CEA China
CEA China
CEA China
Yuanxi Jia, Li Li, Fuqiang Cui, Dongliang Zhang, Guomin Zhang, Fuzhen Wang,
Xiaohong Gong, Hui Zheng, Zhenhua Wu, Ning Miao, Xiaojin Sun, Li Zhang,
Jingjing Lv & Feng Yang
To cite this article: Yuanxi Jia, Li Li, Fuqiang Cui, Dongliang Zhang, Guomin Zhang, Fuzhen
Wang, Xiaohong Gong, Hui Zheng, Zhenhua Wu, Ning Miao, Xiaojin Sun, Li Zhang, Jingjing Lv &
Feng Yang (2014) Cost-effectiveness analysis of a hepatitis B vaccination catch-up program among
children in Shandong Province, China, Human Vaccines & Immunotherapeutics, 10:10, 2983-2991,
DOI: 10.4161/hv.29944
Keywords: catch-up program, Cost-effectiveness Analysis, hepatitis B virus, ICER, QALY, vaccination
Abbreviations: HBV, Hepatitis B Virus; HBsAg, Hepatitis B Surface Antigen; GAVI, Global Alliance on Vaccines and Immuniza-
tion; HepB3, 3-dose Coverage of Hepatitis B Vaccine; MOH, Ministry of Health; NNDRS, National Notifiable Diseases Reporting
System; CHB, Chronic Hepatitis B; CC, Compensated Cirrhosis; DC, Decompensated Cirrhosis; HCC, Hepatocellular Carcinoma;
LT-1, the Year of Liver Transplantation; LT-2, Years after Liver Transplantation; QALYs, Quality-Adjusted Life Years; HRQoL,
Health-Related Quality of Life; Anti-HBs, Antibody to Hepatitis B Surface Antigen; ICER, Incremental Cost-Effectiveness Ratio;
GDP, Gross Domestic Product; BCR, Benefit-Cost Ratio
Objective: The aim of the study was to estimate long-term cost‑effectiveness of a hepatitis B vaccination catch-up
program among children born between 1994 and 2001 (when they were 8‑15 y old) in Shandong province, China, to
provide information for nationwide evaluation and future policy making.
Methods: We determined the cost-effectiveness of the catch-up program compared with the status quo (no catch-up
program). We combined a Decision Tree model and a Markov model to simulate vaccination and clinical progression
after hepatitis B virus (HBV) infection. Parameters in the models were from the literature, a field survey, program files,
and the National Notifiable Disease Reporting System (NNDRS). The incremental cost‑effectiveness ratio (ICER) was used
to compare the 2 alternative strategies. One-way sensitivity analysis, 2-way sensitivity analysis, and probability
sensitivity analysis were used to assess parameter uncertainties.
Results: The catch-up program was dominant compared with the status quo. Using a total of 5.53 million doses of
vaccines, the catch-up program could prevent 21,865 cases of symptomatic acute hepatitis B, 3,088 carrier states with
positive hepatitis B surface antigen (HBsAg), and 812 deaths due to HBV infection. The catch-up program could add
28,888 quality-adjusted life years (QALYs) and save $192.01 million in the targeted population in the future. The models
were robust, considering parameter uncertainties.
Conclusion: The catch-up program in Shandong province among children born between 1994 and 2001 was ‘very
cost-saving.’ It could save life years and reduce total future costs. Our study supported the desirability and impact of
such a catch-up program throughout China.
China has the greatest burden of hepatitis B virus (HBV) dis- (HepB3) exceeded 90% after 2003, compared with less than
ease and liver cancer in the world.1 Data from the 2006 national 60% before 1998 and 20% in 1994. These data indicated
hepatitis B sero-epidemiologic survey showed that about 7.18% that a large proportion of children born between 1994 and
(93 million people) of China’s population carried hepatitis B sur- 2001 were not vaccinated with hepatitis B vaccine and
face antigen (HBsAg).2 remained at risk of HBV infection.
China added hepatitis B vaccine into its national immuni- In order to increase HepB3 coverage among children born
zation program in 2002, allowing the vaccine to be provided between 1994 and 2001 (then aged 8‑15 years), the Ministry of
free to all newborns, nationwide. With support of the Global Health (MOH) initiated a nationwide hepatitis B vaccine catch-
Alliance on Vaccines and Immunization (GAVI), hepatitis B up program for these children that lasted from 2009 to 2011.
vaccination in China’s western region and economically During the catch-up program, approximately 68 million children
weaker areas increased rapidly.3,4 According to China’s Sur- who had not been fully vaccinated were caught up on the doses of
veillance System of Information in the National Immuniza- hepatitis B vaccine that they needed, based on their vaccination
tion Program, reported 3-dose coverage of hepatitis B vaccine history.
