Economic evaluation on infant hepatitis B vaccination combined with immunoglobulin in China, 2013

Introduction

Hepatitis B virus (HBV) infection was highly endemic before the introduction of hepatitis B vaccine (HepB) in China, with HBV surface antigen (HBsAg) prevalence being 9.75% in the general population in 1992.¹ HBsAg prevalence among children <4 years old (9.67%) was nearly the same as the prevalence among adults, which means transmission during birth and early childhood was the predominant mode of HBV transmission in China.¹ Ninety percent of newborns infected with HBV would become chronic carriers, leading to a substantial disease burden of chronic liver disease.²

China has made substantial progress in reducing HBV mother-to-child transmission (MTCT), but challenges remain. In 1992, HepB was first recommended by the Ministry of Health (MOH) for routine vaccination of infants, with the first dose to be administered within 24 hours of birth and subsequent doses at ages 1 and 6 months. In 2002, China integrated HepB into the Expanded Progamme on Immunization (EPI), with the government providing free 5μg/0.5ml/dosage HepB for all infants. In 2011, China changed the HepB dosage form to 10μg/0.5ml/dosage to improve vaccine immunogenicity. Joint HBV immune prophylaxis with HepB vaccine and hepatitis B immunoglobulin (HBIG) after birth could generally interrupt HBV MTCT,^3-5 and it is recommended by the World Health Organization (WHO), the World Gastroenterology Organization (WGO), and the US Centers for Disease Control and Prevention (CDC). At the end of 2010, China implemented a program for preventing mother-to-child transmission (PMTCT) of HIV, syphilis, and hepatitis B, where all children born to HBsAg positive mothers could receive free 100IU HBIG at birth.

Economic analysis of universal infant hepatitis B vaccination has been carried out by several studies in China.^6-9 However, previous studies were focused on the economic analysis of 5μg vaccine used in China. Most of the previous studies did not include screening for pregnant women and HBIG injection. Some of these studies also did not include costs for adverse events following immunization (AEFI) and productivity losses due to HBV- related diseases. The objective of this study is to provide an updated and comprehensive economic evaluation of infant HepB vaccination with 10ug vaccine and HBIG. Results from this study will be important and useful for making future policy decisions regarding HepB vaccination.

Results

Base case

With no universal infant HepB vaccination program, our model estimated that about 12.89 million HBV infections would happen for the entire lifespan of the cohort, resulting in 599,518 early deaths from HBV related diseases. These infections would result in a direct cost of $12.09 billion and a societal cost of $46.93 billion.

Compared with no HepB vaccination, the strategy of universal HepB vaccination combined with HBIG for infants of HBsAg positive mothers could prevent 96.91% (12.49 million) of HBV infections, and 96.11% (0.58 million) of early deaths from HBV related diseases in this cohort. The related direct and societal program costs were estimated to be $0.19 billion and $0.23 billion, respectively. The related direct and societal cost of averted illness would be $11.68 billion and $45.14 billion, respectively. The direct and societal net present value (NPV) of the program would be $11.5 billion and $44.9 billion, respectively. The direct and societal benefit-cost ratios (BCRs ) would be 61.3 and 193.2, respectively (Tables 1 and 2).

Impact of HBIG administration

Compared with HepB vaccination without HBIG, the strategy of HepB vaccination combined with HBIG for infants of HBsAg positive mothers could prevent an estimated 3,500 HBV infections. The direct and societal BCRs for the HBIG administration were 0.4 and 2.7, respectively.

Sensitivity analyses

Sensitivity analyses results for the direct and societal BCRs of the current HepB immunization program are shown in Table 3. We found that the model was very stable, with BCRs remaining above1.0 regardless of changes in key parameter values. The costs of health outcomes, discount rate, and administration costs had a relatively greater influence on the model. For the worst-case scenario, we ran the model using lower bound estimates of disease costs largely obtained from a study in a western city, and the direct and societal BCRs were 30.8 and 161.5, respectively, much lower than the base case BCRs. With higher discount rates, the direct BCR decreased significantly from 41.3 to 172.5. If we doubled the administration costs, direct and societal BCRs would decrease to 44.3 and 147.4, respectively. When we doubled the price of vaccine and HBIG, we saw a negligible change in BCRs.

Table 1. Estimated HBV infections and related deaths in universal vaccination combined with HBIG, 2013.

			Prevented
	No vaccination	vaccination combined with HBIG	Number of cases	(%)
Number of HBV infection	12,889,683	397,733	12,491,950	96.91
Number of HBV-related deaths	599,518	23,309	576,209	96.11

Table 2. NB and BCR of universal vaccination combined with HBIG strategy, 2013.

