Mediated relationship within new-onset atrial fibrillation, disease severity, and mortality 

We read with great interest the recent study by Dr. Hellman,¹ in which they investigated the association between new-onset atrial fibrillation (NOAF) and mortality in patients with renal replacement therapy (RRT). In their study, the key finding is that after adjusting for confounding factors, only NOAF before RRT, but not NOAF during RRT, was an independent risk factor for 1-year mortality. This study is well designed. We want to add some comments.

In this study, understanding whether NOAF before RRT is a real risk factor for mortality is important as it determines whether NOAF before RRT deserves intervention. The authors found that compared with patients with NOAF during RRT, those with NOAF before RRT had significantly higher APACHE II and SOFA scores. Thus, they concluded that NOAF before RRT may be caused by higher disease severity and more compromised hemodynamics, which resulted in higher mortality. However, due to the retrospective nature, whether NOAF before RRT was a cause for increased mortality or this association was only a reflection between severe disease severity and mortality remains unclear. In this case, we suggest that mediation analysis^2,3 may help us explore whether the impact of NOAF before RRT is entirely or partially mediated by disease severity score and understand which one is the ‘real’ risk factor in this relationship.

In Figure 1, coefficient ‘D’ represents the total effect that NOAF before RRT was associated with high mortality (findings in the present study). Coefficient ‘A*B’ (indirect effect) indicates whether the association between NOAF before RRT and mortality was mediated by disease severity. Coefficient ‘C’ represents the direct effect between NOAF before RRT and mortality. In this mediation analysis, if coefficient ‘C’ became insignificant, it means that the association between NOAF before RRT and mortality was entirely mediated by disease severity, which indicates that NOAF before RRT is not a real risk factor for mortality. If coefficient ‘A*B’ was insignificant, it suggests that NOAF before RRT is a direct cause for increased mortality.

Figure 1

A diagram of mediation analysis.

Open in new tabDownload slide

Finally, this study added important information to the current knowledge. Their valuable work is sincerely appreciated.

References

Author notes