Biliary tract cancer's (BTC) treatment main stone for advanced stages is constituted by chemotherapy. Surgical centralization and physicians' confidence in the use of new technologies and molecular analysis turned out to be of interest and potentially influencing survival. After applying a random-effect model, the relationship between each clinical variable on the main outcome was investigated through multilevel mixed-effects logistic regression. The risk-standardized outcomes were calculated for each centre involved. In the unadjusted cohort the median survival was 8.6 months (95%C.I.: 7.8-9.3) with a 9-month survival rate of 48.3% (95%C.I.: 45.0-51.5). A substantial heterogeneity across hospitals was found (I2: 70.3%). In multilevel mixed effect logistic regression, male, being treated for gallbladder cancer, higher ECOG, increased NLR, CEA and Ca 19.9 and low value of haemoglobin showed to increase the odds for 9-month mortality. The model estimated that the residual variance observed in 9-month mortality was attributable for the 2.6% to the treating hospital. Through a multilevel mixed effect model, average risk-standardized mortality within 9 months was 50.1%. As noticeable, all hospital's risk-standardized mortality falls within 95%C.I., thus all participating centres provided similar outcomes when adjusted for patient case-mix. Heterogenicity between hospital did not affect the outcome in term of overall survival.
Keywords: Biliary tract cancer; chemotherapy; cholangiocarcinoma; cisplatin; gemcitabine; physician's expertise; risk-adjusted analysis; variability between institutions.