New atrial fibrillation causes stroke, death in people with severe sepsis? (JAMA)
Severe sepsis makes the heart irritable, probably due to all the evil humors and increased cardiac demand. Between 6-20% of patients with severe sepsis develop atrial fibrillation for the first time; that's old news. What's been unclear is what new-onset atrial fibrillation in severe sepsis means: is it an expected, yeah-so-what marker of critical illness, or does it signal more severe disease, predicting mortality or post-hospital disability?
Allan Walkey et al used the California State Inpatient Database to identify 49,082 hospitalizations with an ICD-9 billing code for severe sepsis. 2,896 of these patients also had a new diagnosis of atrial fibrillation during the hospitalization (14%). (Diagnoses in this database were identified as being present on admission, or not.)
Results:
Compared with hospitalized patients without severe sepsis, patients with severe sepsis had an increased risk of new-onset atrial fibrillation (demographics- and comorbidity-adjusted odds ratio, 6.82; 95% CI, 6.54-7.11; P < .001).
Age, being more severely ill either acutely (organ failures) and/or chronically (chronic illnesses), all increased the likelihood of developing new-onset atrial fibrillation with severe sepsis.
That's all interesting and helpful, and doesn't mess up my day. But then the good Dr. Walkey and team stirred the pot a bit. They further queried the database regarding the incidence of in-hospital stroke, especially embolic stroke, and its potential relationship to mortality (can you see where this may be going)?
People with severe sepsis and new onset atrial fibrillation had an unadjusted in-hospital mortality of 56%, significantly higher than those who came in with pre-existing A-fib (44% mortality) or those with no previous A-fib documented (38%). Sensitivity analyses, including adjusting for number of organ failures, did not change the relationship, although I'm not qualified to critique these methods or their likelihood of uncovering truth here.
The real conundrum emerges from their report of the in-hospital ischemic stroke rates in people with severe sepsis:
2.6% in those with new-onset A-fib;
Only 0.57-0.69% in those with pre-existing or no A-fib.
After controlling for multiple variables using 4 regression models, and nearest-neighbor-matching patients on their likelihood of new-onset a-fib ...
The odds ratio for ischemic stroke was 2.75 - 4.0 for those with new-onset atrial fibrillation, compared to those without.
The odds ratio for mortality was 1.1 - 1.3. (all stat.significant)
As a final epidemiological flourish, they set out to demonstrate that this elevated stroke risk from severe-sepsis-induced atrial fibrillation persisted after hospitalization. They found a handful of people (~400 or so of a cohort of tens of thousands) who were re-hospitalized with a new ischemic stroke after a hospitalization for severe sepsis. Those who had had new-onset a-fib during their severe sepsis admission were disproportionately represented (they had a 2% risk vs. 1.3-1.5% for those without new-onset atrial fib), a hazard ratio of 1.5 which just missed statistical significance.
What are you trying to do, make us anticoagulate all these patients, Dr. Walkey? (I guess at least they won't be on Xigris.)
Walkey AJ et al. Incident Stroke and Mortality Associated With New-Onset Atrial Fibrillation in Patients Hospitalized With Severe Sepsis. JAMA ePub November 13, 2011. FREE FULL TEXT
In an excellent editorial, fellow outcomes numbers-crunching physician-investigators Christopher Goss and Shannon Carson point out some potential limitations of the data, while being cautiously impressed by the robustness of the findings to the multiple sensitivity analyses. (It sounds like they don't want to anticoagulate these patients, either.)
Goss CH, Carson SS. Is Severe Sepsis Associated With New-Onset Atrial Fibrillation and Stroke? JAMA ePub November 13, 2011.