Abstract
Context
New-onset fibrillation (AF) has been reported in 6–20% of patients with severe sepsis. Whereas chronic AF is a known risk factor for stroke and death, the clinical significance of new-onset AF in the setting of severe sepsis is uncertain.
Objective
To determine the in-hospital stroke and in-hospital mortality risks associated with new-onset AF in patients with severe sepsis.
Design
Retrospective population-based cohort of California State Inpatient Database administrative claims data from 1/1/2007 through 12/31/2007.
Setting
Non-Federal acute care hospitals.
Patients
Data was available from 3,144,787 hospitalized adults. Severe sepsis [N=49,082 (1.56%)] was defined by validated ICD-9-CM code 995.92. New-onset AF was defined as AF that occurred during the hospital stay, after excluding AF cases present at admission.
Main Outcome Measures
A priori outcome measures were in-hospital ischemic stroke (ICD-9-CM codes of 433, 434, or 436) and mortality.
Results
Patients with severe sepsis were 69±16 years old and 48% were women. New-onset atrial fibrillation occurred in 5.9% of patients with severe sepsis versus 0.6% of patients without severe sepsis [multivariable-adjusted odds ratio (OR), 6.82; 95% confidence interval (CI), 6.54–7.11; P<0.001]. Severe sepsis was present in 14% of all new-onset AF in hospitalized adults. Compared with severe sepsis patients without new-onset AF, patients with new-onset AF during severe sepsis had greater risks of in-hospital stroke (75/2896 (2.6%) vs. 306/46186 (0.6%) strokes, adjusted OR 2.70; 95% CI, 2.05–3.57; P <0.0001) and in-hospital mortality (1629 (56%) vs. 18027 (39%) deaths, adjusted relative risk, 1.07; 95% CI, 1.04–1.11; P <0.0001). Findings were robust across two definitions of severe sepsis, multiple methods of addressing confounding, and multiple sensitivity analyses.
Conclusion
Among patients with severe sepsis, patients with new-onset AF were at increased risk for in-hospital stroke and death compared with patients with no AF and patients with pre-existing AF.
Keywords: sepsis, atrial fibrillation, strokes, acute, assessment, outcomes, epidemiology, mortality
INTRODUCTION
Atrial fibrillation (AF) is the most common arrhythmia to affect the critically ill.1,2 Previous studies have demonstrated that 6–20% of patients with severe sepsis develop new-onset AF, suggesting that severe sepsis may be a predisposing factor for new-onset AF.2–5 Whereas chronic AF is a major contributor to long-term disability and mortality,6 the relation of new-onset AF in severe sepsis to prognosis is unknown.
Although new-onset AF appears to be relatively common in individuals with severe sepsis, data are sparse to risk stratify or prognosticate these often complicated and critically ill patients. Small, single center studies suggest that new-onset AF is associated with higher mortality and prolonged hospitalization during severe sepsis.5,7,8 However, a population-based assessment of the adverse outcomes associated with new-onset AF that occurs in the context of severe sepsis has not yet been undertaken.
Using a database of individuals hospitalized with severe sepsis identified via California State Inpatient claims, we sought to determine the association of new-onset AF with adverse outcomes of in-hospital mortality and in-hospital ischemic stroke. We hypothesized that new-onset AF during severe sepsis would be associated with increased risks for in-hospital stroke and in-hospital mortality.
METHODS
Participants
We examined hospitalizations from adults (age ≥18 years) using year 2007 discharge data from the California State Inpatient Database, Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality (AHRQ).9 The California AHRQ Inpatient Database contains data for all patients hospitalized in non-Federal acute care California hospitals during 2007. Data elements included demographics, admission and discharge status, length of stay, up to 25 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis, and up to 21 ICD-9-CM procedure codes. Each diagnosis was coded with a separate identifier denoting whether or not it was present on admission, allowing discrimination between pre-existing diagnoses and complications occurring after hospital admission.10–12 All study procedures were approved by the Boston University Medical Campus Institutional Review Board.
The primary severe sepsis analytic cohort was defined by the presence of the ICD-9-CM code for severe sepsis (ICD-9-CM 995.92).
Outcomes
AF was defined by the presence of ICD-9-CM 427.3×.13 New-onset AF was defined as atrial fibrillation or atrial flutter that occurred during the hospital stay, excluding cases that were present on admission. In-hospital ischemic stroke was defined with ICD-9-CM codes of 433, 434, or 43614 that were not present on admission.
Covariates
Demographic data collected in the California State Inpatient Database included age, sex, race and ethnicity (coded as white, black, Hispanic or other). Previously described AF risk factors15–18 were defined by ICD-9-CM codes19,20 present on admission for heart failure, myocardial infarction, hypertension, obesity, diabetes mellitus, and chronic obstructive pulmonary disease (COPD). Acute factors were obtained from ICD-9-CM codes for acute organ dysfunction diagnoses,21,22 electrolyte abnormalities, right heart catheterization, infectious pathogen type (gram positive vs. gram negative bacteria vs. fungal vs. none reported) and infection source. ICD-9-CM coding strategies are shown in eTable 1.
Sensitivity analyses
We performed sensitivity and exploratory analyses to explore the temporality of severe sepsis, AF, and ischemic stroke events. We analyzed a cohort with severe sepsis (ICD-9-CM 995.92) coded as “present on admission”, in whom new-onset AF that was not “present on admission” occurred after the onset of severe sepsis. We performed three sensitivity analyses using increasingly narrow ICD-9-CM codes for embolic stroke, the putative mechanism for stroke associated with AF6 ICD-9-CM 434 or 436:“cerebral artery occlusion” or “acute, but ill-defined, cerebrovascular disease”; ICD-9-CM 434: “cerebral artery occlusion”; and ICD-9-CM 434.1: “embolic stroke”. Because temporality between new-onset AF and new-onset stroke claims could not be established within the severe sepsis hospitalization, we performed an exploratory analysis investigating the risk of re-hospitalization with a new ischemic stroke following the severe sepsis hospitalization (see eSupplement for detailed Methods).