	No vaccination		Vaccination combined with HBIG
	direct	societal	direct	societal
Program cost (billion)	–	–	0.19	0.23
Illness cost (billion)	12.09	46.93	0.42	1.79
Adverted illness cost(billion)	–	–	11.68	45.14
NB (billion)	–	–	11.49	44.91
BCR	–		61.3	193.2

Table 3. The univariate sensitivity analysis.

Parameter	Bound*	direct	societal
Base Case		61.3	193.2
Vaccine prices	lower	62.9	197.4
	upper	54.1	174.4
Wastage rate	lower	61.4	193.6
	upper	61.1	192.8
Administration costs	lower	75.7	228.8
	upper	44.3	147.4
Work loss for vaccination	lower	61.3	212.8
	upper	61.3	176.9
Pregnant woman Screening rate	lower	64.3	200.9
	upper	60.1	190.3
Screening sensitivity and specificity	lower	54.9	176.1
	upper	63.7	199.8
Screening cost	lower	70.5	216.4
	upper	54.2	174.6
Coverage rate	lower	60.2	190.4
	upper	61.3	193.2
Vaccine Efficacy	lower	56.8	179.0
	upper	62.6	197.7
Prenatal infection rate in HBsAg positive mother	lower	60.9	190.8
	upper	61.5	194.6
HBIG injection rate	lower	61.2	193.1
	upper	61.3	193.3
HBIG prices	lower	64.5	201.6
	upper	54.9	176.6
HBV infection rate in pregnant woman	lower	61.4	192.1
	upper	61.1	194.6
Costs of health outcomes	lower	30.8	161.5
	upper	128.1	256.9
Discount rate	lower	142.6	278.6
	upper	41.3	172.5

Discussion

Our results showed that, compared with no HepB immunization program, universal infant HepB vaccination combined with one dose of HBIG would prevent more than 12 million new HBV infections, and 576,209 cases of HBV related deaths. This strategy could result in substantial direct and societal cost savings ($11.68 billion and$ 45.14 billion, respectively) and high BCRs (61.3 and 193.2, respectively) in each vaccinated cohort. Sensitivity analyses indicated our model is very stable with BCRs remaining above 1.0 regardless of changes in parameter values. From the societal perspective, the administration of HBIG was a cost-saving strategy.

These results were consistent with and supported by results from earlier studies, although not directly comparable because of different assumptions and methodology used in the methods. HepB vaccination has proven to be a significant cost saving intervention in China,⁸ with BCRs from 26-216.

China implemented 5μg HepB vaccination in infants from 1992; however, the anti-HBs produced by 5μg HepB were lower compared with 10μg HepB. Some epidemiologists^10,11 suggested that China shouldincrease the HepB vaccine dosage for infants. Price and output were the main challenges for the integration of 10μg HepB into China EPI. With the increase in domestic vaccine production and decrease in prices in recent years, China successfully added 10μg HepB in EPI in 2011. Our results demonstrated the substantial health benefits associated with HepB vaccination, as well as a great return on investment for the vaccine.

Our study had both advantages and limitations. One of its major advantages was that the important parameter values in our model, such as HepB₃ coverage, timely birth dose, and HBIG injection rate, were obtained from the national surveillance system or the official report of PMTCT, which are more reliable and comprehensive than sources used in previous studies. Second, the HBV epidemic parameters in pregnant women were critical for this model, and we obtained the HBsAg prevalence of childbearing women in China (including the e-antigen positive rate) from the most current national serosurvey data in 2006, which fully considered the data differences among eastern, central and western areas, and both urban and rural. Third, although they might be too small to affect the results, we included the costs of AEFI events in our models.

There were several limitations in this study. In China, visits from relatives and friends to patients in informal care are very common, which would incur additional work time loss costs. We did not include them in our analyses, which might be an understatement of the costs of illness that would therefore yield lower NPVs and BCRs. China has substantial geographic differences across regions, with different levels of social and economic development across eastern, central and western areas. It was very difficult to assume that some parameters were nationally representative, such as the costs of health outcomes. Although we tried to reference previous studies carried out in different areas, it was still very difficult to get reliable and representative estimates. Antiviral drugs are very effective in treating chronic HBV infections. However, due to specific economic and medical system issues, these drugs have not been widely used for Chinese patients and thus were not included in our model. Another important limitation of the study is that in the Markov model, it was assumed that the transition probabilities from one health state to another were not age-dependent. Further research is needed to improve the model.

