Selective Hepatitis B Birth-Dose Vaccination in São Tomé and Príncipe: A Program Assessment and Cost-Effectiveness Study

Abstract.

São Tomé and Príncipe (STP) uses a selective hepatitis B birth-dose vaccination (HepB-BD) strategy targeting infants born to mothers who test positive for hepatitis B virus (HBV) surface antigen. We conducted a field assessment and economic analysis of the HepB-BD strategy to provide evidence to guide development of cost-effective policies to prevent perinatal HBV transmission in STP. We interviewed national stakeholders and key informants to understand policies, knowledge, and practices related to HepB-BD, vaccine management, and data recording/reporting. Cost-effectiveness of the existing strategy was compared with an alternate approach of universal HepB-BD to all newborns using a decision analytic model. Incremental cost-effectiveness ratios (ICERs) were calculated in 2015 USD per HBV-associated death and per chronic HBV case prevented, from the STP health-care system perspective. We found that STP lacked national or facility-specific written policies and procedures related to HepB-BD. Timely HepB-BD to eligible newborns was considered a high priority, although timeliness of HepB-BD was not monitored. Compared with the existing selective vaccination strategy, universal HepB-BD would result in a 19% decrease in chronic HBV infections per year at overall cost savings of approximately 44% (savings of USD 5,441 each year). We estimate an ICER of USD 5,012 saved per HBV-associated death averted. The existing selective HepB-BD strategy in STP could be improved through documentation of policies, procedures, and timeliness of HepB-BD. Expansion to universal newborn HepB-BD without maternal screening is feasible and could result in cost savings if actual implementation costs and effectiveness fall within the ranges modeled.

INTRODUCTION

It is estimated that 85–90% of hepatitis B virus (HBV)–associated deaths are vaccine-preventable.¹ Worldwide, 30% of cirrhosis and 53% of hepatocellular cancer deaths are attributable to chronic HBV infection.² The most important risk factor for chronic HBV infection is early infection, especially perinatal infection, which has an approximately 70–90% risk of chronic persistence.^3–5 The WHO recommends hepatitis B birth-dose vaccination (HepB-BD) of all newborns, preferably within 24 hours of delivery, which is 80–90% effective in preventing perinatal HBV transmission, followed by two or three additional doses of HBV vaccine.⁵

São Tomé and Príncipe (STP), with a population of approximately 193,000 and an annual birth cohort of approximately 6,400, is highly endemic for chronic HBV infection, with a seroprevalence of hepatitis B surface antigen (HBsAg) of 10% in men and 6.1% in women.^6,7 São Tomé and Príncipe introduced selective HepB-BD in 2002, targeting infants born to chronically HBV-infected women identified by HBsAg screening during routine prenatal visits. Because only selected infants are targeted for vaccination, national HepB-BD coverage is low (3%); however, STP has high coverage of prenatal services (97.5%, with at least one antenatal care [ANC] visit) and births at a health-care facility (HCF) (91%).^8,9 Hepatitis B birth-dose vaccination coverage among infants born to HBsAg-seropositive women is correspondingly high (national administrative coverage estimate is 92%), and both one-dose (98%) and three-dose coverage (96%) of pentavalent vaccine (which includes diphtheria, tetanus, pertussis, hepatitis B, and Haemophilus influenzae b) are high,⁸ indicating that dropout is minimal and that most⁹ infants receiving HepB-BD will receive the complete series of vaccination.

We conducted an assessment of the current HepB-BD strategy in STP to characterize and document policies, practices, and barriers related to HepB-BD implementation and reporting. In addition, this study aimed to evaluate the cost-effectiveness of the existing selective HepB-BD strategy as compared with a WHO-recommended universal HepB-BD strategy to provide evidence-based recommendations to the country to guide future policy changes.

MATERIALS AND METHODS

Hepatitis B birth-dose vaccination program assessment.

During August 2015, we assessed the five facilities on São Tomé island that provide maternity services—the national referral hospital, three district public hospitals, and one health post. In addition, we collected information from key informants in the Ministry of Health (MOH) and from representatives of partners and stakeholder organizations. The assessment used structured and unstructured survey questions that were developed during a consultative workshop at the WHO Regional Office for Africa headquarters, as part of planned multicountry evaluation HBV-BD programs across the African region.