We repeated mortality and stroke outcome analyses in an alternative severe sepsis cohort defined by the presence of ICD-9-CM codes for both infection and for acute organ dysfunction, as previously described by Angus, et al (eTable 1).22 In addition, we investigated stroke and mortality risks adjusting for potential re-admissions using the utilizing the AHRQ “revisit” database to classify individual patients as a random effects. Because septic embolic disease may result in stroke, we performed a sensitivity analysis excluding all patients with septic emboli claims (ICD-9-CM 449). Finally, indicators of intensive care unit admission (ICU) are not available in California State Inpatient data. As a proxy for intensive care, we performed a subgroup analysis in patients with claims for mechanical ventilation.
ICD-9-CM Validation
Validation of the severe sepsis ICD-9-CM 995.92 at our institution (see eSupplement) demonstrated moderate sensitivity [52%, (95% Confidence Interval (CI), 39–65%)] and high specificity [98%, (95% CI, 92–100%)], similar to validation findings in other hospitals.23 Previous studies have demonstrated 95% sensitivity and 99% specificity for AF ICD-9-CM 427.3× claims.13 We validated present on admission modifiers (see eSupplement) for severe sepsis and AF claims; agreement between severe sepsis present on admission status and blinded chart review was 91% (kappa 0.77) and an agreement between AF present on admission status and blinded chart review was 90% (kappa 0.74), similar to previous findings.24 Prior validation of ischemic stroke ICD-9-CM codes has shown variable accuracy;14,25–27 however, a strategy using ICD-9-CM 433, 434, 436 at any diagnostic position previously demonstrated 86% sensitivity and 95% specificity (kappa=0.82).14 ICD-9-CM 434.11 used to indicate embolic stroke has been demonstrated to be accurate in 73% of patients; patients coded with 434.11 who did not have a clear embolic stroke on chart review were characterized as having either ischemic stroke of athero-thrombotic or uncertain etiology.26
Statistical Analyses
Severe sepsis, new-onset AF and in-hospital stroke
Multivariable logistic regression models were used to determine the association between the presence of severe sepsis and new-onset AF or in-hospital stroke, adjusting for demographics (age, sex, race/ethnicity), and claims for pre-existing comorbidities (heart failure, stroke, myocardial infarction, diabetes, obesity, hypertension, and COPD).
Factors associated with new-onset AF during severe sepsis
We performed multivariable logistic regression models with forward stepwise selection to determine the factors associated with new-onset AF during severe sepsis. Because patients with pre-existing AF are not at risk for new-onset AF, we excluded patients with pre-existing AF from this model.
New-onset AF during severe sepsis and adverse outcomes
We constructed four regression models to evaluate the association of new-onset AF during severe sepsis with in-hospital stroke and in-hospital mortality. Model 1 included demographics and comorbidities as model covariates (claims for prior history of heart failure, myocardial infarction, stroke, diabetes, obesity, hypertension, metastatic or hematologic malignancy, or COPD); Model 2 included demographics and acute factors (number and type of organ dysfunction claims, electrolyte abnormality, right heart catheterization, sources of infection, and pathogen type); Model 3 was a combination of all covariates in Model 1 and Model 2.
Regression Model 4 was performed on a separate cohort matched on the probability of new-onset AF. In order to construct the matched cohort, we calculated the probability of new-onset AF via multivariable logistic regression including measured covariates as independent variables. Nearest-neighbor matching to four decimal places28 was used to match participants with new-onset AF to those without new-onset AF based on the calculated probability of new-onset AF.
We used generalized estimating equations to calculate odds ratios for stroke in all Models. Individual hospitals were defined as random effects due to potentially correlated physician coding practices and patient characteristics within a single hospital. Differences in per-patient stroke rates according to AF status were analyzed with Wilcoxon-Mann-Whitney tests. We used Poisson regression with robust estimates29,30 to calculate relative risks for mortality associated with AF status.
A two-sided alpha level of 0.05 was selected for statistical significance for analyses of outcomes. SAS version 9.0 (Cary, NC) was used for all analyses.
RESULTS
The 2007 California State Inpatient Database contained claims data from 3,144,787 hospitalized adults who had a mean age of 55±21 years, and of whom 62% were women, with a racial/ethnic composition of 57% white, 26% Hispanic, 8% black, and 9% other races. Figure 1 demonstrates the number of patients with or without severe sepsis and outcomes associated with each severe sepsis classification category.
Figure 1.
demonstrates the number of patients, along with in-hospital stroke and mortality outcomes, in each analytic cohort.
Severe sepsis was present during 49,082 [1.56%, 95% confidence interval (CI) 1.55–1.57] hospitalizations. Patients with severe sepsis were on average 69±16 years old and 48% were women, with a racial/ethnic composition of 56% white, 20% Hispanic, 9% black, and 15% other/missing races. Table 1 displays the characteristics of the severe sepsis cohort stratified by AF status. eTable 2 demonstrates that imbalances in measured covariates between AF groups were substantially attenuated in the propensity-score matched severe sepsis cohort.
Table 1.
