Graduate Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan
Genomics Research Center, Academia Sinica, Taipei, Taiwan
Address reprint requests to: Chien-Jen Chen, Sc.D., Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Taipei 11529, Taiwan. E-mail: email@example.com; fax: +886-2-2785-3208.
Potential conflict of interest: Nothing to report.
Hepatitis B immunization has been documented to prevent fulminant hepatic failure (FHF) and hepatocellular carcinoma (HCC) by historical comparison studies in Taiwan. This study aimed to assess long-term risks and predictors of various liver diseases associated with incomplete immunization in 3.8 million vaccinees. Profiles of the National Hepatitis B Immunization Registry, National Cancer Registry, and National Death Certification Registry were linked to ascertain newly diagnosed cases of HCC and deaths from FHF and chronic liver diseases (CLDs) from infancy to early adulthood of 3,836,988 newborn vaccinees. Cox's proportional hazards models were used to estimate hazard ratios (HRs) for various risk predictors. There were 49 newly developed cases of HCC, 73 deaths from FHF, and 74 deaths from CLDs during the follow-up of 41,854,715 person-years. There were striking differences between unvaccinated and vaccinated newborns after the launch of a national immunization program for HCC incidence (0.293 vs. 0.117 per 100,000 person-years), FHF mortality (0.733 vs. 0.174 per 100,000 person-years), and CLD mortality (2.206 vs. 0.177 per 100,000 person-years). Among vaccinees, incomplete immunization was the most important risk predictor of HCC, FHF, and CLDs, showing an HR (95% confidence interval, P value) of 2.52 (1.25-5.05; P = 0.0094), 4.97 (3.05-8.11; P < 0.0001), and 6.27 (3.62-10.84; P < 0.0001), respectively, after adjustment for maternal hepatitis B serostatus. Conclusion: Hepatitis B immunization can significantly prevent the long-term risk of HCC, FHF, and CLDs from infancy to early adulthood. Incomplete immunization with hepatitis B immunoglobulin or vaccines was the most important risk predictor of the liver disease among vaccinees. (Hepatology 2014;60:125–132)
Hepatitis B virus (HBV) is an important risk factor for fulminant hepatic failure (FHF), chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma (HCC). Chronic HBV infection is a worldwide public health challenge.[1, 2] Before the national immunization program was implemented in Taiwan, over 90% of the general population under the age of 40 years was infected with HBV, and 15%-20% had chronic HBV infection.[3-5] The risk of chronic infection depends largely on the age of infection; 40% of infants born to hepatitis B surface antigen (HBsAg)-carrier mothers will also become chronic carriers early in the postnatal period.[6-8] Acquired infections at the perinatal stage causes an estimated 20% of HBV-related deaths worldwide. It is most important to verify and block the major route of HBV transmission to prevent hepatitis B infection.[9, 10]
A national hepatitis B immunization program was launched in Taiwan in July 1984, which had a comprehensive coverage after its implementation.[11, 12] A series of surveys on seroprevalence of HBsAg in children has been carried out before and after the national vaccination program to assess the effectiveness of this national immunization program to prevent chronic HBV infection in vaccinated children in Taiwan. There has been a steady decrease in seroprevalence since 1984,[13-17] and seroprevalence of children born after the launch of the program decreased to 0.9% in 2009. The decrease in HBsAg carrier rate was also reported in China after implementation of the immunization program since 1992.
Several studies have documented the effectiveness of the national hepatitis B immunization program to lower mortality from fulminant hepatitis and incidence of HCC among vaccinated birth cohorts, in comparison with unvaccinated birth cohorts, in Taiwan.[20-24] Reduction in childhood HCC after introduction of selective hepatitis B vaccination for infants born to carrier mothers was also observed in Japan. Most recently, a study in Alaska reported the elimination of HCC and acute hepatitis B in children 25 years after a hepatitis B immunization program. In a recent research letter, Chiang et al. used two secondary databases of the National Death Certificates (1977-2011) and the National Cancer Registry (1977-2009) to show a continuous decline of chronic liver diseases (CLDs) and HCC mortality and HCC incidence for birth cohorts born after implementation of the program in Taiwan. Most studies on the effectiveness of immunization programs were based on the historical comparison of immunized and unimmunized cohorts at the national or community level. The importance of complete immunization and the risk predictors of liver diseases for vaccinees have never been evaluated comprehensively at the individual level in these studies.
