Incidence and Prognosis of Atrial Fibrillation in Patients With Sepsis

Background Although the mortality rate among patients with sepsis is declining, the incidence of both sepsis and sepsis-related deaths is increasing, likely due to its presence in a growing elderly population. As atrial fibrillation is more common in the elderly, we hypothesize that its presence will be associated with greater mortality among patients with sepsis. Methods The Medical Intensive Care Unit (MICU) database of a large tertiary care medical center was queried for sepsis-related codes and atrial fibrillation. Results Atrial fibrillation was associated with older age and a higher mortality in this series of patients with sepsis. Conclusions Whether atrial fibrillation is a marker of disease severity or contributes to mortality is uncertain. Further studies are necessary to determine optimal management.


Introduction
An emerging challenge in the management of the ICU patient with sepsis is atrial fi brillation. While the current mortality rate among patients with sepsis is declining [1], likely due to improvements in the management of the critically-ill patient including early goal-directed therapy [2], the incidence of sepsis and the number of sepsis-related deaths are increasing, due largely to a growing elderly population [3]. As atrial fi brillation is most common in older patients, we performed a retrospective review over a one-year period of ICU patients hospitalized with sepsis, and we found that atrial fi brillation was not only common, but it was associated with increased mortality.

Materials and Methods
From January 1, 2008 to December 31, 2008, 1466 patients were admitted to the Medical Intensive Care Unit (MICU) at Wake Forest University Baptist Medical Center, an 885bed university hospital. All patients admitted to the MICU were queried for sepsis-associated ICD-9 codes (785.52 and 995.92). Patients with a recent myocardial injury during the hospitalization or a malignancy identifi ed by ICD-9 codes were excluded.

Results
Four hundred sixty-fi ve of these patients were identifi ed with either severe sepsis or septic shock [4]. Of these 465 patients (203 women and 262 men) with a sepsis-associated ICD-9 code, 132 (54 women and 78 men) developed atrial fi brillation identifi ed by telemetry and confi rmed by a faculty cardiologist interpretation of an ECG.
The mean age in the ICU population during this period was 63 ± 17 years, and the mean age in the sepsis population was 62 ± 16 years. However, in the patients with both sepsis and atrial fi brillation, the mean age was 72 ± 13 years.
Of the patients with sepsis who developed atrial fi brilla- The table above includes p-values for the overall comparison of any difference between the four groups. When the overall comparison is statistically signifi cant, the odds ratios or mean comparisons are given (depending on the variable type).
For the binary variable-CAD, Odds Ratios and the 95% Confi dence Interval are given with group 4 (Neither) as the reference group. An odds ratio of 1.0 indicates no difference between the group being evaluated and the reference group. Therefore, if the 95% confi dence interval includes one, there is no difference between the two groups.
For age, the continuous variable, multiple comparisons of the mean age between the four groups was done with a Bonferroni correction. There was a signifi cant difference between all groups except 1 versus 3.  tion, there was a much higher percentage of coronary artery disease, chronic obstructive pulmonary disease and diabetes mellitus (40%, 39% and 41%, respectively) compared with those patients with sepsis who did not develop atrial fi brillation (19%, 29% and 33%, respectively). However, it appears that these were also risk factors for the development of atrial fi brillation in the general MICU population (i.e. those without sepsis) as well. (Table 1) Atrial fi brillation was strongly associated with in-hospital mortality in the MICU patient (P < 0.001). Of the patients without sepsis who developed atrial fi brillation, 94 of these 196 individuals died (48%). Even more striking is the mortality in the group with sepsis who developed atrial fi brillation where 95 of 132 (72%) patients died. (Table 2)

Discussion
Critically-ill patients frequently develop cardiac arrhythmias (up to 90% in primary cardiovascular patients); however, these groups often include those patients post major surgery, multiple trauma, those with severe underlying lung disease, malignancies, renal failure, and neurologic diseases as well as sepsis [5,6]. Only one report described 25 of 81 patients admitted with sepsis (31%) developing paroxysmal atrial fibrillation [7]. This fi nding is similar to that of ours in which 132 out of 333 patients with sepsis (40%) developed atrial fi brillation.
Cardiac involvement occurs in septic patients, even without septic shock. The ejection fraction (fraction of enddiastolic volume ejected with each beat) is often reduced, likely through several mechanisms including infl ammatory mediators [8]. It is plausible that the cardiac involvement would include the development of atrial fi brillation. However, one study did not link infl ammation (identifi ed by Creactive protein levels) with the development of post-operative atrial fi brillation [9].
Early goal-directed therapy of sepsis was designed to optimize cardiac preload, afterload, and contractility and was associated with a signifi cant mortality benefi t. However, large-volume infusion results in less of an increase in left ventricular stroke work index in patients with sepsis and septic shock as compared to other critically-ill control subjects [10]. Whether large volume infusion reduces or increases atrial arrhythmias is unknown.
The prevalence of atrial fi brillation is estimated to be 0.4 to 1% of the general population, increasing with age to > 8% in those over 80 years of age. It is complicated by an increased risk of stroke, heart failure and mortality. Anticoagulation, rate control and rhythm control strategies are treatment options; however, large studies have not demonstrated that rhythm control is superior to rate control in certain populations [11]. Some studies have focused on the prevention of atrial fi brillation, e.g., amiodarone, angioten-sin-converting enzyme inhibitors and 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) with limited success [12]. Finally, catheter ablation for paroxysmal atrial fi brillation has emerged as an effective treatment in selected patients [13].
Guidelines and strategies for the acute and chronic management of atrial fi brillation are available for many populations [14,15]. However, after an exhaustive review of the literature, Kanji et al. could not recommend a treatment strategy for atrial fi brillation in the noncardiac, critically-ill adult patient due to a lack of clinical trials) [16]. Management decisions are even more complicated among typical ICU patients who are elderly, have multiple comorbidities, and have been excluded from clinical trials of atrial fi brillation [17].
Our series, as well as others, demonstrate that atrial fibrillation frequently complicates the course of sepsis and results in increased mortality [5,7,[18][19][20], although one series reported no impact on the risk of in-hospital mortality [6]. Whether this arrhythmia is a marker of critical illness or the cause of death is uncertain. Further studies are warranted to determine the optimal management of atrial fi brillation in this older, high-risk, critically-ill population.