Characteristics of participants with severe sepsis with and without new-onset atrial fibrillation (AF).
| Variable | No AF N=36200 |
Pre-existing AF N=9986 |
New-onset AF N=2896 |
|---|---|---|---|
| Age, mean (SD), y | 66 (17) | 76 (12) | 74 (12) |
| Sex, female | 17690 (49.0) | 4756 (47.7) | 1280 (44.2) |
| Race/ethnicity | |||
| White | 19006 (52.5) | 6485 (64.9) | 1861 (64.3) |
| Black | 3547 (10.1) | 671 (6.7) | 209 (7.2) |
| Hispanic | 8045 (22.2) | 1397(14.0) | 384 (13.3) |
| Other | 5502 (15.2) | 1433 (14.4) | 442 (15.3) |
| Comorbidities | |||
| Hypertension | 17373 (48.0) | 5751(57.6) | 1336 (46.1) |
| Diabetes mellitus | 12135(33.5) | 3546 (35.5) | 800 (27.6) |
| Obesity | 2914 (8.1) | 711(7.1) | 208 (7.2) |
| Malignancy | 4079 (11.3) | 905 (9.1) | 348(12.0) |
| Congestive heart failure | 2312 (6.4) | 1365 (13.7) | 306(10.6) |
| COPD | 1920 (5.3) | 745 (7.5) | 189 (6.5) |
| Myocardial infarction | 1659 (4.6) | 713 (7.1) | 165 (5.7) |
| Stroke | 880 (2.4) | 317 (3.2) | 126 (4.4) |
| Acute organ dysfunction | |||
| Total organ failures, mean (SD) | 2.44 (1.35) | 2.51 (1.30) | 3.10 (1.28) |
| Circulatory | 23264 (64.3) | 6690 (67.0) | 2049 (70.7) |
| Renal | 21343 (59.0) | 6338 (63.5) | 2048 (70.7) |
| Respiratory | 20833 (57.6) | 6080 (60.1) | 2385 (82.4) |
| Hematologic | 7301 (20.2) | 1974 (19.8) | 933 (32.2) |
| Neurologic | 5318 (14.7) | 1414 (14.2) | 526 (18.2) |
| Metabolic | 7246 (20.0) | 1933(19.4) | 706(24.4) |
| Hepatic | 3150 (8.7) | 673 (6.7) | 334(11.5) |
| Electrolyte abnormality | 23963 (66.2) | 6733 (67.4) | 2051 (70.8) |
| Right heart catheterization | 722 (2.0) | 236 (2.4) | 172 (5.9) |
| Source of infection | |||
| Respiratory | 14178 (39.2) | 4358 (43.6) | 1419 (49.0) |
| Urinary tract | 13748 (38.1) | 4025 (40.3) | 950 (32.8) |
| Primary bacteremia | 8910 (24.6) | 2240 (22.4) | 600 (20.7) |
| Abdominal | 5532 (15.3) | 1496 (15.0) | 735 (25.4) |
| Skin or soft tissue | 2888 (8.0) | 729 (7.3) | 229 (7.9) |
| Pathogen type | |||
| Gram positive | 8675 (24.0) | 2546 (25.5) | 822 (28.4) |
| Gram negative | 8776 (24.2) | 2342(23.5) | 677 (23.4) |
| Fungal | 632(1.8) | 156(1.6) | 104(3.6) |
| None specified | 20268 (56.0) | 5530 (55.4) | 1538 (53.1) |
N (%), unless otherwise specified; SD denotes standard deviation; COPD: Chronic obstructive pulmonary disease
Severe sepsis and new-onset AF
New-onset AF occurred during 20,608 (0.65%, 95% CI 0.65–0.66) of hospitalizations (including sepsis and non-sepsis) and during 2896 (5.9%, 95% CI 5.7–6.1%) hospitalizations of patients with severe sepsis. Therefore, 14% (95% CI 13.6–14.5%) of all hospital-associated, new-onset AF occurred in the context of severe sepsis. Compared with hospitalized patients without severe sepsis, patients with severe sepsis had an increased risk of new-onset AF [demographic and comorbidity-adjusted odds ratio (OR), 6.82; 95% confidence interval (CI), 6.54–7.11; P <0.001].
Factors associated with new-onset AF during severe sepsis
Results of the multivariable analysis of factors associated with new-onset AF during severe sepsis are shown in Table 2. Factors associated with increased risk for new-onset AF during severe sepsis included demographics (increasing age, male sex, and white race), comorbidities (history of heart failure, obesity, malignancy, and stroke), and acute factors (increasing number of organ failures, respiratory failure, hematologic failure, renal failure, use of right heart catheter, pulmonary or abdominal source of infection, and gram positive or fungal organisms).
Table 2.
Factors associated with new-onset atrial fibrillation among patients with severe sepsis.
| Variable | OR (95% Confidence interval) |
|---|---|
| Age, per 10 years | 1.52 (1.47–1.56) |
| Female Sex | 0.83 (0.76–0.90) |
| Race/ethnicity | |
| White | Ref |
| Black | 0.67 (0.58–0.78) |
| Hispanic | 0.58 (0.50–0.63) |
| Other | 0.78 (0.69–0.87) |
| Comorbidities | |
| Hypertension | 0.88 (0.81–0.95) |
| Diabetes mellitus | 0.82 (0.75–0.90) |
| Obesity | 1.20 (1.03–1.40) |
| Congestive heart failure | 1.61 (1.41–1.83) |
| Metastatic or hematologic malignancy | 1.23 (1.09–1.39) |
| Prior Stroke | 1.64 (1.35–2.01) |
| Acute Organ dysfunction | |
| Per organ failure | 1.12 (1.05–1.19) |
| Respiratory failure | 2.81 (2.48–3.19) |
| Renal failure | 1.40 (1.26–1.56) |
| Hematologic failure | 1.50 (1.34–1.68) |
| Acidosis | 0.87 (0.77–0.97) |
| Right heart catheterization | 2.25 (1.87 –2.70) |
| Source of infection | |
| Respiratory | 1.27 (1.14–1.40) |
| Urinary tract | 0.89 (0.81–0.99) |
| Abdominal | 1.77 (1.59–1.97) |
| Primary bacteremia | 1.17 (1.02–1.36) |
| Skin or soft tissue | 1.33 (1.14–1.55) |
| Pathogen type | |
| Gram positive | 1.29 (1.18–1.55) |
| Fungal | 1.59 (1.27–2.00) |
c-statistic 0.760
In-hospital ischemic stroke
In-hospital stroke claims occurred during 3310 (0.11%, 95% CI 0.10–0.11) adult hospitalizations (both sepsis and non-sepsis) and 381 (0.78%, 95% CI 0.70–0.86%) hospitalizations of patients with severe sepsis. Thus, 11% (95% CI 10–13%) of in-hospital strokes occurred in patients with severe sepsis. Compared with hospitalized patients without severe sepsis, patients with severe sepsis had an increased risk for in-hospital ischemic stroke, (demographic and comorbidity-adjusted OR 6.0; 95% confidence interval (CI), 5.38–6.69; P <0.001).