In summary, the current infant HepB immunization program in China was effective in preventing new HBV infection, and it could reduce HBV related morbidity and mortality while also reducing health care costs. These results supported China's health policy makers in continuing central and regional funding for vaccination of high risk newborns combined with HBIG.

Method

Model

The decision tree upon which our model is based appears in Figure 1. We evaluated the base case scenario: the effect of the infant HepB vaccination combined with HBIG for infants of HBsAg-positive mothers on disease and economic outcomes over the lifetime of a birth cohort and compared with outcome with no vaccination policy. We also evaluated the impact of HBIG. The assumptions in our model included: (1) all the vaccinated infants completed the entire 3-dose series of HepB; (2) vaccination results in lifelong protection; (3) there is no vaccine-induced herd-immunity. We used Microsoft Excel 2010 to construct the model and perform the analyses.

edical costs) and societal cost (direct cost and indirect cost).¹² Direct medical costs included those associated with treating acute infections, as well as costs associated with HBV related diseases. Direct non-medical costs included travel costs, costs for special nourishment, and patient care. Societal costs included the productivity losses due to HBV infection and premature death. Benefits of immunization were quantified as the savings in direct and societal costs that accrue from averting morbidity and mortality by vaccination. The costs associated with the infant HepB immunization program included the vaccines and HBIG, their administration, and parents' work time lost associated with vaccination. All costs were adjusted to 2013 US dollars using general Consumer Price Indices (CPI)¹³ and Medical CPI,¹⁴ and future costs and benefits were discounted at a 3% annual rate. We calculated the NPV and BCRs. NPV is the discounted benefit from the immunization program minus the discounted immunization program cost, and BCR is equal to the discounted benefit divided by the discounted immunization program cost. If the BCR≥1.0, we consider the immunization program to be cost saving.

Figure 1. Decision-tree for the model.

The data for burden of diseases, costs of diseases, and costs of vaccination used in our analysis were compiled from a variety of sources: the published literature, including surveillance data, sero-survey data, study data, and expert consensus; several large computerized data sets; and Chinese Center for Disease Control and Prevention (China CDC) unpublished data.

Estimating the burden of HBV infected without immunization program

The target population was the 2013 birth cohort of 16,437,497 infants.¹⁵ We divided them into 3 subgroups by mother's HBV infection status: (1) both HBsAg and HBV e antigen (HBeAg) were positive; (2) HBsAg was positive, but HBeAg was negative; (3) HBsAg was negative. In each subgroup the HBV infection risks for newborns were different. We used the age-specific HBV infection rate for susceptible populations to determine the number infected without immunization,¹⁶ and tracked the hepatitis B-related events for the entire lifespan of all the infants.

HBV infection was assumed to occur in one of 3 age periods: perinatal (at birth); early childhood (after birth through 5 years), and late (>5 years). Symptoms occur in 1% of perinatal infections, 10% of early childhood infections, and 30% of late infections.² We assumed all the symptomatic HBV infected patients would be hospitalized in China. Among symptomatic patients, risk of fulminant hepatitis was 0.1% and 0.6% for perinatal infection and other cases, respectively.² Mortality from fulminant hepatitis B was assumed to be 70%.^17,18 Among the fulminant patients, the liver transplantation rate would be 0.1%.¹⁹ In China, the successful liver transplantation rate would be 92% (Table 4).

Table 4. Point and interval estimation of parameters in the model.

