Threemonth landmark analysis death, N 3,843 Male gender, n Age year Platelet 09L Missing, n Prothrombin time second Missing, n Albumin gL Missing, n Total bilirubin lmolL Missing, n Creatinine lmolL Missing, n ALT IUL Missing, n AST IUL Missing, n MELD score Missing, n ChildPugh score Missing, n APRI Missing, n Forns index Missing, n FIB4 index Missing, n AFP IUml Missing, n Positive HBeAg, n Missing, n HBV DNA, log IUml Missing, n HCC Title Loaded From File treatment options, n OT only LAT only TACE only LT only Combination therapy, n OT other people LT other people LAT other individuals TACE other individuals Any TACE NA use, n Duration of NA use year Lamivudine Adefovir dipivoxil Entecavir Telbivudine TDFTAF Any NA Concomitant drugs, n Oral hypoglycaemic agents Metformin Insulin Statins NSAID No NA, n 407 328 80.six 65.9 11.6 174.0 102.five 1 0.two 13.two two.2 4 1.0 30.eight six.two 0 0.0 25.0 15.8 0 0.0 87.5 54.four 0 0.0 95.0 46.0 200.five 0 0.0 86.0 42.5 189.0 180 44.two 9.six 2.six 0 0.0 6.four 1.0 0 0.0 1.five 0.7 three.five 180 44.2 8.0 2.0 316 77.6 4.3 two.0 eight.six 180 44.two 65.5 7.8 1,097.8 three 0.7 33 21.three 252 61.9 two.74 1.00 five.38 371 91.two 91 22.four 38 9.three 251 61.7 0 0.0 27 six.6 20 four.9 0 0.0 17 4.2 18 four.four 269 66.1 0 0 0.0 0.0 0.0 0.0 0.0 0.0 PreHCC NA, n 2,932 two,402 81.9 61.7 10.three 130.7 65.6 8 0.three 13.9 3.0 11 0.four 33.eight six.1 0 0.0 27.four 27.3 0 0.0 91.eight 66.six 0 0.0 120.5.Imputation model. A 3month landmark analysis i.e. by excluding patients who died in the course of the three months exposure period and those that had followup of less than three months, with followup duration up to 5 years from the landmark date, was adopted to prevent immortal time bias in patients who received remedy immediately after baseline. A 6month landmark analysis was conducted as sensitivity evaluation. Inverse probability of therapy weighting IPTW utilizing propensity scores PSs was utilised in our secondary evaluation. We estimated PS, the conditional probability of receiving NA prior to HCC, by generalised boosted models. The balance of covariates by PS was assessed by 2 summary statistics the absolute standardised mean distinction ASMD plus the KolmogorovSmirnov KS statistic. Both imply and maximum of either ASMD or KS were the four stopping rules for assessing the covariate balance among no NA, preHCC NA, and postHCC NA i.e. ASMD 0.two andor KS 0.1 as an indication of imbalance.18 PS weights of average therapy effect around the treated had been calculated by the stopping rule with the greatest productive sample size plus the best subgroup balance.