In patients with severe sepsis, in-hospital ischemic stroke occurred in 75 of 2896 (2.6%, 95% CI 2.0–3.2%) individuals with new-onset AF compared to 57 of 9986 (0.57%, 95% CI 0.43–0.74%) with pre-existing AF and 249 of 36200 (0.69%, 95% CI 0.61–0.78%) without AF. The average stroke rate was 0.15±1.21% per hospital-day for patients with new-onset AF as compared with 0.05±1.01% for patients with no AF or pre-existing AF (p<0.001).
Amongst individuals with severe sepsis, new-onset AF was associated with increased adjusted risks for in-hospital ischemic stroke (Table 3). In contrast, participants with severe sepsis and pre-existing AF did not have an increased risk of in-hospital ischemic stroke compared to those with severe sepsis and no AF (multivariable-adjusted OR, 0.74; 95% CI, 0.55–1.01; P =0.054). Further, individuals with new-onset AF had greater stroke risk than those with pre-existing AF (multivariable-adjusted OR, 3.63; 95% CI 2.51–5.25, P <0.0001). A sensitivity analysis using different ischemic stroke ICD-9-CM definitions demonstrated increasing strength of association between new-onset AF and in-hospital stroke in severe sepsis as ICD-9-CM codes increased in specificity for “embolic stroke” (Table 4). Additional sensitivity analyses in patients meeting the alternative severe sepsis definition, in patients with severe sepsis present on admission, in patients without septic emboli claims, and in patients who required mechanical ventilation did not result in substantially different estimates of stroke risk associated with new-onset AF (Table 4).
Table 3.
Association between new-onset atrial fibrillation and adverse outcomes.
| Model | Outcome | |
|---|---|---|
| In-hospital ischemic stroke |
In-hospital Mortality | |
| Odds ratio (95% CI) |
Relative Risk (95% CI) |
|
| Age, sex and race-adjusted | 4.05 (3.09–5.30) | 1.36 (1.32–1.41) |
| Demographic and comorbidity-adjusteda | 3.84 (2.93–5.04) | 1.33 (1.28 –1.37) |
| Demographic and acute factor-adjustedb | 2.84 (2.15–3.76) | 1.09 (1.06–1.13) |
| Combined multivariable-adjustedc | 2.70 (2.05–3.57) | 1.07 (1.04–1.11) |
| New-onset AF, probability-matchedd | 2.75 (1.76–4.29) | 1.13 (1.08–1.19) |
Abbreviations: AF: atrial fibrillation CI: Confidence Interval
Adjusted for age, sex, race, and prior history of comorbidities diabetes mellitus, hypertension, obesity, heart failure, stroke, myocardial infarction, chronic obstructive pulmonary disease, metastatic or hematologic malignancy.
Adjusted for age, sex, race, and sepsis-associated factors including the number of organ failures, presence of electrolyte disturbances, source of sepsis, type of organ failure, type of pathogenic organism, use of right heart catheterization.
Adjusted for all variables from both the comorbidity-adjusted and sepsis-associated factor-adjusted models
Cohorts matched on calculated probability for risk of new-onset AF.
Table 4.
Sensitivity analyses of association between new-onset atrial fibrillation and in-hospital stroke in patients with severe sepsis.a
| Group | Severe sepsis: (ICD-9-CM 995.92) |
Severe sepsis: (Infection + Organ failure)22 |
|---|---|---|
| Odds ratio (95% confidence interval) N in model and # of events |
||
| Severe sepsisb | 2.70 (2.05–3.57) N=48961 381 strokes |
2.85 (2.41–3.37) N=228677 1172 strokes |
| Severe sepsisc | 3.04 (2.26–4.10) N=48961 327 strokes |
3.44 (2.86–4.13) N=228677 814 strokes |
| Severe sepsisd | 3.10 (2.30–4.17) N=48961 324 strokes |
3.46 (2.86–4.15) N=228677 809 strokes |
| Severe sepsise | 3.94 (2.21–7.01) N=48961 70 strokes |
4.81 (3.38–6.87) N=228677 182 strokes |
| Severe sepsis, exclude septic embolib | 2.70, (2.06–3.54) N=47290, 376 strokes |
2.84 (2.40–3.37) N=228649 1169 strokes |
| Severe sepsis, mechanical ventilationb | 2.47 (1.79–4.42) N=23335 284 strokes |
2.50 (2.05–3.03) N=52595 668 strokes |
| Severe sepsis, patient random effectb (account for multiple sepsis admissions) | 2.86 (2.16–3.78) N=45037 353 strokes |
- |
| Severe sepsis, present on admissionb | 2.17 (1.37–3.42) N=38591 196 strokes |
- |
Abbreviations: ICD-9-CM: International Classification of Diseases, Ninth Revision, Clinical Modification, - analysis not performed
Models adjusted for age, sex, race, and prior history of diabetes mellitus, hypertension, obesity, heart failure, myocardial infarction, stroke, chronic obstructive pulmonary disease, metastatic or hematologic malignancy and sepsis-associated factors including the number of organ failures, presence of electrolyte disturbances, source of sepsis, type of organ failure, type of pathogenic organism, use of right heart catheterization.
