Table 3.

Multivariable model comparing TIPS vs. serial LVP to predict change in eGFR after 90 days, stratified by baseline eGFR and for all subjects

	Parameter Estimate (95% CI)		P value
Baseline eGFR <60 mL/min/1.73 m2
TIPS, vs serial LVP as reference	21.27 (13.07 to 29.47)		<0.001
Age, per 1-y older	−0.21	[−0.61, 0.20]	0.3
Baseline eGFR, per 1-mL/min/1.73 m2 greater	−0.08	[−0.34, 0.19]	0.6
Etiology of cirrhosis, vs alcohol as reference
NASH	6.25	[−10.17, 22.66]	0.5
HCV infection	11.37	[−0.08, 22.66]	0.05
Mixed etiology	9.34	[−2.25, 20.92]	0.1
Other	3.71	[−6.87, 14.30]	0.5
Baseline eGFR ≥60 mL/min/1.73 m2
TIPS, vs serial LVP as reference	1.37	[−9.48, 12.21]	0.8
Age, per 1-y older	−0.51	[−1.09, 0.06]	0.08
Baseline eGFR, per 1-mL/min/1.73 m2 greater	−0.12	[−0.45, 0.20]	0.5
Etiology of Cirrhosis, vs alcohol as reference
NASH	3.38	[−16.65, 23.41]	0.7
HCV infection	3.66	[−9.95, 17.27]]	0.6
Mixed etiology	3.05	[−13.48, 19.58]	0.7
Other	−3.68	[−18.62, 11.28]	0.6
All Participants
TIPS, vs serial LVP as reference	12.71	[6.50, 18.92]	<0.001
Baseline eGFR*, per 1-mL/min/1.73 m2 greater	0.01	[−0.13, 0.14]	0.9
Interaction Term: eGFR + TIPS	−0.33	[−0.53, −0.13]	0.001
Age, per 1-y older	−0.32	[−0.63, 0.00]	0.05
Etiology of cirrhosis, vs alcohol as reference
NASH	2.83	[−9.40, 15.06]	0.7
HCV infection	6.96	[−1.47, 15.39]	0.1
Mixed etiology	6.13	[−3.26, 15.52]	0.2
Other	0.04	[−8.22, 8.30]	0.9

^*Baseline eGFR centered at 60 mL/min/1.73 m² for interaction analysis.

Note: Baseline eGFR was measured immediately before TIPS placement, or closest available to day 0 of follow-up or the LVP cohort. Parameter Estimate refers to the expected change in eGFR (in mL/min/1.73 m²) over 90 days, with respect to each predictor, while holding other factors constant.

HCV, hepatitis C virus; TIPS (transjugular intrahepatic portosystemic shunt), LVP (large-volume paracentesis), eGFR (estimated glomerular filtration rate), CI (confidence interval), NASH (non-alcoholic steatohepatitis)