During the assessment of STP’s HBV-BD program, MOH officers were interviewed regarding HepB-BD policy, training, supervision, vaccine management, and coverage monitoring. At each facility, one key informant from the Expanded Programme on Immunization (EPI) and/or a supervising maternity staff member was interviewed on birth-dose policy, practice, and knowledge. Written childbirth and newborn care policy and procedure documents were reviewed. Newborn log books and monthly monitoring reports were reviewed to evaluate the number of newborns who received HepB-BD and the timing of the dose. Simple descriptive analyses using R statistical programming language (R Foundation for Statistical Computing, Vienna, Austria) were performed to summarize the findings using one-way frequencies.¹⁰

Cost-effectiveness analysis.

We conducted a cost-effectiveness analysis to guide the STP MOH in formulating policy to reduce HBV-related health burden using HepB-BD. We created a decision analytic model (Supplemental Appendix A) using @RISK (Palisade Software, Ithaca, NY) to compare the current selective HepB-BD strategy in STP (“base case”) against a hypothetical alternative strategy (“alternative”) of universal facility-based newborn HepB-BD without maternal HBsAg screening.

Selective and universal vaccination policy assumptions.

Under the “base-case” strategy, pregnant women undergo screening for HBsAg using a rapid test (Alere, Waltham, MA) as part of routine preventive ANC, which includes integrated antenatal screening for syphilis and HIV. Following delivery in a HCF, infants born to HBsAg-positive women receive HepB-BD.¹¹ Women whose HBsAg status is unknown are tested if they present to a HCF in labor. Under the alternative strategy, routine prenatal HBsAg testing would be discontinued from integrated antenatal screening without disrupting the other components of the existing antenatal testing program, and instead, all infants would be targeted to receive HepB-BD within 24 hours of birth. We assumed that there would continue to be no outreach program to provide HepB-BD outside a HCF.

Decision analytic model assumptions and data sources.

We used a static decision analytic model to assess the costs and effects of providing the two vaccination strategies to a cohort of 6,487 hypothetical mothers (corresponding to the annual pregnancy cohort in STP and their newborns).^6,12 The primary cost-effectiveness measure evaluated was the incremental cost-effectiveness ratio (ICER) per death averted by HepB-BD, expressed as U.S. dollars per death averted. We also assessed other outcomes, including the total number of chronic HBV infections, total number of HBV-related deaths, and the ICER per chronic HBV infection averted. We did not consider any secondary benefits of vaccination, and in this model, we only considered hepatitis B infections in the first year of life for the cohort due to vertical transmission.

The decision analytic model included the following health states that could be experienced during the lifetime of a birth cohort member: infected or uninfected with HBV, progression to chronic HBV, death due to hepatitis B complications, or death unrelated to hepatitis B. We used transition probabilities, vaccine-effectiveness, epidemiological data, and cost assumptions from the published literature whenever available (Table 1).^{3,7,9,12–21} Transition from chronic infection to HBV-attributable death is modeled as a single parameter that includes the risk of death due to either cirrhosis or hepatocellular carcinoma; the range of this parameter is sourced from a modeling study that incorporated administrative health record data from Taiwan, the United States, China, and the Gambia and from a prospective cohort study of more than 22,000 individuals in Taiwan.^3,17 We assumed that maternal HBV prevalence was 6.1% based on recent serological survey data.⁷ We accounted for the differential vertical transmission risk associated with the hepatitis B e antigen (HBeAg) status of the mother.^15,22,23 Although high-quality age- and gender-specific serological data exist on the prevalence of chronic HBV infection in STP, no data exist on HBeAg prevalence in STP. We, therefore, assumed that 30.6% of HBsAg-positive mothers were HBeAg positive, based on published estimates for HBeAg prevalence in West sub-Saharan Africa among women aged 20–29 years.²¹ Estimates for the age range 10–19 years and 40–49 years were used as upper (41.1) and lower (15.5) bounds, respectively.²¹ For HepB-BD coverage in hospitals, we used the national estimate for percent coverage of birth dose among HBV-positive mothers (92%) as the only available empirical estimate from STP. Coverage among HBsAg-positive mothers delivering in health facilities was assumed to be the same (92%) under both the base case and the alternative scenario in which screening was discontinued, which was considered reasonable given STP’s relatively well-performing immunization system and high Bacillus Calmette-Guerin coverage.

