Abstract: Background：Thoracic fluid content (TFC) is a volume indicator derived from electrical cardiometry (EC) based on the standardization of thoracic basic impedance. It can reflect the total fluid content in intravascular areas, extravascular areas, and pleural tissue gaps. There are very few clinical studies using TFC to evaluate volume status during continuous renal replacement therapy (CRRT) in critically ill children at home and abroad. Objective：To identify whether TFC can be used as an assessment indicator of volume status during CRRT and guide volume management during CRRT. Design：Prognosis study. Methods：Critically ill children with CRRT ≥ 24 hours in PICU were enrolled into the study. The internal jugular vein, femoral vein or subclavian vein double lumen tube were the vascular pathway for CRRT. We collected the baseline, laboratory, EC measurements, and clinical fluid status indicators. The occurrence of hemodynamic disturbance was taken as the outcome, including hypotension or arrhythmia, requiring volume expansion treatment, and the maximum inotropic score (ISmax) greater than IS. We evaluated the correlation between stroke volume variation (SVV), TFC, and B-type natriuretic peptide (BNP) with ultrafiltration volume at different time points of CRRT(start, 24 hours and end), and investigated whether volume indicators can predict the outcome. Main outcome measures：Predictive factors for hemodynamic disturbance. Results：A total of 43 children with 59 CRRTs were included into the analysis. Compared with the beginning of CRRT, there was a statistically significant difference in TFC, Hct, and Hb at 24 hours of CRRT (P<0.05). Compared with the start of CRRT, there was a statistically significant difference in SVV, TFC, ICON, BNP, Hct, Scr, BUN, Hb, and ALB (P<0.05) at the end of CRRT. FO is significantly positively correlated with TFC and SVV at the start of CRRT. There is a significant positive correlation between ultrafiltration volume/body weight and TFC (start), and a significant negative correlation with ΔTFC(24 h), ΔTFC(end), TFCd0% (24 h) and TFCd0%(end). The ultrafiltration rate was positively correlated with SVV, TFC, BNP at the start of CRRT,and was negatively correlated with ΔTFC(24 h),ΔTFC(end), and TFCd0% (24 h). The balance/weight was positively correlated with ΔTFC(24 h),ΔTFC(end), TFCd0% (24 h), TFCd0%(end) and ΔBNP(end), and was negatively correlated with BNP (start). A total of 33 hemodynamic disturbance events occurred during CRRT. There is a statistical difference in BNP (start), ΔBNP(end), ultrafiltration rate, TFCd0% (24 h) between groups with and without adverse hemodynamic events. Multivariate analysis showed that TFCd0%（24 h）≤-3.64 kΩ^-1（OR=8.84，95%CI：1.16~67.44，P=0.036）and ΔBNP(end)≤-133.8 pg·mL^-1（OR=5.67，95%CI：1.24~26.03，P=0.007）were more likely to cause hemodynamic disorders. Conclusion：TFC can be used to monitor the fluid status of critically ill children during CRRT treatment. TFCd0%≤-3.64 kΩ^-1 after 24 hours of CRRT can be a predictive factor for hemodynamic disturbance.

Key words: Critically ill children, Continuous renal replacement therapy, Thoracic fluid content, Volume assessment, Volume assessment