		Interval Estimation
Parameters	Point Estimation	low	up	Ref.
Newborns in 2013	16,437,497			^27
Vaccination related data
10μg vaccine prices ($/dose)	0.5	0.4	1	^28
HBIG prices ($/dose)	12.92	7.27	25.83	a
Wastage rate of vaccine (%)	2	0	4	^29,30
Administration costs ($/dose)	1.47	0.74	2.11	^31
Work loss for vaccination ($/h)	1.05	0.52	1.57	^32
Pregnant woman screening rate (%)	94.20	80.00	100.00	^33
Screening
cost ( $/person)	2.42	0.81	4.04	^34
Screening reagent sensitivity(%)	96	91	99	^12,35,36
Screening reagent specificity(%)	95	83	99	^12,35,36
Coverage(%)
HepB3 coverage	99.60	95.00	100.00	b
TBD for all newborns	98.00	95.00	99.99	b
TBD for screening HBsAg (+) mothers' newborns	95.88	85.00	99.99	b
HBIG injection rate (%)	91.20	80.00	99.00	report
Efficacy (%)
3 dose for general newborns	97.0	90.0	99.0	^7,37
TBD+HBIG: HBsAg+HBeAg-	94.5	85.1	98.9	^38-41
TBD+HBIG: HBsAg +HBeAg+	91.0	87.0	95.0	^7,41-43
TBD: HBsAg+HBeAg-	93.3	84.0	99.1	^7,39,41
TBD: HBsAg +HBeAg+	83.1	75.0	88.0	^41-43
Duration of protection	lifelong			^2,46,47
AEFI incidence(1/1,000,000 doses)
Thrombocytopenic Purpura	0.03			^c
Angioedema	0.02			^c
AEFI treatment cost ( $/person)
Thrombocytopenic Purpura	617.63			^25
angioedema	513.71			^26
Epidemiology data
HBsAg positive rate among pregnant women	6.61%	5%	7%	^48
HBeAg positive rate among HBsAg(+) women	23.51%	20%	50%	^48
Prenatal infection in HBeAg (+ )mothers	87.5%	70%	90%	^49-52
Prenatal infection in HBeAg (−) mothers	25%	20%	30%	^49-51
Hepatitis B onset proportion after 20 y of neonatal infection	25%			^21
Symptomatic
Prenatal (neonatal)	1%			^2
Early infection(1-5)	10%			^2
Late infection(≥6)	30%			^2
Hospitalization after symptomatic	100%
Fulminant cases among symptomatic
Prenatal (neonatal)	0.10%			^2
Early infection(1-5)	0.60%			^2
Late infection(≥6)	0.60%			^2
Liver transplantation among fulminant	0.1%			^19
Successful transplantation	92%	–	–	^53
Death among fulminant	70%	63%	93%	^17,18
Proportion developing chronic hepatitis
Prenatal (neonatal)	90%	84%	93%	^2,54,55
Early infection(1-5)	30%	23%	40%	^2,54,55
Late infection(≥6)	5%	2.7%	10%	^2,54,55
HBV infected rate in by age-susceptible people	0.13074116-0.01362531×age+0.0004646×age^2-0.00000489×age^3	^16
Transition probabilities (>20 years old)
Natural Clearance of HBV Carriers	1.8%	1%	3%	^56,57
Carrier to CHB:
First year	0.073	–	–	^19
Subsequent years	0.029	–	–	^19
Carrier to CC	0.005			^58
Carrier to HCC	0.0006			^58
CHB to Carrier
First year	0.17324	0.115	0.243	^19,59
Subsequent years	0.105	0.060	0.163	^19,59
CHB to CC	0.129	0.004	0.153	^19,59
CHB to HCC	0.005	0.002	0.007	^19,59
CC to DC	0.054	0.035	0.1	^19,59,60
CC to HCC	0.024	0.01	0.12	^19,59,60
Mortality of CC	0.037	0.03	0.045	^19
DC to HCC	0.024	0.002	0.081	^19
Liver transplantation among DC	0.002
Mortality of DC	0.39	0.3	0.5	^19,59
Liver transplantation among HCC	0.001	–	–
Mortality of HCC	0.56	0.3	0.7	^19,59
Mortality of Liver transplantation	0.06	0.02	0.12	^60,61
Direct Medical costs ( $/year in 2013)
Inactive carrier	105	–	–	^62
Hospitalization for non-fulminant (acute infection)	2,813	1,749	5,853	^62-69
Hospitalization for fulminant	11,066	2,932	13,051	^65-70
CHB	2,830	774	4,086	^62-72
CC	4,012	1,961	7,996	^64,65,67,68,70-72
DC	5,818	2,534	8,875	^64-68,70-72
HCC	6,613	1,772	11,664	^62-72
Liver transplantation (1^st year)	60,189	20,774	63,997	^69,73,74
Liver transplantation (subsequent year)	10,298	–	–	^75
Direct non-Medical costs ($/year in 2013)
Inactive carrier	24	–	–	^62
Hospitalization for non-fulminant (acute infection)	672	118	594	^62,63,65-70
Hospitalization for fulminant	736	326	849	^65-70
CHB	364	80	1,943	^62,64-72
CC	464	204	2,277	^64-68,70-72
DC	640	181	2,589	^64,65,67,68,70-72
HCC	825	222	1,774	^62,64-72
Liver transplantation	680	–	–	^69
Indirect costs ($/year in 2013)
Inactive carrier	83	–	–	^7
Hospitalization for non-fulminant (acute infection)	206	219	753	^66,67,70
Hospitalization for fulminant	1,039	411	1,987	^66,67,70
CHB	515	122	1,153	^64,66,67,70,71
CC	1,240	238	2,208	^64,67,70,71
DC	1,578	352	2,688	^64,67,70,71
HCC	1,787	232	2,636	^64,67,70,71
Liver transplantation (1^st year)	1,431	–	–	^76
Liver transplantation (subsequent year)	83	–	–	^76
Others
Discount rate (%)	3	0	5
Exchange rate in 2013 RMB/USD	6.19			^77