Table 1.

Modeling assumptions and parameters for HepB-BD vaccination cost-effectiveness—STP, 2015

Parameter	Point estimate*	Range†	Sensitivity range‡	Source, or reference number
Vaccine and rapid test performance
Effectiveness of HepB-BD	88%	50–95	–	3,13
HBV test sensitivity	98.2%	94.7–99	–	14
HBV test specificity	99.9%	99.3–100	–	14
Perinatal transmission risk
Transmission risk for HBeAg-positive mothers	49.6%	9.2–90	–	3,15
Transmission risk for HBeAg-negative mothers	15.3%	1.8–28.7	–	3,15
Disease model
Transmission to symptomatic acute infection	1%	0–8	–	3,16
Symptomatic to acute fulminant infection	10%	–	–	3,16
Acute fulminant infection to death	70%	–	–	3,16
Acute perinatal infection to chronic infection	88.5%	84–94	–	3,15
Chronic infection to HBV-attributable death	25%	10–40	–	3,17
Costs§
Per-unit cost of vaccine dose and supplies‖	$0.38	–	0.00–0.76	26
Costs of HepB-BD (selective)	$0.28	–	±100%	18
Cold chain maintenance	$0.01	–	–	–
Program management	$0.02	–	–	–
Recordkeeping	$0.11	–	–	–
Routine facility-based service delivery	$0.07	–	–	–
Supervision	$0.04	–	–	–
Surveillance	$0.03	–	–	–
Additional per-dose costs of universal strategy	$0.54	–	±100%	18
Training	$0.05	–	–	–
Other‖¶	$0.05	–	–	–
Advocacy and social mobilization	$0.44	–	–	–
Unit cost of rapid test kit and supplies	$0.95	–	0.00–1.90	Market purchase price for São Tomé
Nursing time required for test	17.5 minutes	–	8.8–26.3	19,20
Nursing time required for posttest counseling (positive test)	22.5 minutes	–	16.9–33.8	19,20
Nursing salary per hour	$1.89	$1.69–2.10		20¶
Epidemiological assumptions and health system parameters
Birth cohort	6,385	–	–	6
Rate of stillbirth in STP	1.6%	–	–	12
Maternal prevalence of HBsAg	6.1%	5.2–7.1	–	7
Prevalence of HBeAg among HBsAg-positive mothers	30.6%	15.5–41.1	–	22
Overall facility birth proportion	91.0%	88.8–93.2	–	9
ANC coverage (at least one visit)	97.5%	96.2–98.7	–	9
Timely BD coverage among hospital births	92%	–	80–100	National surveillance data (% coverage of HBsAg-positive mothers)
Timely birth-dose coverage among home births	50%	–	0–100	Authors’ assumption
Rate of screening for HBsAg if ANC	95%	–	80–100	Authors’ assumption
Percent of mothers without ANC who seek hospital births	75%	–	50–100	Authors’ assumption
Probability of rapid test for a facility birth who did not receive ANC	80%	–	50–100	Authors’ assumption

ANC = antenatal care; HBV = hepatitis B virus; HBeAg = hepatitis B e antigen; HBsAg = hepatitis B surface antigen; HepB-BD = hepatitis B birth-dose vaccination; STP = São Tomé and Príncipe.

* When a range of parameter estimates was sourced from the published literature, the mean value was chosen as a point estimate.

† When a range of parameter estimates was sourced from the published literature, we incorporated uncertainty into the estimates by modeling those parameters as β-PERT distribution using the point estimate and ranges as mode, minimum, and maximum inputs.

‡ For sensitivity analyses, parameters were modeled as a uniform distribution across the sensitivity range.