All costs were adjusted with respect to the cost of US dollar in August, 2013.16 We first confirmed the applicability of EQ-5D-
2013 according to the consumer price index and the exchange 5L in the targeted population. HRQoL of 5 HBV infectious con-
rate.14,15 ditions, including carriers, CHB, CC, DC, and HCC, were
Effectiveness Parameters: Effectiveness was quantified by obtained from a field survey. HRQoL of patients with liver trans-
QALY in the form of HRQoL. We performed a field survey plantation were obtained from the published literature. Since dif-
using EQ-5D-5L, the 5-level version of EQ-5D to determine the ferent general instruments using the same population may result
HRQoL of the targeted population in Shandong province in in different outcomes and lead to opposite conclusions,17 only
CC: Compensated cirrhosis; CHB: Chronic hepatitis B; DC: Decompensated cirrhosis; HBs: Hepatitis B surface; HBV: Hepatitis B virus; HCC: Hepatocellular carci-
noma; HRQoL: Health-Related Quality of Life; LT-1: The year of liver transplantation; LT-2: Years after liver transplantation; NNDRS: National Notifiable Disease
Reporting System.
studies using EQ-5D as the survey instrument were included in The proportion of HBV susceptible persons becoming
our model. infected each year was estimated using data from NNDRS.
Since there was no study of HRQoL of acute hepatitis B NNDRS data showed that the incidence of acute hepatitis B in
patients using EQ-5D, we estimated this parameter by compar- Shandong province was 3.58/100,000 in 2008. We assumed that
ing 2 instruments. According to a model used in Japan, the value 30% of the patients with acute hepatitis B would have clinical
by EQ-5D ranged from ¡0.106 to 0.848,18 while for SF-6D, symptoms; 31.51% of the total population was susceptible for
the value ranged from 0.30 to 1.00.19 Therefore, HRQoL for HBV infection; and that 50% of patients were hospitalized for
acute hepatitis B patients was estimated as 0.766.20 treatment after onset of symptoms. We therefore assumed that
Other Parameters: Information on the targeted cohorts was the proportion of HBV susceptible persons becoming infected
extracted from program files, including the number of children each year could be estimated as 75.6/100,000. Since the estimate
in each cohort and the number of doses received previously in of this proportion has varied considerably in studies,7,9 we used a
each group (Table 2). wide interval from 20/100,000 to 500/100,000 in the sensitivity
Since there were no serological tests available, and since it was analysis.
not possible to trace targeted children by the information col-
lected during the program, the proportion of HBV susceptible Model analysis
persons in each cohort was estimated based on a sero-epidemio- We used the Incremental Cost-Effectiveness Ratio (ICER) to
logic survey by age groups,21 using a formula given below: compare the 2 alternative strategies. The ICER was determined by
calculating the difference in costs of the 2 strategies divided by the
Ps D .1 ¡ P1 ¡ P2 / £ .1 ¡ P3 / difference in health effects. According to WHO, an intervention
can be considered ’very cost-effective’ if the ICER is lower than
the annual per capita Gross Domestic Product (GDP) or 3 times
Ps represents the proportion of HBV susceptible persons in each the annual per capita GDP.47 These values were $5,414 and
group and P1 and P2 represent the HBV prevalence proportion $16,242,14,15 respectively. We used one-way sensitivity, 2-way sen-
and natural immunization proportion in each group, respec- sitivity, and probabilistic sensitivity analysis based on Monte-Carlo
tively. P3 represents the antibody to hepatitis B surface antigen simulation with 10,000 iterations to evaluate the impact of param-
(Anti-HBs) positive conversion rate after different doses of eter uncertainty on the ICER. Since the distributions of parameters
vaccination. were unclear due to limited literature-based estimates, all parame-
ters were assigned triangular distributions.7
Table 2. Information on the Targeted Cohorts
*All costs were adjusted with respect to the cost of dollar in 2013.
the costs of the catch-up strategy were composed of the imple- probabilistic sensitivity analysis. All the points are in the fourth
mentation costs ($7.89 million) and the reduced treatment costs quadrant, indicating that all the samples derived supported our
($8.38 million). The saved treatment costs of $199.88 million conclusion that the catch-up program was always dominant with
were greater than the implementation costs of the vaccination reduced costs and increased QALYs. Taking all parameter uncer-
program ($7.89 million). tainties into account, the model remained robust and reliable.
