By Michael H. Crawford, MD, Editor
Synopsis: In a Canadian administrative database study of patients who developed atrial fibrillation during a hospitalization for other reasons, researchers followed the patients for one year to ascertain the risk of subsequent stroke. The incidence of stroke in those not anticoagulated generally was below the 2% per year threshold recommended for treatment with anticoagulants. However, in those with a CHA2DS2-VA score ≥ 5 (sex not included based on new data) and in those admitted for cardiac medical problems, the 95% confidence intervals of stroke risk did cross 2%. Thus, selected patients may be candidates for anticoagulation.
Source: Abdel-Qadir H, Gunn M, Fang J, et al. Risk for stroke after newly diagnosed atrial fibrillation during hospitalization for other primary diagnoses: A retrospective cohort study. Ann Intern Med. 2025; Apr 22. doi.org/10.7326/ANNALS-24-01967. [Online ahead of print].
Stress-induced atrial fibrillation (AF) often occurs in older patients hospitalized for infections, respiratory diseases, and surgery who have no prior history of AF. It usually is self-limiting, and there is a paucity of data to define the appropriate anticoagulant strategy for such patients. Thus, these investigators from Canada conducted a population-based cohort study of patients with a first occurrence of AF during hospitalization for other reasons.
The aim of the study was to quantitate the risk of subsequent stroke in such patients. A secondary objective was to observe anticoagulation patterns post-discharge and the risk of subsequent bleeding or mortality. These analyses were contrasted with those of a cohort of patients admitted for new onset AF.
The study population originated from the administrative health databases of Ontario, Canada, and included patients 66-105 years of age discharged alive from a hospital in 2013-2023 with a first diagnosis of AF in the hospital. Excluded were those whose discharge diagnosis was stroke, AF diagnosed in the prior five years, long-term care patients, valve disease patients, and those taking oral anticoagulants or any antiarrhythmic drug within 180 days of hospitalization.
The responsible stressor was categorized as cardiac medical or surgical or noncardiac medical or surgical. Patients were categorized by the CHA2DS2-VA score (CVS; sex excluded based on new data) as 1-4 vs. 5-8 and followed for one year. The primary outcome was stroke. Secondary outcomes were oral anticoagulant (OAC) use and hospitalization for bleeding.
Among 38,909 patients hospitalized with newly diagnosed AF, 20,639 were hospitalized for other reasons (mean age 77 years, 58% men). Of the latter, 40% were in the noncardiac medical group, 34% were in the cardiac surgical group, 17% were in the noncardiac surgical group, and 8% were in the cardiac medical group.
At one year, 26% of patients with a CVS of 1-4 were taking an OAC, as were 35% of patients with a CVS of 5-8. The primary outcome of stroke in those not taking an OAC was highest in the CVS 5-8 group (1.8%) compared to the CVS 1-4 group (0.7%). The primary outcome also varied by hospital diagnostic group: cardiac medical group (1.3%), noncardiac medical (1.2%), noncardiac surgical (1.1%), and cardiac surgical (1.0%). Among those admitted for new onset AF who were not taking an OAC, the one-year stroke rate was 2.3%. Bleeding was most common in the cardiac medical group, which had the highest rate of OAC use, and was least common in the cardiac surgical group.
The authors concluded that most patients with stress-induced AF were not put on an OAC in the year after hospital discharge. In those with a lower CVS, the incidence of stroke was below the threshold for guideline-recommended OAC (≥ 2% per year), but those with a higher CVS may exceed the OAC threshold.
Commentary
Whether to anticoagulate patients admitted to the hospital for other reasons who develop stress-related or stress-provoked AF is quite controversial. Such episodes usually are self-limited, making the decision more difficult. Especially if surgery or invasive procedures are involved, the risk of bleeding by anticoagulating comes into the risk-benefit calculation.
Since comorbidities increase the risk of subsequent AF, some use the CVS to help make the decision, since the most relevant comorbidities for AF risk are included in the score. (Note that the authors eliminated the use of sex since recent studies have shown that female sex should not influence the decision to administer anticoagulation.) The Canadian study provides some support for this practice, since those with higher scores had higher one-year stroke rates. However, the overall stroke rate in nonanticoagulated patients with a higher CVS did not exceed the 2% cutoff recommended by guidelines to initiate anticoagulation, but the confidence intervals did cross this level. Also, those admitted with a cardiac medical condition also had an observed stroke risk < 2%, but, again, the confidence intervals crossed this threshold. These observations suggest that individualized patient decisions on whether to prescribe anticoagulation in patients with stress-induced AF is a valid approach.
The Canadian study has limitations. The most important is that persistent AF may have been misdiagnosed as new AF. Also, the researchers could not determine the subsequent course of the AF and whether it was treated with rhythm control or presumed to be permanent. The type of stroke was not determined, so not all of them may have been cardioembolic. Since this was an administrative database study, there was no information on specific risk factors, such as atrial size, body weight, and ethanol consumption. Finally, only those > 65 years of age were included because there was no drug information in the database for those of younger age.
In summary, this study supports the practice of not routinely prescribing long-term anticoagulation in patients hospitalized for other reasons who develop new AF. However, in select high-risk patients (determined by CVS or comorbidities), anticoagulation may be appropriate.
Michael H. Crawford, MD, is Professor Emeritus of Medicine and Consulting Cardiologist, UCSF Health, San Francisco.
In a Canadian administrative database study of patients who developed atrial fibrillation during a hospitalization for other reasons, researchers followed the patients for one year to ascertain the risk of subsequent stroke. The incidence of stroke in those not anticoagulated generally was below the 2% per year threshold recommended for treatment with anticoagulants. However, in those with a CHA2DS2-VA score ≥ 5 (sex not included based on new data) and in those admitted for cardiac medical problems, the 95% confidence intervals of stroke risk did cross 2%. Thus, selected patients may be candidates for anticoagulation.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content