§ All costs are reported in 2015 USD.

‖ Does not include wastage due to the use of multidose vaccine vials; uncertainty related to wastage is captured in the sensitivity range.

¶ Includes material costs and physical supplies related to social mobilization and training.

Cost assumptions and data sources.

We conducted the cost-effectiveness analysis from the STP health-care system perspective inclusive of all payers, with a lifetime analytic horizon and a 12-month time-frame for costs, which are expressed in 2015 USD. We did not include societal indirect costs because of lost productivity from hepatitis B morbidity and mortality. We chose an incremental approach to assess costs in this study and, therefore, only included costs that would be directly attributable to the startup costs, increased personnel demand, and materials specifically associated with the implementation of an alternative universal HepB-BD strategy.^24,25 We assumed that costs associated with initial introduction of the existing HepB-BD strategy and shared overhead costs, such as the cost of phlebotomy, and posttest counseling, as part of the existing integrated antenatal screening program for syphilis and HIV, were absorbed by the health system. Direct nonmedical costs borne by parents (e.g., transportation to health-care facilities) and indirect costs (e.g., productivity losses due to time spent obtaining health-care services) related to HBsAg screening or vaccination are not included, as both interventions would occur as part of routine integrated antenatal screening for syphilis and HIV, and postpartum care under either scenario. Under the universal vaccination strategy, infants born at home would be brought to a hospital to receive HepB-BD. Under the existing system, infants born at home are expected to be brought immediately to a hospital to receive other services; thus, there would be no incremental transportation costs associated with administration of HepB to infants born at home. Costs of treatment of HBV complications were not included because the focus of the analysis was on the differential program costs of screening and vaccination interventions, and the treatment costs were assumed to be directly proportional to the chronic infections modeled as an outcome of these interventions.

Under the base-case strategy, we included the unit cost of hepatitis vaccine procurement in STP and generic per-dose costs of cold chain maintenance, program management, recordkeeping, routine facility-based service delivery, supervision, and surveillance associated with immunization program implementation, drawn from a 2011 cost study of pneumococcal vaccination in Benin.^18,26 This study was used because the primary cost data were more recent and more comprehensive than those of cost studies of Hepatitis B introduction in other African countries; however, sensitivity analyses were conducted to include the range of costs identified in these previous studies.^24,27 Costs of screening included the unit cost of the maternal screening test (L. M. do Sacramento Bonfim, personal communication, UNICEF STP), personnel time associated with rapid testing, and pre- and posttest counseling. We assumed posttest counseling in the case of negative tests would be minimal and would be conducted together with other posttest counseling, and the costs would be negligible and absorbed into the overhead shared health system costs. Personnel costs were extrapolated using the estimated personnel time burden for rapid syphilis testing in routine ANC in Haiti, and nursing salary costs from Comoros and Cote d’Ivoire, West African nations with similar gross domestic product (GDP) to STP.^19,20

For the alternative strategy, additional incremental costs associated with training, advocacy, and social mobilization associated with the change in the program from selective to universal vaccination were included, also sourced from the Benin pneumococcal vaccine cost study, with sensitivity analyses to include the range of incremental costs identified in older cost studies specific to Hepatitis B introduction.^18,24,27 Guided by information provided by STP MOH logistics and cold chain staff, we assumed that vaccine transportation and storage capacity in the Maternal, Neonatal, and Child Health/EPI system were not at maximum and that no additional specific personnel or capital equipment costs were associated with the increased number of HBV monovalent vaccine required for universal vaccination. In the base-case strategy, HBV rapid testing results are not used to make decisions on referral of women for clinical management, or for surveillance purposes, and are only used to identify mothers whose infants will receive HBV-BD. Guided by consultation with MOH staff, we, therefore, assumed that hepatitis B rapid testing would not be continued as part of routine ANC if universal HBV-BD was adopted, and, therefore, did not include those costs.

Model uncertainty and sensitivity analyses.