^a. procurement prices by government in 2013

^b. reported coverage data from China Information Management System for Immunization Program

^c. reported data from AEFI Information System in NIP

Chronic HBV infection can be characterized by a variety of disease states, such as inactive carrier, chronic hepatitis (CH), compensated cirrhosis (CC), decompensated cirrhosis (DC), hepatocellular carcinoma (HCC) and death from hepatitis-related causes. A Markov cohort model (Fig. 2) with yearly cycles was used to model the HBV chronically infected people who transition from one state to another (Table 4). At any time, people could die of other causes. “Carrier” refers to an HBV infected person who is asymptomatic and has no obvious hepatic inflammation with a normal alanine aminotransferase (ALT) level in the blood. CHB is defined as hepatic inflammation with elevated ALT levels continuing for at least 6 months^6,20. Previous studies demonstrated that chronic active hepatitis developed in more than 25% carriers and often resulted in cirrhosis.²¹ The immune tolerant phase as carriers for early child infection can be as long as 20-30 years, so in our model we assumed that among those with neonatal HBV infections, 25% would become chronic hepatitis B patients at 20 y old.

Figure 2. Markov cohort model for the natural history of HBV infection.

Estimating the burden of HBV infection with infant hepatitis B immunization program

To determine the number of unprotected infants, we used the reported coverage from the China Information Management System for Immunization Programming (CIMSIP): 99.60% for HepB3 and 95.88% for timely birth dose (TBD) among all infants. According to the PMTCT project report, the HBsAg screening rate in pregnant women was 94.2%, and 91.2% of the infants born to screened HBsAg-positive motherss were administered HBIG. For the infants of HBsAg+ and HBeAg+mothers, the efficacy of HepB combined HBIG could be 91.0%, compared to 83.1% for HepB vaccine only. And for the infants of HBsAg+ and HBeAg- mothers, the related efficacies for HepB combined HBIG and HepB only were 93.3% and 94.5%, respectively (Table 4).

In the unprotected susceptible population, the age-specific HBV infection rates were the same as above.

Costs associated with disease

Direct health care costs included those associated with the treatment of the hepatitis-related events after HBV infection. Both outpatient and inpatient costs were included. All the direct-medical and direct-nonmedical cost data were obtained from previous studies (Table 4).

For indirect costs, our model estimated the total economic value of life lost from hepatitis B related deaths. To estimate the economic value of life, we used the human capital approach, which assumes that the value to society of an individual's life can be measured by his or her future production potential, or the present discounted value of lost earnings.²² The earnings estimate used in the analysis was a weighted average of the proportion of participation in the workforce and non-participation in the workforce for urban and rural areas. Future earnings were adjusted for an expected 1% annual increase in labor productivity. Annual income and workforce participation data were obtained from the National Bureau of Statistics of China.^23,24

Costs associated with vaccination and HBIG administration

In our model, the program costs included screening, vaccine, HBIG, administration cost (including AD-syringe cost, cold chain transportation, staff remuneration, training, supervision, and surveillance), and AEFI treatment cost. HBIG is administered with the first dose of HepB in the hospital, so its administration cost was not calculated separately. We include 2 series AEFI events that may possibly occur after HepB vaccination: hrombocytopenic purpura and angioedema. Based on the National AEFI surveillance system, hrombocytopenic purpura incidence was 0.03/1,000,000 doses and angioedema incidence was 0.02/1,000,000 doses.^25,26

Sensitivity analyses

We conducted one-way sensitivity analyses to assess the robustness of our economic estimates and to estimate the impact of potential changes to the immunization program. We assessed the effect on the model of varying parameters such as vaccine and HBIG prices, wastage rate, administration cost, work loss for vaccination, direct coverage rate, screening sensitivity and specificity, screening rate and cost, HBIG infection rate, vaccine and HBIG efficacy, the cost of health outcomes, discount rate, etc. The lower and upper bounds of most parameters were shown in Table 4.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgements

We would like to express our thanks to Zhou Fangjun, PhD for his helpful comments.

Funding

The Chinese Ministry of Science and Technology Program for Important Infectious Diseases Control and Prevention (grant 2008ZX10002-001 and 2012ZX10002001).