We read the recent study by Ferrera et al highlighting the role of atrial fibrillation (AF) in patients with infective endocarditis. We would like to offer our thoughts on the correlation of heart failure (HF) with AF and the microbiology in these patients. As correctly pointed out by the investigators it is indeed difficult to establish a temporal relation between HF and AF in these patients. However, on closer evaluation of their cohort, we note the following points: (1) in patients with new-onset AF, the left atrial (LA) diameter was significantly larger than in patients maintaining sinus rhythm and (2) 53% of their patients presented with “heart failure” per the investigators, which as we understand, is an episode of acute decompensation. LA dimension (defined by LA volume index) in patients with normal filling pressures and normal left ventricular (LV) ejection fraction (EF) is an independent mortality predictor. LAVI has been noted to be an accurate marker of chronic HF with preserved EF and along with filling pressures and tissue Doppler imaging, a marker of acute HF. Taken in combination, it may be speculated that these patients had chronic HF with acute decompensation during their index admission. Furthermore, the distinction between HF with reduced or preserved EF in this cohort is crucial because the consequences of AF would be more pronounced in the HF with preserved EF cohort because of loss of late diastolic filling.

Although the investigators have used radiographic criteria for diagnosis of HF, a meta-analysis demonstrated chest radiography to be only 76% specific and 67% sensitive in contrast to more robust diagnostic techniques. In addition, the present study does not elaborate on New York Heart Association classes, cardiac biomarker levels, LV function, including LVEF, filling pressures, and tissue Doppler imaging velocities. This lack of distinction between acute versus chronic and compensated versus decompensated HF could have altered the assessment of HF during multivariate analysis because of inappropriate risk stratification. Furthermore, in their multivariate analysis, the investigators highlighted HF as an independent predictor of mortality and that has been shown to be true across multiple studies in patients with infective endocarditis resulting in a class IB indication for surgery. However, their conclusion (as highlighted in the discussion and Figure 1) that AF leads to HF in a unidirectional fashion, and not using a bidirectional relation, is not robustly supported by their data because of the reasons mentioned previously.

The previously mentioned observations lead us to speculate that HF may have been underdiagnosed in their cohort and could have been a driver for AF in addition to the opposite relation. As such, Ferrera et al’s hypothesis of AF being a marker rather than a mediator for mortality might be extendable to HF too and merits further investigation evaluating causation rather than association.

From a microbiological standpoint, we note that non–community-acquired (i.e., health care–acquired) infective endocarditis was higher in the new-onset AF cohort, which is an independent marker of worse prognosis. This information, in combination with the lack of data on how many eligible patients did not receive cardiac surgery due to increased risk of surgery, severity of illness, or patient/provider preference, can potentially confound the assessment of mortality in this cohort and need further elaborating before attributing these differences to AF.