In-hospital ischemic stroke defined by ICD-9-CM 433, 434, or 436, (“occlusion and stenosis precerebral arteries, (“cerebral artery occlusion” or “acute, ill-defined, cerebrovascular disease”), not present on admission
In-hospital ischemic stroke defined by ICD-9-CM 434 or 436 (“cerebral artery occlusion” or “acute, ill-defined, cerebrovascular disease”), not present on admission
In-hospital ischemic stroke defined by ICD-9-CM 434 (“cerebral artery occlusion”), not present on admission
In-hospital ischemic stroke definied by ICD-9-CM 434.11 (“embolic stroke”), not present on admission
In order to establish a temporal relationship between new-onset AF during severe sepsis and new stroke, we performed an exploratory analysis investigating the risk for incident stroke occurring after the severe sepsis hospitalization. We identified 27325 severe sepsis survivors without a prior stroke claim. Re-hospitalization with a new ischemic stroke occurred in 23/1171 (2.0%) with new-onset AF during severe sepsis, 81/5300 (1.5%) with pre-existing AF during sepsis, and 261/20854 (1.3%) without AF during severe sepsis. Compared to patients with no AF, patients with new-onset AF during the severe sepsis hospitalization had a non-significantly increased risk for re-hospitalization with incident ischemic stroke (multivariable-adjusted hazard ratio: 1.51, 95% CI 0.98–2.33, p=0.06).
In-hospital mortality
In patients with severe sepsis, 1629 (56.3%, 95% CI 54.4–58.1%) with new-onset AF died in the hospital, compared with 4375 (43.8%, 95% CI 42.8–44.8%) with pre-existing AF and 13652 (37.7%, 95% CI 37.2–38.2%) without AF. Thus, compared with patients without new-onset AF, individuals with new-onset AF during severe sepsis experienced increased in-hospital mortality (Table 3). Sensitivity analyses did not show substantially different effect estimates for mortality (eTable 4). eFigure 2 demonstrates that the increased mortality risk for subjects with severe sepsis and new-onset AF persisted regardless of the number of severe sepsis-associated organ failures.
DISCUSSION
Our investigation of new-onset AF during severe sepsis presents a number of clinically relevant findings. First, severe sepsis was associated with increased risk of both new-onset AF and in-hospital ischemic stroke. In addition, we identified multiple demographic and clinical factors associated with new-onset AF during severe sepsis. Importantly, patients with new-onset AF during severe sepsis had increased risks for both in-hospital ischemic stroke and mortality. The increased stroke and mortality risks observed with new-onset AF were robust across two definitions of severe sepsis, multiple methods of addressing confounding, and multiple sensitivity analyses.
Consistent with previous reports of increased stroke risk following infection,31–35 our study demonstrates that patients with severe sepsis had a six-fold increased risk for in-hospital stroke compared with hospitalized patients without severe sepsis. Importantly, patients with severe sepsis who developed new-onset AF had a greater risk of in-hospital stroke than patients with pre-existing AF and individuals without a history of AF. As far as we are aware, the increased risk of ischemic stroke in patients with severe sepsis and new-onset AF has not been previously reported.
Several potential mechanisms might explain the increased ischemic stroke risk in patients with severe sepsis and new-onset AF. Severe sepsis alone may be associated with an increased risk of stroke through hemodynamic collapse, increased systemic inflammation, and coagulopathy.36,37 New-onset AF may simply be a marker for greater severity of illness, and thus, greater stroke risk. However, within the limitations of claims data, adjustment for clinical and demographic factors associated with severity of illness did not eliminate the strong associations of new-onset AF claims with incident stroke claims.
Alternatively, new-onset AF may be a potential source of cardio-embolic stroke. In fact, we identified an almost four-fold increased risk for in-hospital embolic stroke claims associated with new-onset AF during the severe sepsis hospitalization. Although new-onset AF in the critically ill is often transient,2 prior studies have shown that atrial thrombi may form within 2 days of the onset of AF.38 Of note, individuals with severe sepsis and pre-existing AF did not have an increased stroke risk when compared with patients without AF. Potential differences in anticoagulation practice patterns between patients with pre-existing and new-onset AF could not be ascertained from the California administrative data. Whether anticoagulation practices differ between patients with pre-existing AF and new-onset AF and whether benefits of systemic anticoagulation for AF during severe sepsis outweigh risks cannot be ascertained from our data source and warrants further investigation.
New-onset AF was associated with a 7% increase in the adjusted risk of in-hospital death. The hospital mortality of patients with new-onset AF during severe sepsis was similar to the mortality of those without AF, but with one additional acute organ failure. Whether new-onset AF functions as a marker for increased illness severity and poor prognosis (e.g., new-onset AF represents an additional “organ dysfunction”) or directly contributes to mortality (e.g., through refractory hypotension, stroke, or heart failure) cannot be elucidated from observational claims data and also warrants further investigation.
