In 2017, we interviewed 23 non-hepatologists regarding chronic hepatitis B (CHB) using in-depth interviews. The rate of introducing patients to hepatologists was significantly higher among physicians (13/17, 76%) than that among surgeons (1/6, 17%, p = 0.036). Most non-hepatologists considered that the main reasons they did not introduce patients with CHB to hepatologists were a lack of time for preparing patient referral documents and ignorance regarding how introduce patients. They also highlighted that to improve the introduction rate, simplifying the patient referral document and creating an easy medical reservation system are required. Based on this research, in August 2017 we introduced the "patient referral document for hepatitis" in our hospital. Since then, the number of patients being introduced has increased by 1.6 times compared with the previous year. Based on the additional data, the construction of a more effective introduction system is recommended.

We performed a retrospective validation and accuracy study for the prognosis prediction model in 100 end-stage hepatocellular carcinoma patients who underwent palliative care at our hospital. Using the receiver operating characteristic analysis and the area under the curve (AUC), we selected the cut-off value for 3-week survival, and the predictive ability was evaluated using sensitivity, specificity, positive predictive value, negative predictive value, and accurate diagnosis rate. An examination of the various models showed that the Palliative Prognostic Index and the Biological Prognostic Score (BPS) version 3 had excellent predictive performance with AUC 0.89 and 0.82, respectively and diagnostic accuracy rate of 80 and 79, respectively. The BPS version 2 and Model for End-Stage Liver Disease score had fair predictive performance with AUC 0.72 and 0.71, respectively, and diagnostic accuracy rate of 72 and 70, respectively. The use of these models enables the stratification of prognostic prediction.

In order to demonstrate the evaluation of basic assessments and the clinical utility of a fully automated highly sensitive chemiluminescent enzyme immunoassay for detecting hepatitis B surface antigen (Lumipulse Presto HBsAg-HQ), the HBsAg-HQ assay was compared to a conventional assay (HISCL HBsAg). The sensitivity of HBsAg-HQ (0.005 IU/mL) is approximately 6-fold higher than that of HISCL (0.03 IU/mL). The concordance rates between HBsAg-HQ and HISCL for 1,312 clinical sera were 99.6%; 5 samples were positive for only HBsAg-HQ. The specificity of HBsAg-HQ was 99.84% (1,256/1,258). Higher sensitivity of HBsAg-HQ was also confirmed with seroconversion panels and dilution tests of each genotype. Moreover, in 9/13 cases with HBV reactivation, HBsAg-HQ was able to detect HBsAg from the sera undetectable with HISCL, and 3 of them had earlier detection than HBV-DNA. Thus, HBsAg-HQ is more sensitive for detecting occult HBV infection and HBV reactivation in the early phase.