Through the linkage of individual data from computerized profiles of the National Hepatitis B Immunization Registry, National Cancer Registry, and National Death Certification Registry of 3.8 million vaccinees, this study aimed to (1) assess the long-term incidence of HCC and mortality of FHF and CLDs associated with incomplete immunization and (2) identify risk predictors of HCC, FHF, and CLD risk among vaccinated children.
Patients and Methods
National Hepatitis B Immunization Program
The national hepatitis B immunization program in Taiwan was implemented in 1984. During the first 2 years (July 1984 to June 1986) of the program, only newborns to high-risk mothers (seropositive for HBsAg) were vaccinated. Healthy newborns of highly infectious carrier mothers (seropositive for hepatitis B e antigen [HBeAg] or with a high serotiter of HBsAg) received an additional 0.5 mL of hepatitis B immunoglobulin (HBIG) within 24 hours after birth. All newborns were vaccinated with four doses of plasma-derived hepatitis B vaccine at 0, 1, 2, and 12 months of age since July 1986. After November 1, 1992, the vaccine used in the program was changed to recombinant yeast vaccine with three doses at 0, 1, and 6 months of age. The overall vaccine coverage rates from July 1984 to December 2002 were 96.6%, 95.2%, and 92.8% for the first, second, and third doses, respectively.
In the National Hepatitis B Immunization Registry, all the information of hepatitis B seromarkers of pregnant women and immunization records of newborns has been mandatorily registered, updated, and regularly scrutinized by the Taiwan Center for Disease Control (Taipei, Taiwan). The profiles were considered very complete and accurate.
Study Population and Liver Disease Ascertainment
The profile of the National Household Registry, which contained national identification numbers and sociodemographic characteristics of all household members, including live births, was used to link the data of maternal hepatitis B seromarkers and newborn immunization records in the computerized profiles of the National Hepatitis B Vaccination Registry from July 1, 1984 to March 31, 2000 in Taiwan. Maternal serostatus of HBsAg and HBeAg were determined by prenatal care providers and reported to the National Hepatitis B Immunization Registry. Prenatal hepatitis B seromarkers were tested by radioimmunoassay (RIA), enzyme immunoassay (EIA), or reverse passive hemagglutination (HA) assay before births of their children. For the first several years after initiation of the national vaccination program, the reverse passive HA assay was used as an alternate testing method if the standard RIA or EIA was unavailable at some laboratories. Previous experience with use of the reverse passive HA assay as a substitute for RIA or EIA was satisfactory.[11, 28]
Figure 1 shows a flow diagram of vaccinees included in the analysis of this study. There were 11,685 vaccinees with unknown maternal hepatitis B seromarkers during pregnancy or immunization status; this analysis included a cohort of 3,836,988 newborns with complete information on maternal hepatitis B seromarkers and immunization status with HBIG and vaccines. Frequency distribution of maternal hepatitis B serostatus was 82.2% for seronegative of both HBsAg and HBeAg, 11.4% for HBsAg-seropositive and HBeAg-seronegative, and 6.4% for seropositive of both antigens. There were 75.3% of vaccinees whose mothers were seropositive for HBsAg and HBeAg received HBIG and 80.3% vaccinees received complete doses of hepatitis B vaccines.
Newly diagnosed cases of hepatocellular carcinoma (International Classification of Diseases [ICD]-O code 155.0 with histology code 81703) were ascertained by computerized linkage with the National Cancer Registry, whereas deaths from fulminant hepatic failure (ICD-9 code 570) and CLDs (ICD-9 code 571) were ascertained by computerized linkage with the National Death Certification Registry. During the follow-up of 41,854,715 person-years, there were 49 newly diagnosed HCC cases, 73 deaths of FHF, and 74 deaths of CLD in this study.
