Hatim Attar, MD

Pennlert J, Overholser R, Asplund K, Carlberg B, Rompaye BV, Wiklund PG, et al. Optimal Timing of Anticoagulant Treatment After Intracerebral Hemorrhage in Patients With Atrial Fibrillation. Stroke. 2017

To answer this question, Pennlert et al completed a large observational study in Swedish patients. The timing for anticoagulation (AC) after Intracerebral Hemorrhage (ICH) has been brought up several times, with a recent systemic review and meta-analysis published in Stroke by Murthy et al (Restarting Anticoagulant Therapy After Intracranial Hemorrhage: A Systematic Review and Meta-Analysis). Further, the specific time point at when it is considered safe to restart anticoagulation is in flux without any current international guideline. However, this paper by Pennlert et al provides clarity specifically targeting AC in A. fib patients who have had an ICH and at what time is it optimal to start anticoagulation.

The authors used the national Swedish Stroke Register, Riksstroke, which included first-time ICH with concurrent diagnosis of atrial fibrillation between July 1, 2005 and December 31, 2012 who survived their hospitalization. Patients with a first-time ICH with a concomitant diagnosis of AF were included. Two primary outcome events were defined; the first was overall ischemic stroke events and deaths related to any vascular thrombotic event. The second outcome was recurrent ICH, as well as death attributable to other hemorrhages. Follow-up was initiated only after day 28 of the first event. Patients on dual therapy antiplatelet and anticoagulant agents were excluded. Patients were then stratified into two groups, low risk and high risk, based on patient demographics and co-morbidities via the CHA2DS2-VASc scoring system.

For statistical analysis, in order to explore the risk between time of anti-thrombotic agent onset to adverse event, this was assessed by using Cumulative Incidence Functions (CIF), the probability of observing an event before a specified time. For a given set of characteristics, the cause-specific hazards could be combined to determine the CIF for thrombotic event, hemorrhagic event, and combined event.

Study flow chart.

Figure 1: Study flow chart.

A total of 2619 patients were included, which amounted to 5759 person-years of follow-up. Total of 379 thrombotic events were noted, with 79.7% being ischemic strokes with 28-day mortality of 17.5%. Total of 115 hemorrhagic events were noted, and 83.5% were due to recurrent ICH with 28-day mortality of 37.5%. At 3 years, the cumulative incidence of thrombotic events was 14.5%, and the incidence of severe hemorrhagic events was 4.4%.

The authors have graphically represented the CIFs, for men and women separately for the primary outcomes. Anticoagulation was found to have a statistically significant risk reduction compared with no treatment in 4-16 week interval between both low- and high-risk groups. Though initiation of AC was not associated with increased risk of hemorrhage, very early initiation cannot be excluded. The positive benefit from earlier initiation is negated by risk of hemorrhage. The optimal time for initiation of AC, represented by the lowest estimated CIFs of vascular death, was 7-8 weeks. Patients’ risk factors were adjusted for, and still anticoagulation was found to be overall beneficial. With regards to antiplatelet therapy initiation, it was not associated with a lowered event risk at any point and was associated with increased overall risk.

Per Pennlert, this serves as the largest observational study to address this question and to take particular risk factors into consideration during statistical analysis. It was mentioned that there was no statistical difference between different risk factors. It should also be noted that decision to initiate either anticoagulation or antiplatelet therapy was left up to individual physician preference, which could be a confounding factor in the analysis. Another source of bias was that a more conservative management approach was taken towards high-risk patients which may explain a higher risk of thrombosis in high-risk patients treated with antiplatelet therapy.

Some of the key limitations of the study was that the type of hemorrhage was identified only via ICD-10 codes rather than reviewing imaging for each. However, hemorrhage characteristics like location, size and etiology have to be placed into the equation. Intuitively, a patient with amyloid angiopathy with multiple microbleeds on susceptibility weighted imaging would definitely carry a much higher risk for re-bleeds. Also, there was no means to assess compliance to therapy during follow-up, which may have affected outcomes. Further, this study ended in 2012; novel modalities of anticoagulation have been approved, yet the optimal timing of initiation of these agents was unable to be evaluated.

While this study has effectively outlined the picture with broad brush ‘strokes’, we need the patient details painted in to get a complete picture before making a clinical decision.