Table 1

Markov inputs in cost-effectiveness analysis

Variable	Baseline	Range low	Range high	Distribution	Reference
Survival input
HR of HAIC-FO versus sorafenib for OS	0.408	0.301	0.553	Normal	13
HR of HAIC-FO versus sorafenib for PFS	0.451	0.340	0.598	Normal	13
Weibull OS survival model with HAIC-FO	λ 0.0170272, γ 1.3635872				13
Weibull PFS survival model with HAIC-FO	λ 0.071169, γ 1.037127				13
Weibull OS survival model with sorafenib	λ 0.0151598, γ 1.8083652				13
Weibull PFS survival model with sorafenib	λ 0.0888648, γ 1.3537234				13
Utility input
Utility of PFS	0.760	0.610	0.910	Beta	17
Utility of PD	0.680	0.540	0.820	Beta	17
Disutility due to neutropenia	0.090	0.006	0.120	Beta	30
Disutility due to fatigue	0.073	0.037	0.110	Beta	30
Cost input
Sorafenib (per month)	3077.760	2462.208	3693.312	Gamma	17
Oxaliplatin (per month)	364.413	291.531	437.296	Gamma	17
Cisplatin (per month)	14.310	11.448	17.172	Gamma	31
Fluorouracil (per month)	686.200	548.960	823.440	Gamma	17
Leucovorin (per month)	31.653	25.323	37.984	Gamma	17
HAIC procedure (per month)	2422.707	1938.165	2907.248	Gamma	17
Hospitalisation (per month)	502.560	402.048	603.072	Gamma	17
Test (per month)	469.587	375.669	563.504	Gamma	17
Second line (per month)	959.160	767.328	1150.992	Gamma	9
Body surface area	1.720	1.380	2.060	Normal	17
Discount rate	0.030	0.000	0.050	Fixed	17

Regarding the choice of survival distribution, we evaluated multiple potential distributions, including Weibull, log-logistic, log-normal, gamma, Gompertz and exponential distributions. We selected the Weibull distribution based on the principle of minimum Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) values, as smaller AIC and BIC values indicate a better model fit.
AST, aspartate aminotransferase; FO, oxaliplatin+fluorouracil; HAIC, hepatic arterial infusion chemotherapy; OS, overall survival; PD, progression disease; PFS, progression-free survival.