When a range of parameter estimates was sourced from the published literature, we incorporated uncertainty into the estimates by modeling those parameters as β-PERT distribution using the point estimates and ranges listed in Table 1 as mode, minimum, and maximum inputs and performed a total of 10,000 simulations, using random draws from the probability distributions as inputs to the model.²⁸ We used the 50, 2.5, and 97.5 percentiles of the Monte Carlo simulation results to report the estimate and 95% CI for acute infections, chronic infections, HBV-associated deaths, and ICERs. In addition, we conducted one-way sensitivity analyses to estimate the impact of variations to key assumptions that were not directly sourced from published data, by modeling those parameters as a uniform distribution across the sensitivity ranges listed in Table 1. We varied the assumed birth-dose coverage for hospital births and home births, the proportion of mothers tested for HBV during ANC, the proportion of women without ANC who deliver in a facility, and the rate of hospital-based HBV rapid testing of women who did not receive ANC. We varied the supply unit price and service delivery costs of both HepB-BD and HBsAg rapid testing by ± 100%; this range includes the unit costs identified in earlier cost studies of hepatitis B introduction in Mozambique and Ethiopia.^24,27 Uncertainty ranges and sensitivity analyses were applied to both the base-case and alternative strategies in the model.

RESULTS

Hepatitis B birth-dose strategy assessment.

All five facilities that offered maternity services reported providing HepB-BD according to the national selective vaccination policy. Findings of the HepB-BD program assessment are summarized in Table 2. National or hospital-specific written policies and procedures related to HepB-BD do not exist; however, knowledge regarding HepB-BD was high, and reported practices were generally uniform among HCFs. Health-care facility staff universally believed that breastfeeding should be withheld until after HepB-BD to prevent vertical HBV transmission, although this policy was not explicitly written in guidelines and is not supported by WHO recommendations.^1,5,29

Table 2.

Facility characteristics and HepB-BD vaccine knowledge, practices, and vaccine management—São Tomé and Príncipe, 2015

	All facilities (n = 5)	National public hospital (n = 1)	District public hospital (N = 3)	Health center (N = 1)
Background characteristics
Total number of deliveries per facility (median [range])	336 [31–4,383]	4,383	336 [177–357]	31
% Of HF where mothers stay ≥ 24 hours postdelivery	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Knowledge (maternity ward supervisor)
Received training on HepB-BD	4 (80%)	1 (100%)	3 (100%)	0 (0%)
Know that a mother can transmit hepatitis B virus to her baby	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Know that recommended HepB-BD administration is < 24 hours of birth	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Practices
Vaccinate all newborns with HepB-BD	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Follow standard written protocols for HepB-BD administration	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Provide written documentation of HepB-BD to mother	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Vaccinate in the delivery room	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Administer to low-weight babies (2–2.5 kg)	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Administer to very low-weight babies (< 2 kg)	4 (80%)	1 (100%)	3 (100%)	0 (0%)
Administer to premature babies	3 (60%)	1 (100%)	2 (67%)	0 (0%)
Administer to unstable babies	1 (20%)	0 (0%)	1 (33%)	0 (0%)
Provide outreach HepB-BD to home births	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Charge patient for HepB-BD administration	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Patients sometimes refuse HepB-BD	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Offer HepB-BD daily	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Require a physician order for HepB-BD	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Vaccine management
Stockout > 2 weeks in 2014	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Vaccine fridge is EPI-approved	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Observed VVM stage 3–4 in fridge	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Fridge is monitored at least × 2/day	5 (100%)	1 (100%)	3 (100%)	1 (100%)
Vaccine is obtained from the Ministry of Health EPI	5 (100%)	1 (100%)	3 (100%)	1 (100%)

EPI = Expanded Program on Immunization; HepB-BD = hepatitis B birth dose; HF = health facility; VVM = vaccine vial monitor.

We were unable to calculate HepB-BD coverage among infants of HBsAg-positive mothers because HBsAg status of mothers admitted to maternity wards was not documented unless the infant was vaccinated. We chose not to use all pregnant women as a denominator for the assessment because our goal was to assess the performance of the existing program rather than calculate an estimate of overall coverage. Date of HepB-BD was not recorded by any of the sampled facilities; therefore, we could not assess the timeliness of HepB-BD.