Although there were no registered records for the unimmunized newborns (3.4%), their incidence of HCC and mortality of FHF and CLDs were estimated from incidence and mortality rates of the registered immunized newborns and the entire population born after July 1984 derived from national death certification and cancer registry databases, taking vaccination coverage (96.6%) into consideration.
Duration of follow-up for each cohort member was calculated from the date of birth to the date of the diagnosis of newly developed HCC, date of the death from fulminant hepatic failure, CLD, or other causes, or the last date when the national cancer registry and death certification registry data were available, whichever came first. Cumulative incidence and mortality by age was derived using the Nelson-Aalen method. Cox's proportional hazards regression analysis was used to estimate the hazard ratios (HRs) with corresponding 95% confidence intervals (CIs) for risk predictors of HCC, FHF, and CLD. All statistical tests were two-tailed.
Maternal serostatus of HBsAg and HBeAg were dichotomized into binary variables (seropositive or seronegative). In addition to the seropositivity of HBeAg tested by RIA or EIA, a serotiter of HBsAg of 1:2,560 or above was also considered as the seropositivity for HBeAg. Based on doses of vaccines, for cohort members born from July 1984 to October 1992, vaccination with four doses of plasma-derived vaccine was considered to be complete; complete vaccination status was defined as vaccination with three doses of recombinant vaccines for those born from November 1992 to March 2000. Newborns of highly infectious carrier mothers received an additional 0.5 mL of HBIG, plus complete vaccination status was defined as complete immunization status.
Risk of Liver Diseases in Immunized and Unimmunized Newborns
There were 49 newly diagnosed HCC cases, 73 deaths of FHF, and 74 deaths of CLD during the follow-up of 3,836,988 immunized newborns for 41,854,715 person-years, showing an incidence rate of 0.117 per 100,000 person-years for HCC and a mortality rate of 0.174 and 0.177 per 100,000 person-years, respectively, for FHF and CLDs. Derived from the national death certification and cancer registry databases, HCC incidence, FHF mortality, and CLDs mortality for the entire population born after the launch of the immunization program were 0.123, 0.193, and 0.246 per 100,000 person-years. Taking the hepatitis B immunization coverage of 96.6% into account, HCC incidence, FHF mortality, and CLDs mortality were as high as 0.293, 0.733, and 2.206 per 100,000 person-years, respectively, indicating that hepatitis B immunization had a very high efficacy to prevent HCC (60.1%), FHF (76.3%), and CLDs (92.0%).
Among HCC, 41% of cases were diagnosed at ages 5-10 years and 49% of cases were diagnosed at ages 10-20 years. Most vaccinees did not reach the ages of 15-20 until the last date when the national cancer registry and death certification registry data were available for data analysis in this study, so there were few cases in this age group. Most FHF and CLD deaths in this vaccinated cohort occurred in those younger than 5 years (85% and 73%, respectively).
Figure 2 shows the cumulative incidence of HCC and mortality of FHF and CLDs by maternal hepatitis B serostatus and immunization status of 3.8 million vaccinees. Cumulative incidence per 100,000 of HCC was 0.79 for vaccinees whose mothers were seropositive for HBsAg and HBeAg, 0.16 for vaccinees whose mothers were seropositive for HBsAg only, and 0.03 for vaccinees whose mothers who were seronegative for both antigens (Fig. 2A). Cumulative mortality per 100,000 of FHF was 0.30 for vaccinees whose mothers were seropositive for HBsAg and 0.12 for vaccinees whose mothers who were seronegative for the antigen (Fig. 2B). Cumulative mortality from CLDs was similar for vaccinees born to mothers seropositive and seronegative for HBsAg (Fig. 2C). Incomplete immunization was significantly associated with cumulative incidence of HCC in vaccinees whose mothers were seropositive for HBsAg (Fig. 2D) and with cumulative mortality of FHF (Fig. 2E) and CLDs (Fig. 2F).
Risk Predictors for HCC
Table 1 shows the incidence of HCC by gender and maternal hepatitis B seromarker of vaccinees. Male vaccinees had a significant higher incidence of HCC than female vaccinees (maternal hepatitis B seromarker-adjusted HR:, 2.53; 95% CI: 1.34-4.77; P = 0.0042). The gender-adjusted HR (95% CI; P value) of developing HCC was 3.90 (1.68-9.04; P = 0.0015) for vaccinees whose mothers were seropositive for HBsAg only and 17.58 (9.41-32.84; P < 0.0001) for vaccinees who were seropositive for both HBsAg and HBeAg, in comparison with vaccinees whose mothers were seronegative for both seromarkers as the referent group.
