CVVHD results in longer filter life than pre-filter CVVH: Results of a quasi-randomized clinical trial

Abstract

Background Filter clotting is a major issue in continuous kidney replacement therapy (CKRT) that interrupts treatment, reduces delivered effluent dose, and increases cost of care. While a number of variables are involved in filter life, treatment modality is an understudied factor. We hypothesized that filters in pre-filter continuous venovenous hemofiltration (CVVH) would have shorter lifespans than in continuous venovenous hemodialysis (CVVHD). Methods This was a single center, pragmatic, unblinded, quasi-randomized cluster trial conducted in critically ill adult patients with severe acute kidney injury (AKI) at the University of Iowa Hospitals and Clinics (UIHC) between March 2020 and December 2020. Patients were quasi-randomized by time block to receive pre-filter CVVH (convection) or CVVHD (diffusion). The primary outcome was filter life, and secondary outcomes were number of filters used, number of filters reaching 72 hours, and in-hospital mortality. Results In the intention-to-treat analysis, filter life in pre-filter CVVH was 79% of that observed in CVVHD (mean ratio 0.79, 95% CI 0.65–0.97, p = 0.02). Median filter life (with interquartile range) in pre-filter CVVH was 21.8 (11.4–45.3) and was 26.6 (13.0–63.5) for CVVHD. In addition, 11.8% of filters in pre-filter CVVH were active for >72 hours, versus 21.2% in the CVVHD group. Finally, filter clotting accounted for the loss of 26.7% of filters in the CVVH group compared to 17.5% in the CVVHD group. There were no differences in overall numbers of filters used or mortality between groups. Conclusions Among critically patients with severe AKI requiring CKRT, use of pre-filter CVVH resulted in significantly shorter filter life compared to CVVHD. Trial registration ClinicalTrials.gov, NCT04762524. Registered 02/21/21—Retroactively registered, https://clinicaltrials.gov/ct2/show/NCT04762524?cond=The+Impact+of+CRRT+Modality+on+Filter+Life&draw=2&rank=1.