Cost-effectiveness of selective versus universal HepB-BD.

We found that implementation of a universal HepB-BD strategy could result in overall cost savings to the STP health-care system of approximately 44% (95% CI, 42–46%), while preventing the same number of chronic HBV infections per year (Table 3). The CIs in the estimates reflect the deterministic uncertainty in the model resulting from the ranges of available data used in the model. Figure 1 shows the relative impact of each parameter on the uncertainty of the ICER estimate. Uncertainty around risk of death due to chronic HBV infection and risk of perinatal transmission had the greatest impact on the total variation in the ICER for deaths and infections averted. Higher risk of death from chronic HBV infection and higher risk of perinatal transmission resulted in increasing cost savings of a universal strategy.

Table 3.

Monte Carlo estimates from decision analytic model of HepB-BD vaccination cost-effectiveness for a single year’s birth cohort—São Tomé and Príncipe, 2015

Modeled HepB-BD approach	Number of infections	Number of chronic infections	Number of HBV deaths	Cost for one pregnancy cohort of 6,487 mothers (in 2015 USD)	ICER (USD per chronic HBV infection averted)	ICER (USD per HBV-related death averted)
Universal vaccination	24 (14–40)	21 (13–35)	5 (3–10)	6,870 (6,824–6,915)	–	–
Selective vaccination	29 (19–45)	26 (17–40)	6 (3–11)	12,311 (11,885–12,736)	1,234 (819–1,880)	5,012 (2,901–9,239)

HBV = hepatitis B virus; HepB-BD = hepatitis B birth-dose vaccination; ICER = incremental cost-effectiveness ratio.

Base-case estimates and 95% CI (in parentheses) calculated by reporting the 50th, 5th, and 95th percentiles of 10,000 Monte Carlo iterations of the decision analytic model. Estimates are rounded to the nearest whole. A positive value for ICER is interpreted as cost savings if a universal vaccination strategy is adopted.

Figure 1.

Tornado plot of stochastic sensitivity analysis accounting for parameter uncertainty in hepatitis B birth-dose (HepB-BD) modeling, São Tomé and Príncipe, 2015. Parameters on the y axis are ranked by their effect on the median incremental cost-effectiveness ratio (ICER) estimate after 10,000 simulations. Range of bars indicates the impact on the 50th percentile of Monte Carlo simulation results across the range of possible values for each parameter. Positive ICER indicates higher cost-effectiveness of a universal vaccination strategy.

Figure 2 shows the result of sensitivity analyses to evaluate the impact of different scenarios and assumptions related to the performance of the health-care system on the ICER. The estimated coverage of HBsAg screening during ANC had the highest impact on the ICER, and the higher the rate of antenatal screening, the more cost-effective the universal vaccination strategy becomes. Across the range of scenarios considered by the sensitivity analyses, including the most optimistic and pessimistic assumptions for the performance of the health-care system in implementing these programs, selective screening was only more cost-effective if the personnel costs of rapid HBsAg testing or purchase price per rapid test are assumed to be negligible (< USD 0.10 and < USD 0.06 per test, respectively). Even if the service delivery cost of antenatal rapid HBsAg testing is zero, the incremental cost of a universal birth-dose strategy would be 509 USD per additional HBV death averted; recognizing the limitations of GDP-based thresholds (which do not consider affordability, budget impact, or the feasibility of implementation),³⁰ if historically recommended WHO thresholds were used for comparison, this ICER would be considered highly cost-effective, given that the GDP per capita for STP in 2014 was 1,811 USD.^31,32

Figure 2.

Tornado plot of sensitivity analysis evaluating the impact of different scenarios and assumptions on the incremental cost-effectiveness estimate in hepatitis B birth-dose (HepB-BD) modeling, São Tomé and Príncipe, 2015. Parameters on the y axis are ranked by their effect on the median incremental cost-effectiveness ratio (ICER) estimate after 10,000 simulations. Range of bars indicates the impact on the 50th percentile of Monte Carlo simulation results across the range of possible values for each parameter. Positive ICER indicates higher cost-effectiveness of a universal vaccination strategy.