Our results expand upon those of prior studies2–4 demonstrating severe sepsis to be strongly associated with new-onset AF. Severe sepsis was associated with 14% of all episodes of new-onset AF occurring in hospitalized adults. Standard demographic risk factors for community-acquired AF such as older age, male sex and white race15,16 and previously-described AF-associated comorbidities such as heart failure, obesity15,16 and malignancy,39–41 were also associated with new-onset AF during severe sepsis. In addition, prevalent stroke was associated with new-onset AF, a finding that suggests some patients with new-onset AF may have had undiagnosed paroxysmal AF.42 Our seemingly paradoxical findings that hypertension and diabetes were associated with reduced risk for new-onset AF in severe sepsis may be an artifact of less frequent coding of chronic comorbid conditions such as hypertension and diabetes in patients with critical illness and warrant prospective study. Multiple acute factors were also associated with increased risk of first-diagnosed AF during severe sepsis. Whether these acute factors serve as markers of illness severity or represent other potential mechanisms for AF is unclear, though the increased risk of arrhythmias associated with use of right heart catheters in the critically ill has been previously reported.43,44
Our study has several limitations. First, the incidence of severe sepsis-associated, new-onset AF in our study (5.9%) was on the lower end of previously reported rates (6–20%).2–5 AF incidence in our study may be lower because claims data are less sensitive for detecting new-onset AF as compared with the chart abstraction or prospective identification used in prior studies. Additionally, new-onset AF that occurred upon hospital admission in the setting of severe sepsis may be classified as “pre-existing AF” by the “present on admission” coding strategy used in our study, which might falsely lower the observed incidence of AF or bias outcome analyses. Second, because new-onset AF during critical illness is often self-limited,2 clinicians may not record episodes of AF that are deemed clinically insignificant. Thus, “circular coding” may be present in which new-onset AF is preferentially coded when it is felt to be associated with an adverse outcome such as stroke. However, we did not find associations between stroke and pre-existing AF, which would theoretically be susceptible to a similar circular coding bias. Third, because our definition of new-onset AF necessitated that AF occur after hospital admission, an “immortal time” bias45 may exist in which patients had to survive long enough to be diagnosed with AF. Immortal time bias may have falsely lowered the risk of mortality associated with new-onset AF; thus the mortality risk associated with new-onset AF during severe sepsis may actually be greater than reported in our study. Death may be considered as a competing risk for in-hospital stroke; however, we could not ascertain time to in-hospital stroke necessary to perform competing risk analyses. The composite vector of these potential biases is not clear. Fourth, though sensitivity analyses suggest a temporal order between severe sepsis, AF and in-hospital stroke, the administrative data used for the present study is limited in ascertaining the timing of clinical events. However, an exploratory analysis of incident ischemic stroke occurring after the severe sepsis hospitalization suggests that stroke risks may remain elevated up to 1-year following severe sepsis. Finally, due to the observational nature of our study and the possibility for unmeasured confounding we cannot prove a causal relation between new-onset AF in the setting of severe sepsis and increased risk of stroke and death.
Clinical Implications and Future Directions
Given projected estimates of severe sepsis incidence in 1,000,000 Americans in 2011,22 it is likely that new-onset AF occurs in greater than 60,000 patients with severe sepsis in the US each year. Our findings suggest that new-onset AF during severe sepsis is associated with especially high short-term nosocomial stroke and mortality risks; most patients with new-onset AF during a hospitalization with severe sepsis did not survive. Current Societal guidelines do not address AF that occurs in the setting of severe sepsis or acute infection,6,46 suggesting that new-onset AF that occurs during severe sepsis is an under-recognized public health problem. If our findings of increased stroke and death in the setting of AF and severe sepsis are replicated in other data sets, then it will be important to examine management strategies that might diminish the risk of adverse outcomes associated with AF during severe sepsis.
Supplementary Material
ACKNOWLEDGEMENTS
Funding: UL1RR025771 Boston University Clinical Translational Research Institute (AJW), the Evans Center for Interdisciplinary Biomedical Research ARC on Atrial Fibrillation at Boston University (http://www.bumc.bu.edu/evanscenteribr/) (AJW, LHC, EJB); 1R01HL092577 (EJB); 1RC1HL101056 (EJB); 1R01 HL102214 (EJB, LHC); K07 CA138772 (RSW) and the Department of Veterans Affairs (RSW).
No funding organization had a role in the design or conduct of the study.
ABBREVIATIONS
- AF
Atrial fibrillation
- AHRQ
Agency for Healthcare Research and Quality
- CI
Confidence interval
- ICD-9-CM
International Classification of Diseases, Ninth Revision, Clinical Modification
- OR
odds ratio
- RR
relative risk
Footnotes
Dr. Walkey had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Walkey, Wiener, Benjamin.
Acquisition of data: Walkey, Wiener, Ghobrial.
Analysis and interpretation of data: Walkey, Wiener, Benjamin, Curtis.
Drafting of the manuscript: Walkey, Wiener.
Critical revision of the manuscript for important intellectual content: Walkey, Wiener, Benjamin, Curtis.
Statistical analysis: Walkey, Wiener.
No author has conflicts of interest to declare.
Contributor Information
Allan J. Walkey, The Pulmonary Center, Division of Pulmonary and Critical Care Medicine, Department of Medicine, Boston University School of Medicine, Boston, MA, USA..
Renda Soylemez Wiener, The Pulmonary Center, Division of Pulmonary and Critical Care Medicine, Boston University School of Medicine, Boston, MA USA. Center for Health Quality, Outcomes, & Economic Research, Edith Nourse Rogers Memorial VA Hospital, Bedford, MA, USA. The Dartmouth Institute for Health Policy and Clinical Practice, Dartmouth Medical School, Hanover, NH, USA.
Joanna M. Ghobrial, Department of Medicine, Division of Cardiology, University of Washington School of Medicine, Seattle, WA USA..
Lesley H. Curtis, Duke Clinical Research Institute, Duke University School of Medicine, Durham, NC, USA. Department of Medicine, Duke University School of Medicine, Durham, NC, USA..
Emelia J. Benjamin, National Heart Lung and Blood Institute’s and Boston University’s Framingham Heart Study, Framingham, MA, USA. Cardiology and Preventive Medicine, Whitaker Cardiovascular Institute, Department of Medicine, Boston University School of Medicine, Boston, MA, USA. Epidemiology Department, Boston University School of Public Health, Boston, MA, USA..