Table 1. Incidence and Adjusted HR of HCC by Gender and Maternal Hepatitis B Seromarkers of Vaccinees
Gender-adjusted or maternal hepatitis B seromarker-adjusted HR.
2.53 (1.34-4.77) P = 0.0042
Maternal status of HBsAg/HBeAg
3.90 (1.68-9.04) P = 0.0015
17.58 (9.41-32.84) P < 0.0001
The gender-adjusted HR was substantially higher among vaccinees whose mothers were seropositive for HBsAg and HBeAg (with HBIG at birth, HR = 13.03 [6.30-26.92; P < 0.0001], and without HBIG, HR = 29.95 [13.99-64.12; P < 0.0001]) and moderately higher among vaccinees whose mothers were seropositive for HBsAg only (with hepatitis B vaccines completed, HR = 2.41 [0.81-7.16; P = 0.114], and with vaccines incompleted, HR = 10.33 [3.47-30.73; P < 0.0001]), in comparison with vaccinees whose mothers were seronegative for both seromarkers as the referent group.
Because the incidence of HCC was too low to assess its difference between completely and incompletely immunized vaccinees born to mothers seronegative for hepatitis B markers, Table 2 shows gender-adjusted HRs of HCC by maternal serostatus of HBeAg and immunization status of 679,056 vaccinees whose mothers were seropositive for HBsAg. The gender-adjusted HRs increased with maternal serostatus of HBeAg and incomplete immunization status. The gender-adjusted HR (95% CI; P value) was 12.71 (5.60-28.81; P < 0.0001) for HBIG-unimmunized newborns of mothers seropositive for HBeAg, compared with completely immunized newborns whose mother were seronegative for HBeAg. Incomplete vaccination, either HBIG or vaccines, had an independent effect on the incidence of HCC, showing an HR of 2.52 (95% CI; 1.25-5.05; P = 0.0094) after adjustment for gender and maternal HBeAg serostatus.
Table 2. Gender-Adjusted HR of HCC by Maternal Serostatus of HBeAg and Immunization Status of Vaccinees Whose Mothers Were Seropositive for HBsAg
Maternal Serostatus of HBeAg
Person-Years of Follow-up
Number of HCC Cases
Incidence Rate (10−5)
Gender-Adjusted HR (95% CI)
4.40 (1.42-13.65) P = 0.0103
5.51 (2.51-12.08) P < 0.0001
12.71 (5.60-28.81) P < 0.0001
Risk Predictors for FHF
Table 3 shows mortality and HR of FHF by maternal serostatus of HBsAg and immunization status of vaccinees. HRs increased with maternal serostatus of HBsAg and incomplete immunization status. The HR (95% CI; p value) was 11.23 (6.23-20.24; P < 0.0001) for incompletely immunized newborns of mothers seropositive for HBsAg, compared with completely immunized newborns whose mother were seronegative for HBsAg. Incomplete vaccination had an independent effect on the mortality of FHF, showing an HR of 4.97 (3.05-8.11; P < 0.0001) after adjustment for maternal HBsAg serostatus.
Table 3. Mortality and HR of FHF by Maternal Serostatus of HBsAg and Immunization Status of Vaccinees
Maternal Serostatus of HBsAg
Number of Vaccinees
Person-Years of Follow-up
Number of FHF Cases
Mortality Rate (10−5)
HR (95% CI)
3.47 (1.96-6.14) P < 0.0001
1.25 (0.52-3.01) P = 0.6183
11.23 (6.23-20.24) P < 0.0001
Risk Predictors for CLDs
Table 4 shows mortality and HR of CLDs by maternal serostatus of HBsAg and immunization status of vaccinees. HRs increased with maternal serostatus of HBsAg and incomplete immunization status. The HR (95% CI; P value) of incomplete immunization, compared with complete immunization, was 3.78 (2.30-6.20; P < 0.0001) and 3.00 (1.27-7.11; P = 0.013) for newborns whose mothers were seronegative and seropositive for HBsAg, respectively. Incomplete vaccination was associated with an increased mortality of CLDs in vaccinees, showing an HR of 6.27 (3.62-10.84; P < 0.0001) after adjustment for maternal HBsAg serostatus.