The catch-up program dominated in all the cohorts,indicat-
ing that the program was not sensitive to the age parameter of
the targeted population (Table 3). By using a total of 5.53 million Discussion
doses of vaccines, the catch-up program could prevent 21,865
cases with symptomatic acute hepatitis B, 3,088 carriers with pos- In recent years, China has made great efforts to control and
itive HBsAg, and 812 deaths due to HBV infection compared prevent HBV infection. But before the extensive use of vaccines,
with the status quo (Table 4). The catch-up program could many newborns were not vaccinated and remained susceptible to
increase 28,888 QALYs and reduce $192.01 million in the HBV infection. According to WHO, implementation of catch-
future. up immunization in such populations will produce broad immu-
Sensitivity analysis: According to the one-way sensitivity anal- nity to HBV infection and eventually prevent transmission
ysis (Table 5), none of the single parameters could change the among all age groups.10
conclusion that the program was ’cost-saving.’ The model was A review showed that in regions with intermediate or high lev-
robust to the uncertainty of any single parameter. Parameters els of HBV epidemic, universal vaccination of newborns would
that showed the greatest impact on the ICER were the probabil- be cost-effective,5 and this conclusion extends to Taiwan,8 Gam-
ity of HBV carriers becoming CHB, the proportion of HBV sus- bia,9 and India.48 Many studies using cost‑effectiveness analysis
ceptible persons becoming HBV infectious each year, and the or cost-benefit analysis in small populations showed that vaccina-
discount rate. We conducted 2-way sensitivity analyses using tion of newborns,49-52 and of some specific group (soldiers,53,54
these parameters, and no combination could change the conclu- workers,55 or middle school students,56 or children under 15 y of
sion of being ‘cost-saving.’ Figure 2 shows the results of the age57) in mainland China, can also be economically acceptable.
Birth Status Catch-up Infections Status Catch-up Carriers Status Catch-up Deaths
Year Quo Vaccination Prevented Quo Vaccination Prevented Quo Vaccination Prevented
In our study, we examined a nationwide catch-up program using capita GDP has increased as well. It can be inferred that in
original data collection to improve the accuracy and reliability of coming years such catch-up programs may be necessary and
parameters and obtain credible results. Although there were dif- affordable.
ferences in benefit-cost ratio (BCR) and ICER between studies, In previous research on this catch-up strategy, the annual inci-
the conclusion always remained that universal vaccination of dence of acute HBV infection was the only parameter could
newborns and catch-up programs for specific groups was a cost- reverse the conclusion.6 Additionally, the discount rate and the
saving strategy. transition rate from carrier to CHB were among the critical
Our study showed that the catch-up program not only parameters35 that we found in our study. According to different
increased the QALYs of the targeted population, but also populations and vaccination strategies, there are differences in
reduced total costs substantially. Although the incremental the contribution of parameter uncertainties. Since there were not
cost and incremental effectiveness varied in different cohorts, sufficient data on the distribution of parameters, we applied tri-
the program was always ‘cost-saving’ for all the age groups. angular distributions in our models,7 while previous studies used
Since most parameters from the literature had uncertainties other parameter distributions.8,50 Compared with prior work6
of their true values, these uncertainties might influence model arriving at the same conclusion, the accuracy and reliability of
results. We used sensitivity analysis to explore the direction parameters have been improved by our study. It was the first
and degree of effects from each and all parameters. The time in mainland China that EQ-5D-5L was used as the instru-
results showed that even within a wide range, no single ment of HRQoL, and that data from program files was used
parameter could reverse the conclusion of program of being directly as a resource in a cost‑effectiveness analysis. We received
‘cost-saving.’ The models were always robust, considering the data and support from NNDRS managed by Chinese Center for
combination of all uncertainties from parameters. Our study Disease Control and Prevention to form parameters in the Mar-
provided evidence to support the economic benefits of the kov model.
catch-up program in Shandong province. Recently, China has We conducted our study in Shandong province, while the
experienced rapid economic progress, and its annual per program was implemented nationally. We believe that the