Reply to the letter regarding ‘Prognostic value of left atrial reverse remodelling in patients hospitalized with ADHF’

Abstract

We thank Hao Zhang et al. for their interest in our work and would like to respond to their comments regarding our study.¹

In our study, because we defined the primary endpoint as a composite of heart failure (HF) rehospitalization or cardiovascular death, a Cox regression analysis was conducted. We acknowledge that using a Cox regression model can lead to an overestimation of the cumulative event incidence in the presence of competing events. Therefore, additional analyses using Fine–Gray competing risks models were conducted. Fine–Gray competing risks models also demonstrated that patients with left atrial reverse remodelling (LARR) had significantly better prognoses for cardiovascular death (P = 0.026) and HF rehospitalization (P = 0.0007). Among the 117 individuals who reached the primary endpoint, 109 experienced rehospitalization for HF, and 8 died due to cardiovascular events (5 related to HF and 3 due to non-HF causes). This finding indicated that the majority of events were related to HF and that the incidence of competing events (i.e., cardiovascular death without HF worsening) was quite low. Therefore, we believe that the implication of using a competing risk model in this context is limited.

We agree that cardiogenic shock (CS) is associated with poor outcomes; however, other factors such as age and chronic kidney disease also affect long-term prognosis.^{2, 3} To investigate the relationship between CS and long-term outcomes, it is crucial to consider the presence of CS during the current hospitalization and also any previous CS history. However, collecting this historical information poses challenges. In this present study, patients with in-hospital death were excluded, and selected patients who underwent two-point echocardiography with adequate quality for calculating LARR were included. Thus, the number of patients with CS was low, leading us to conclude that CS was not an appropriate factor for multivariate analysis in this study.

Frailty is another important factor for patients with HF. Recent reports indicate that physical frailty, cognitive dysfunction and social isolation are associated with poor outcomes in patients with HF.^{4, 5} In this study, because of the lack of data on cognitive dysfunction and social environment, we focused our additional analysis solely on physical frailty. Similar to previous reports,^{6, 7} physical frailty emerged as a significant prognostic factor (P = 0.031). However, in the multivariate analysis, the significance of frailty was attenuated, suggesting that factors such as male sex, anaemia, no beta-blocker use, non-LARR and history of prior HF hospitalization had more substantial effects on adverse events.

Although diabetes itself may worsen HF, it is also associated with other risk factors such as ischaemic heart disease and renal dysfunction. In the Seattle Heart Failure Model, a well-established risk score for mortality in patients with HF, diabetes is not included in the calculation.⁸ In our study, because diabetes did not show a significant difference in the univariate analysis (P = 0.358), we did not include it in the multivariate analysis. Hao Zhang et al. proposed a subgroup analysis of diabetes; however, since only 40% of the participants had diabetes, the power was insufficient for a meaningful subgroup analysis. Furthermore, the age at diabetes onset was not recorded in our study, making it difficult to incorporate this variable into additional analyses. We believe that further investigation is warranted in the future.