Table 4. Mortality and HR of CLDs by Maternal Serostatus of HBsAg and Immunization Status of Vaccinees
Maternal Serostatus of HBsAg
Number of Vaccinees
Person-Years of Follow-up
Number of CLD Cases
Mortality Rate (10−5)
HR (95% CI)
3.78 (2.30-6.20) P < 0.0001
0.32 (0.08-1.34) P = 0.120
3.00 (1.27-7.11) P = 0.013
In this prospective cohort study of 3.8 million vaccinees in Taiwan, we found that the incidence of HCC and mortality of FHF and CLDs had not been eradicated after implementation of the national hepatitis B immunization program. The large newborn population with detailed information of maternal hepatitis B seromarkers and immunization records in this study allowed us to assess the long-term risk and predictors of end-stage liver diseases of immunized newborns from infancy to early adulthood. There were striking differences between unvaccinated and vaccinated newborns after the launch of national immunization program for incidence of HCC (0.293 vs. 0.117 per 100,000 person-years), mortality of FHF (0.733 vs. 0.174 per 100,000 person-years), and mortality of CLDs (2.206 vs. 0.177 per 100,000 person-years). The findings demonstrated the high efficacy of hepatitis B immunization to prevent these liver diseases (60.1%-92.0%).
Incomplete immunization with HBIG or vaccines was the most important risk predictor of the liver disease among vaccinees. The multivariate-adjusted HR of developing HCC, FHF, and CLDs was 2.52, 4.97, and 6.27, respectively, suggesting that approximately 60% of HCC, 80% of FHF, and 84% of CLDs in incompletely immunized vaccinees would have been prevented if they were immunized completely.
Male gender and maternal serostatus of HBsAg and HBeAg were significantly associated with risk of HCC in this study, which were consistent with findings of our previous ecological study. Administration of HBIG was important for newborns whose mother were seropositive for HBeAg in this study. It might control mother-to-infant transmission of HBV and reduce the rate of chronic HBV infection. Maternal serostatus of HBsAg was another important risk predictor for FHF, but not for CLDs, among vaccinees in this study. Previous studies have shown that maternal serostatus of HBeAg, delayed immunoprophylaxis, escape mutant of HBVs, and in utero infection factors may be attributable to immunization failure of newborns whose mothers were seropositive for HBaAg.[29-32] Moreover, HBV might also be transmitted in early childhood through horizontal routes, which had decreased drastically after implementation of the national immunization program.[10, 33]
This large-scale, population-based cohort study has the advantage over previous ecological studies on evaluation of the efficacy of the hepatitis B immunization program. Because individualized data were obtained from several complete and accurate national registries, there was no bias on immunization status of newborns, prenatal serostatus of HBsAg and HBeAg, as well as ascertainment of incidence of HCC and deaths of FHF and CLDs. However, there are some limitations to our study. This study did not have the individual responses to immunization; it was thus difficult to assess the effect of vaccine failure directly. There might be other etiological agents other than HBV to cause liver diseases of this vaccinee cohort. But, the high HRs for incomplete immunization seem to suggest that a high proportion of these liver diseases may be attributable to HBV through either perinatal or horizontal transmission. The first-dose coverage rate was 93.9%, 95.9%, and 93.2% for incident cases of HCC and deaths of FHF and CLDs, respectively. As a result of the relatively small number of these clinical outcomes, we were limited to analyze the effect of the timing of each dose of vaccination on development of HCC, FHF, and CLDs.
In conclusion, prenatal maternal hepatitis B serostatus and incomplete immunization of HBIG and vaccines after birth were important determinants for HCC development, FHF death, and CLD death among the vaccinated cohort. To achieve optimal prevention of HBV-related liver diseases, administration of HBIG and vaccines timely and completely is essential.