The prevalence of atrial fibrillation is estimated at 0.1% among adults younger than 55 years, rising with age to 9% in those older than 80 [1, 2], with clear implications in resource utilization by an aging population. It is estimated that the number of patients with atrial fibrillation will rise to about 5.6 million in 2050 from its current number of 2.3 million cases [3]. Atrial fibrillation is a resource-intensive arrhythmia as 70-80% of these patients are admitted during the course of their illness [4]. More importantly, a number of studies have shown that the hospitalization rates for atrial fibrillation have increased, up to 2-3 fold in the last 2 decades [5, 6].

Though there are few recommendations about the use of cardiac markers in the investigation and management of atrial fibrillation [7, 8], these tests are often ordered to diagnose myocardial infarction as the cause or consequence of the arrhythmia in patients with atrial fibrillation or flutter that present to emergency departments. Unnecessary use of cardiac markers may contribute to emergency department laboratory costs, delays in patient discharge if using serial tests, and unnecessary admission and cardiac investigations for positive troponins of unknown significance. Currently, it is unknown how many patients with atrial fibrillation/flutter undergo troponin testing and how positive troponin results are managed in the emergency department.

We sought to look at the rates of troponins ordered by emergency physicians, the rates of positive troponins, and how many of those with positive troponins were diagnosed as acute coronary syndrome by admitting services. To our knowledge, no studies have looked at the use of troponins in atrial fibrillation/flutter or the impact of positive troponins on the management of these patients.

Our study identified that a high percentage of patients with atrial fibrillation/flutter is kept in the emergency department for troponin testing. Eighty-six percent of patients had at least one set of troponins ordered, and 40% were kept for two or more sets. Our result is similar to another study, where 42.7% of patients were kept for a "rule-out myocardial infarction" protocol consisting of three sets of CK-MB cardiac markers [9].

In a recent Canadian study, 16.7% (range 10-27%) of patients with recent onset atrial fibrillation were admitted from the emergency department [10]. Admission rates were higher (55%) in our patient population; however, our study included all patients with atrial fibrillation. In our study, 14.7% of all those with at least one set of troponins and 26.6% of those with two or more sets had positive troponins. Of importance, 98% of patients with positive troponins were referred on to consulting services; however, only a third were treated as ACS by consulting services (4.9% of the entire patient cohort). It is possible that patients with positive troponins who were not treated as ACS could have required admission for other reasons; however, our study did not account for that.

The findings of our study have important implications for clinical practice and resource utilization. Cardiac troponins, while relatively cardiac-specific, are not disease-specific and can be positive in the absence of infarction [11]. For example, in a study of 1,000 consecutive patients who presented to an urban emergency department with potential symptoms or signs of coronary ischemia, 45% of patients with significantly elevated troponin I levels had a final diagnosis other than ACS [12]. In our study, 79% of patients with a positive troponin had a potential alternative diagnosis for a rise in troponin other than ACS, corroborating findings from other studies [13]. Demand-related rise in troponin could account for 35.7% of cases, as has been seen in numerous other studies [14–17]. In a study of patients presenting with elevated TnI levels with subsequent normal coronary angiogram, tachyarrythmias were the cause of TnI release in 28% of cases [18]. Failure to consider these other potential causes of elevated troponin can lead to unnecessary and invasive cardiac investigations and resource utilization, which will become of more importance with the introduction of newer, more sensitive troponin assays [19]. While it was not within the scope of this study, our work raises interesting questions about the possibility of identifying patients with specific symptoms or characteristics who warrant further diagnostic workup versus patients who are safe to discharge home. These types of questions are best answered in a prospective study, for which our work sheds light on important characteristic in the study population.

Our study has a few limitations. There was no gold standard to establish the diagnosis of acute coronary syndrome, and since the diagnosis of acute coronary syndrome given by cardiologists (and internists) can be open to interpretation, we cannot say with certainty that those who were treated for ACS truly required the treatment. Further, due to the design of our study, we did not look at adverse events in patients who were and were not treated for ACS. There are also inherent limitations of a retrospective chart review, such as the lack of clinical homogeneity among the different sites, missing clinical data, and variability in data abstraction. Specifically, the intra-rater reliability was lower for hypotension compared to other variables, as reflected in the lower kappa statistic. This is likely related to the fact that while even the data abstraction protocols stated that the 'presenting' blood pressure be measured, we did not explicitly define that this blood pressure should be the triage blood pressure, and not any other record. This could have led to confusion regarding which blood pressure was actually recorded. Unfortunately, this was not picked up during our pilot study and random checks. Despite this, we feel our results are sufficiently robust given that a kappa of 0.64 still reflects "substantial agreement," and all other kappa values were high.

In summary, a high percentage of patients with atrial fibrillation/flutter are kept in Canadian emergency departments for troponin testing, and 4.9% of the total cohort was treated for acute coronary syndrome. Further prospective studies are required to study the clinical implications and prognosis of patients with positive troponins in patients with atrial fibrillation/flutter and to identify those patients who should be tested for troponin and be treated for acute coronary syndrome.