 1.62 (1.22, 2.14)].  Compared to the stroke, no AF group, those with KAF had a higher mortality risk that was rendered non-significant after adjustment.  

Conclusions: The AFDAS group had the highest risk of death, which was not explained by comorbidities or use of antithrombotic therapies.  

Methods

### 研究目标
为根据心房颤动（atrial fibrillation, AF）确诊时机评估缺血性脑卒中患者的预后，我们构建了新发脑卒中事件的起始队列，对比三类脑卒中患者的死亡风险：a）窦性心律患者；b）确诊已知心房颤动（known atrial fibrillation, KAF）患者；c）卒中后确诊心房颤动（AF diagnosed after stroke, AFDAS）患者。

### 研究方法
本研究依托宾夕法尼亚州无房颤研究（Penn AF Free study）构建新发脑卒中患者的起始队列。通过与截至2017年6月30日的美国国家死亡索引（National Death Index）进行数据关联，识别研究对象的死亡事件。同时评估研究周期内抗凝药物与抗血小板药物的启用情况。采用Cox比例风险模型分析脑卒中亚型与死亡的关联。

### 研究结果
本研究共纳入1489例新发缺血性脑卒中患者，其中985例在研究全程未发生房颤，215例于卒中前确诊已知心房颤动（KAF），160例于卒中后≤6个月检出房颤，129例于卒中后>6个月检出房颤。中位随访时长为4.9年[四分位距（IQR）1.9, 6.8]，期间共发生686例死亡。无房颤脑卒中组年化死亡率为8.8%，KAF组为12.2%，卒中后≤6个月检出房颤（AFDAS ≤6个月）组为15.8%，卒中后>6个月检出房颤（AFDAS >6个月）组为12.7%。即便对人口统计学特征、临床危险因素及抗血栓治疗使用情况进行多变量校正后，卒中后≤6个月检出房颤组的全因死亡独立风险仍为各组最高，风险比（HR）为1.62（1.22, 2.14）。与无房颤脑卒中组相比，KAF组患者的死亡风险更高，但经校正后该差异无统计学意义。

### 研究结论
卒中后确诊房颤（AFDAS）组的死亡风险最高，且该差异无法通过合并症或抗血栓治疗的使用情况加以解释。