Association of fluid balance trajectories with multi-timepoint prognosis in heart failure: a group-based trajectory model analysis
Yue Xi, Yining Han, Q Sun, Pengxiang Zhu, Jiaxin Guo, Beining Zhang, Jiacheng Fan, Xiaofeng Li
Heart failure (HF) exhibits persistently high mortality rates in intensive care unit (ICU) with extremely poor prognosis. Abnormal fluid load and fluid retention constitute the core pathological mechanisms underlying the increased mortality risk in HF patients. Fluid balance (FB) management is a critical component of HF treatment; however, existing studies predominantly focus on single-point assessments of fluid status, failing to comprehensively elucidate the dynamic relationship between fluid dynamics and patient outcomes. We extracted clinical data of ICU HF patients from MIMIC-IV v3.1. The primary outcome was 28-day all-cause mortality after ICU admission, and secondary outcomes were 90-day and 365-day all-cause mortality. Group-based trajectory modelling (GBTM) was used to identify 7-day FB trajectory groups, supplemented by survival analysis, multivariate Cox regression, restricted cubic spline, subgroup and sensitivity analyses. A total of 1,593 patients were enrolled, and four distinct FB trajectories were identified with high posterior probabilities: stable balance (62.52%), significantly decreased to negative balance (25.86%), low fluctuation (6.21%), and high fluctuation persistent positive balance (5.40%). Survival analysis showed significant differences in mortality across groups ( P < 0.001), with the high fluctuation positive balance group having the worst prognosis and the significant negative balance group having the best. Adjusted regression confirmed early negative FB reduced mortality risk, while persistent positive FB increased risk. Total FB was linearly positively correlated with mortality risk. The GBTM method facilitates the identification of subgroups of HF patients in the ICU with distinct risk characteristics. Short-term, medium-term, and long-term survival outcomes are optimal when fluid status significantly declines to negative balance, whereas HF patients with fluid status in positive balance require focused monitoring for mortality risk.