References
- 1.Artucio H, Pereira M. Cardiac arrhythmias in critically ill patients: Epidemiologic study. Crit Care Med. 1990;18(12):1383–1388. doi: 10.1097/00003246-199012000-00015. [DOI] [PubMed] [Google Scholar]
- 2.Annane D, Sebille V, Duboc D, et al. Incidence and prognosis of sustained arrhythmias in critically ill patients. Am J Respir Crit Care Med. 2008;178(1):20–25. doi: 10.1164/rccm.200701-031OC. [DOI] [PubMed] [Google Scholar]
- 3.Goodman S, Shirov T, Weissman C. Supraventricular arrhythmias in intensive care unit patients: Short and long-term consequences. Anesth Analg. 2007;104(4):880–886. doi: 10.1213/01.ane.0000255759.41131.05. [DOI] [PubMed] [Google Scholar]
- 4.Arora S, Lang I, Nayyar V, Stachowski E, Ross DL. Atrial fibrillation in a tertiary care multidisciplinary intensive care unit--incidence and risk factors. Anaesth Intensive Care. 2007;35(5):707–713. doi: 10.1177/0310057X0703500508. [DOI] [PubMed] [Google Scholar]
- 5.Meierhenrich R, Steinhilber E, Eggermann C, et al. Incidence and prognostic impact of new-onset atrial fibrillation in patients with septic shock: A prospective observational study. Crit Care. 2010;14(3):R108. doi: 10.1186/cc9057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fuster V, Ryden LE, Cannom DS, et al. 2011 ACCF/AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: A report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation. 2011;123(10):e269–e367. doi: 10.1161/CIR.0b013e318214876d. [DOI] [PubMed] [Google Scholar]
- 7.Christian SA, Schorr C, Ferchau L, Jarbrink ME, Parrillo JE, Gerber DR. Clinical characteristics and outcomes of septic patients with new-onset atrial fibrillation. J Crit Care. 2008;23(4):532–536. doi: 10.1016/j.jcrc.2007.09.005. [DOI] [PubMed] [Google Scholar]
- 8.Salman S, Bajwa A, Gajic O, Afessa B. Paroxysmal atrial fibrillation in critically ill patients with sepsis. J Intensive Care Med. 2008;23(3):178–183. doi: 10.1177/0885066608315838. [DOI] [PubMed] [Google Scholar]
- 9.HCUP state inpatient database (SID), California. Healthcare Cost and Utilization Project (HCUP) Rockville, MD: Agency for Healthcare Research and Quality; 2007. www.hcup-us.ahrq.gov/sidoverview.jsp. [Google Scholar]
- 10.Glance LG, Dick AW, Osler TM, Mukamel DB. Accuracy of hospital report cards based on administrative data. Health Serv Res. 2006;41(4 Pt 1):1413–1437. doi: 10.1111/j.1475-6773.2006.00554.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Glance LG, Dick AW, Osler TM, Mukamel DB. Does date stamping ICD-9-CM codes increase the value of clinical information in administrative data? Health Serv Res. 2006;41(1):231–251. doi: 10.1111/j.1475-6773.2005.00419.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Coffey R, Milenkovic M, Andrews RM. The case for the present-on-admission (POA) indicator. [Accessed 7/7/2011];HCUP Methods Series Report # 2006-01. 2006 [Google Scholar]
- 13.Glazer NL, Dublin S, Smith NL, et al. Newly detected atrial fibrillation and compliance with antithrombotic guidelines. Arch Intern Med. 2007;167(3):246–252. doi: 10.1001/archinte.167.3.246. [DOI] [PubMed] [Google Scholar]
- 14.Tirschwell DL, Longstreth WT., Jr Validating administrative data in stroke research. Stroke. 2002;33(10):2465–2470. doi: 10.1161/01.str.0000032240.28636.bd. [DOI] [PubMed] [Google Scholar]
- 15.Schnabel RB, Sullivan LM, Levy D, et al. Development of a risk score for atrial fibrillation (Framingham Heart Study): A community-based cohort study. Lancet. 2009;373(9665):739–745. doi: 10.1016/S0140-6736(09)60443-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chamberlain AM, Agarwal SK, Folsom AR, et al. A clinical risk score for atrial fibrillation in a biracial prospective cohort (from the Atherosclerosis Risk in Communities [ARIC] study) Am J Cardiol. 2011;107(1):85–91. doi: 10.1016/j.amjcard.2010.08.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Winkelmayer WC, Patrick AR, Liu J, Brookhart MA, Setoguchi S. The increasing prevalence of atrial fibrillation among hemodialysis patients. J Am Soc Nephrol. 2011;22(2):349–357. doi: 10.1681/ASN.2010050459. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Buch P, Friberg J, Scharling H, Lange P, Prescott E. Reduced lung function and risk of atrial fibrillation in the Copenhagen City Heart Study. Eur Respir J. 2003;21(6):1012–1016. doi: 10.1183/09031936.03.00051502. [DOI] [PubMed] [Google Scholar]
- 19.Birman-Deych E, Waterman AD, Yan Y, Nilasena DS, Radford MJ, Gage BF. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care. 2005;43(5):480–485. doi: 10.1097/01.mlr.0000160417.39497.a9. [DOI] [PubMed] [Google Scholar]
- 20.Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–1139. doi: 10.1097/01.mlr.0000182534.19832.83. [DOI] [PubMed] [Google Scholar]
- 21.Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med. 2003;348(16):1546–1554. doi: 10.1056/NEJMoa022139. [DOI] [PubMed] [Google Scholar]
- 22.Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis in the United States: Analysis of incidence, outcome, and associated costs of care. Crit Care Med. 2001;29(7):1303–1310. doi: 10.1097/00003246-200107000-00002. [DOI] [PubMed] [Google Scholar]
- 23.Poulose JT, Cartin-Ceba R, Shoja A, et al. Comparison of international classification of Disease–Ninth revision (ICD–9) coding with retrospective case review for the diagnosis of septic shock [abstract] Am J Respir Crit Care Med. 2009;179:A4691. [Google Scholar]
- 24.Quan H, Parsons GA, Ghali WA. Assessing accuracy of diagnosis-type indicators for flagging complications in administrative data. J Clin Epidemiol. 2004;57(4):366–372. doi: 10.1016/j.jclinepi.2003.01.002. [DOI] [PubMed] [Google Scholar]
