Beaubien-Souligny William, Gamarian Ehsan, Côté Jean-Maxime, Neyra Javier A, Baroz Frederic, Adhikari Neill K J, Thorpe Kevin, Bagshaw Sean M, Wald Ron
Division of Nephrology, Centre Hospitalier de l'Université de Montréal, Montreal, Canada.
Centre de Recherche du Centre Hospitalier de l'Université de Montréal, Montreal, Canada.
Crit Care. 2025 May 27;29(1):216. doi: 10.1186/s13054-025-05447-y.
Fluid management is an essential component of renal replacement therapy (RRT) in critically ill patients. Both a positive cumulative fluid balance (CFB) and a high net ultrafiltration (NUF) rate have been reported to be associated with adverse outcomes in epidemiological studies, although the overall trajectory of fluid balance after RRT initiation is not well-described. We aimed to characterize trajectories of fluid management parameters during RRT and analyse the effect of CFB/NUF on outcomes as a trajectory rather than single or aggregated time points over the first week after initiation of RRT.
This is a secondary analysis using fluid balance data focusing on individuals enrolled in the standard-strategy arm of the STARRT-AKI trial who initiated RRT. Cumulative fluid balance (CFB) following RRT initiation and daily net ultrafiltration (NUF) adjusted for body weight during the first 7 days after initiation of RRT were the main independent exposures. We modeled the trajectory of fluid parameters using spline functions and used latent trajectory analysis methods to identify predominant trajectories to compare patients' characteristics and outcomes. We employed logistic regression and multivariable joint longitudinal models to compare the odds and determine the time-dependent association between fluid parameters (CFB and NUF) and 90-day mortality across and within the trajectory classes identified.
We included 855 patients in the primary analysis. After excluding erroneous fluid balance data, we identified two distinct CFB/NUF trajectories. Class A (82.8%) was characterized by a slight increase in CFB and low/stable NUF during the week following RRT initiation while class B (17.2%) was characterized by an increasingly negative CFB with initially higher daily NUF during the first 4 days followed by a stabilization after day 4. In an adjusted analysis, individuals classified in class B were at lower risk for 90-day mortality (aOR: 0.48 CI 0.32; 0.70) p < 0.001) compared to class A. Time-dependent analysis revealed higher CFB was associated with mortality only in those with a class A trajectory (aHR 1.29, 95% CI 1.03-1.55, p = 0.03).
Distinct CFB/NUF trajectories convey prognostic information beyond single-day fluid balance or NUF values and should be considered when formulating or interpreting fluid management strategies.
液体管理是危重症患者肾脏替代治疗（RRT）的重要组成部分。在流行病学研究中，累积液体平衡（CFB）为正和高净超滤（NUF）率均与不良结局相关，尽管RRT开始后液体平衡的总体轨迹尚未得到充分描述。我们旨在描述RRT期间液体管理参数的轨迹，并分析CFB/NUF作为一种轨迹而非RRT开始后第一周内的单个或汇总时间点对结局的影响。
这是一项二次分析，使用的液体平衡数据聚焦于参与STARRT-AKI试验标准策略组且开始RRT的个体。RRT开始后的累积液体平衡（CFB）以及RRT开始后前7天根据体重调整的每日净超滤（NUF）是主要的独立暴露因素。我们使用样条函数对液体参数的轨迹进行建模，并使用潜在轨迹分析方法来识别主要轨迹以比较患者的特征和结局。我们采用逻辑回归和多变量联合纵向模型来比较比值，并确定在已识别的轨迹类别之间和之内液体参数（CFB和NUF）与90天死亡率之间的时间依赖性关联。
我们在初步分析中纳入了855例患者。在排除错误的液体平衡数据后，我们识别出两种不同的CFB/NUF轨迹。A类（82.8%）的特征是RRT开始后一周内CFB略有增加且NUF低/稳定，而B类（17.2%）的特征是CFB越来越负，在开始的4天内每日NUF最初较高，随后在第4天后趋于稳定。在一项校正分析中，与A类相比，分类为B类的个体90天死亡率风险较低（校正比值比：0.48，置信区间0.32；0.70，p<0.001）。时间依赖性分析显示，仅在具有A类轨迹的患者中，较高的CFB与死亡率相关（校正风险比1.29，95%置信区间1.03-1.55，p=0.03）。
不同的CFB/NUF轨迹所传达的预后信息超出了单日液体平衡或NUF值，在制定或解释液体管理策略时应予以考虑。
