Selection of Estimation Influences the Observed Mortality Threshold at 13 mL/kg/h

By
Shuhao Mei
Hailian Yi
Yuyin Han
Boran Dong
Yong Liu
Xiaoyang Gong
September 15, 2026

Critical Care

Background

The association between CRRT dosing and patient outcomes is critical in managing acute kidney injury in critically ill patients. Variability in analytical methods can significantly impact the interpretation of mortality data.

Data Highlights

Estimate	Hazard Ratio (HR)	95% Confidence Interval (CI)
Conventional Analysis	1.70	1.16–2.49
Marginal Structural Analysis	1.08	0.74–1.58
Alternative Estimate	0.87	0.66–1.14

Key Findings

Very-low-dose CRRT (<13 mL/kg/h) is associated with increased 90-day mortality (OR, 1.72; 95% CI, 1.31–2.24).
Different analytical models yield varying hazard ratios for CRRT dosing, affecting mortality interpretations.
The conventional Cox model reported an HR of 1.70, while the marginal structural model indicated an HR of 1.08.
Adjustment for time-dependent factors in the marginal structural analysis nullified the baseline association.
Selection of estimates for meta-analysis can significantly influence pooled results and their clinical interpretation.

Clinical Implications

Clinicians should be aware of the variability in mortality estimates based on the analytical methods used in studies of CRRT dosing.

Conclusion

The findings underscore the importance of methodological transparency in research on CRRT dosing and its impact on mortality, highlighting the need for consistent reporting practices.