- 25.Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: Results from the national registry of atrial fibrillation. JAMA. 2001;285(22):2864–2870. doi: 10.1001/jama.285.22.2864. [DOI] [PubMed] [Google Scholar]
- 26.Goldstein LB. Accuracy of ICD-9-CM coding for the identification of patients with acute ischemic stroke: Effect of modifier codes. Stroke. 1998;29(8):1602–1604. doi: 10.1161/01.str.29.8.1602. [DOI] [PubMed] [Google Scholar]
- 27.Benesch C, Witter DM, Jr, Wilder AL, Duncan PW, Samsa GP, Matchar DB. Inaccuracy of the international classification of diseases (ICD-9-CM) in identifying the diagnosis of ischemic cerebrovascular disease. Neurology. 1997;49(3):660–664. doi: 10.1212/wnl.49.3.660. [DOI] [PubMed] [Google Scholar]
- 28.Parsons LS. Using SAS® software to perform a case-control match on propensity score in an observational study. SUGI. 1999:225–229. [Google Scholar]
- 29.Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159(7):702–706. doi: 10.1093/aje/kwh090. [DOI] [PubMed] [Google Scholar]
- 30.Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol. 2005;162(3):199–200. doi: 10.1093/aje/kwi188. [DOI] [PubMed] [Google Scholar]
- 31.Smeeth L, Thomas SL, Hall AJ, Hubbard R, Farrington P, Vallance P. Risk of myocardial infarction and stroke after acute infection or vaccination. N Engl J Med. 2004;351(25):2611–2618. doi: 10.1056/NEJMoa041747. [DOI] [PubMed] [Google Scholar]
- 32.Syrjanen J. Central nervous system complications in patients with bacteremia. Scand J Infect Dis. 1989;21(3):285–296. doi: 10.3109/00365548909035698. [DOI] [PubMed] [Google Scholar]
- 33.Syrjanen J, Valtonen VV, Iivanainen M, Kaste M, Huttunen JK. Preceding infection as an important risk factor for ischaemic brain infarction in young and middle aged patients. Br Med J (Clin Res Ed) 1988;296(6630):1156–1160. doi: 10.1136/bmj.296.6630.1156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Zurru MC, Alonzo C, Brescacin L, et al. Recent respiratory infection predicts atherothrombotic stroke: Case-control study in a Buenos Aires healthcare system. Stroke. 2009;40(6):1986–1990. doi: 10.1161/STROKEAHA.108.535559. [DOI] [PubMed] [Google Scholar]
- 35.Elkind MS, Carty CL, O'Meara ES, et al. Hospitalization for infection and risk of acute ischemic stroke: The Cardiovascular Health Study. Stroke. 2011;42(7):1851–1856. doi: 10.1161/STROKEAHA.110.608588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Zeerleder S, Hack CE, Wuillemin WA. Disseminated intravascular coagulation in sepsis. Chest. 2005;128(4):2864–2875. doi: 10.1378/chest.128.4.2864. [DOI] [PubMed] [Google Scholar]
- 37.Kreutz RP, Bliden KP, Tantry US, Gurbel PA. Viral respiratory tract infections increase platelet reactivity and activation: An explanation for the higher rates of myocardial infarction and stroke during viral illness. J Thromb Haemost. 2005;3(9):2108–2109. doi: 10.1111/j.1538-7836.2005.01474.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Stoddard MF, Dawkins PR, Prince CR, Ammash NM. Left atrial appendage thrombus is not uncommon in patients with acute atrial fibrillation and a recent embolic event: A transesophageal echocardiographic study. J Am Coll Cardiol. 1995;25(2):452–459. doi: 10.1016/0735-1097(94)00396-8. [DOI] [PubMed] [Google Scholar]
- 39.Fatema K, Gertz MA, Barnes ME, et al. Acute weight gain and diastolic dysfunction as a potent risk complex for post stem cell transplant atrial fibrillation. Am J Hematol. 2009;84(8):499–503. doi: 10.1002/ajh.21459. [DOI] [PubMed] [Google Scholar]
- 40.Gibbs HR, Swafford J, Nguyen HD, Ewer MS, Ali MK. Postoperative atrial fibrillation in cancer surgery: Preoperative risks and clinical outcome. J Surg Oncol. 1992;50(4):224–227. doi: 10.1002/jso.2930500405. [DOI] [PubMed] [Google Scholar]
- 41.Hidalgo JD, Krone R, Rich MW, et al. Supraventricular tachyarrhythmias after hematopoietic stem cell transplantation: Incidence, risk factors and outcomes. Bone Marrow Transplant. 2004;34(7):615–619. doi: 10.1038/sj.bmt.1704623. [DOI] [PubMed] [Google Scholar]
- 42.Elijovich L, Josephson SA, Fung GL, Smith WS. Intermittent atrial fibrillation may account for a large proportion of otherwise cryptogenic stroke: A study of 30-day cardiac event monitors. J Stroke Cerebrovasc Dis. 2009;18(3):185–189. doi: 10.1016/j.jstrokecerebrovasdis.2008.09.005. [DOI] [PubMed] [Google Scholar]
- 43.Harvey S, Harrison DA, Singer M, et al. Assessment of the clinical effectiveness of pulmonary artery catheters in management of patients in intensive care (PAC-man): A randomised controlled trial. Lancet. 2005;366(9484):472–477. doi: 10.1016/S0140-6736(05)67061-4. [DOI] [PubMed] [Google Scholar]
- 44.Wheeler AP, Bernard GR, Thompson BT, et al. Pulmonary-artery versus central venous catheter to guide treatment of acute lung injury. N Engl J Med. 2006;354(21):2213–2224. doi: 10.1056/NEJMoa061895. [DOI] [PubMed] [Google Scholar]
- 45.Suissa S. Effectiveness of inhaled corticosteroids in chronic obstructive pulmonary disease: Immortal time bias in observational studies. Am J Respir Crit Care Med. 2003;168(1):49–53. doi: 10.1164/rccm.200210-1231OC. [DOI] [PubMed] [Google Scholar]
- 46.Camm AJ, et al. European Heart Rhythm Association, European Association for Cardio-Thoracic Surgery. Guidelines for the management of atrial fibrillation: The Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC) Europace. 2010;12(10):1360–1420. doi: 10.1093/europace/euq350. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.