Implementation quality as a modifiable determinant of filter lifespan in regional citrate anticoagulation: a real-world clustering-adjusted study

Figures

• 

  Figure 1.. RCA group monitoring quality analysis. (A) Distribution of monitoring quality based on timely measurement: 58.4% of RCA sessions had post-filter iCa2+ measurement within 2 h of initiation (defined as timely monitoring). (B) Impact of timely monitoring on filter clotting: timely monitoring was associated with significantly lower clotting incidence (16.1% vs. 30.4%, p = 0.008). (C) Distribution of initial citrate infusion rates. (D) Target attainment by initial citrate infusion rate: proportion of sessions achieving post-filter iCa2+ ≤0.35 mmol/L (on-target) versus >0.35 mmol/L (off-target). Lower citrate rates (<200 mL/h) were associated with higher off-target proportions. (E) Dose–response relationship: probability of filter clotting by time to first post-filter iCa2+ measurement (continuous), with 95% confidence bands.

• 

  Figure 2.. Statistical modeling framework. Overview of the analytical approach used to evaluate anticoagulation strategies in 420 CRRT sessions. Patient-level clustering was addressed using mixed-effects models or robust standard errors. The primary outcome (filter lifespan) was analyzed using mixed-effects Cox and Fine–Gray competing risk models. Secondary binary outcomes (filter clotting) were analyzed using mixed-effects logistic regression. Predefined sensitivity analyses are listed at the bottom.

• 

  Figure 3.. Anticoagulation strategies distribution. (A) Overall distribution across 420 CRRT sessions. RCA was the most frequently used strategy (58.3%, n = 245), followed by heparin (23.3%, n = 98) and nafamostat (18.3%, n = 77). (B) Anticoagulation strategy by machine type. RCA and nafamostat were predominantly used with Fresenius machines, while heparin was more commonly used with other machine brands.

• 

  Figure 4.. Kaplan–Meier curves for filter survival free of clotting by anticoagulant group. RCA (blue) and nafamostat (green) present similar survival curves, both superior to heparin (red). Log-rank p < 0.001.

• 

  Figure 5.. Mixed-effects Cox regression forest plot for filter clotting risk. Hazard ratios with 95% CIs for predictors of filter clotting, accounting for patient-level clustering. Heparin use is associated with a significantly increased risk compared with RCA (reference). Error bars represent 95% CIs. The vertical dashed line indicates a hazard ratio of 1.0.