DISCUSSION

For the current selective HepB-BD strategy, coverage among at-risk infants in STP could be improved through standardization and documentation of key policies and procedures, including reporting, monitoring, and having clear guidelines on indications and contraindications. Specific recommendations were given to the STP MOH on improvements to recordkeeping and reporting, in particular, ensuring that timely birth dose is recorded and tracked at the facility, district, and national level and ensuring that the HBsAg status of the mother, when known, is recorded and tracked together with birth-dose administration logs and in the health records of the infant. Although a specific written physician order is not needed in STP to administer HepB-BD, establishing written postnatal “standing orders” for care that include HepB-BD (e.g., incorporating into routine postnatal ward newborn checks) could help improve standardization of care according to the national policies.⁵ The lack of standardized written guidelines and refresher training could lead to inconsistency in implementation through staff turnover, as has been previously observed in other settings.³³ In addition, data recording and reporting policies and practices should be modified and strengthened to ensure that timely vaccination coverage of all eligible infants can be calculated and monitored at district and national levels. We also determined that maternity staff routinely encouraged and enforced withholding of breastfeeding initiation until HepB-BD administration, a practice that is not supported by the existing recommended policy, and risks leading to unnecessary delay in breastfeeding, with possible consequences for neonatal mortality.^1,5,34

The principal reason for not providing HepB-BD to all newborns in STP was the selective vaccination policy. The MOH estimates 92% coverage of at-risk infants in STP under this policy. However, this may be an overestimate; STP does not routinely monitor the timeliness of HepB-BD, and the effectiveness of a selective HepB-BD strategy might be limited because of incomplete screening and inconsistent application of policies.³³ We, therefore, applied a wide uncertainty range around the MOH estimate of 92% HepB-BD coverage; nevertheless, this was not the most influential source of uncertainty in the model (Figure 2), and all uncertainties were in a direction that favored universal HepB-BD, indicating that coverage of screening affects costs but not outcomes. Our study found that the selective vaccination strategy, at best, prevented the same number of chronic infections and HBV-related deaths per year as a universal vaccination strategy, but at nearly double the cost. The CIs for the effectiveness of the universal and selective strategies overlap, indicating that the universal strategy was not always more effective than the selective strategy in each model run; however, given the dramatically lower costs of the universal strategy than that of the selective strategy, the universal strategy was the more cost-effective strategy under almost every scenario examined. After these findings were presented to the MOH of STP in late 2015, a switch from selective screening to universal vaccination was discussed for possible inclusion in the next comprehensive multiyear strategic plan for immunization, which was in the process of being finalized; however, this was ultimately not included.

The universal vaccination model, in which a small percentage of mothers and infants are still missed, still resulted in suboptimal immunization coverage and new perinatal HBV infections, some of which could be prevented through concurrent administration of hepatitis B immunoglobulin (HBIG).^5,35 This intervention, guided by antenatal screening, is recommended by both the U.S. CDC³⁵ and the WHO⁵; however, as it is not presently available in STP, it was not modeled. In well-resourced settings with adequate health infrastructure, universal vaccination plus HBIG administration to infants born to HBsAg-positive mothers is commonly implemented; however, in resource-limited settings such as STP, this strategy is not cost-effective.³⁶

Our model assumed, guided by discussions with MOH staff, that under the alternative case scenario, antenatal HBsAg testing would be discontinued from the existing integrated antenatal testing program, which includes testing for syphilis and HIV. Although we assume this discontinuation would not have an unintended consequence of decreasing screenings for other infectious diseases, the potential for negative impact is unknown. In addition, antenatal HBsAg testing offers several theoretical benefits to infants of HBsAg-positive mothers, including affording the opportunity to follow up these infants to confirm seroconversion after HepB-BD,^5,35 and to provide other directed clinical interventions such as antiviral therapy to pregnant mothers testing positive for HBsAg and e antigen,³⁷ none of which are presently available in STP, but which might be available in the future. If antenatal HBsAg screening were discontinued to adopt a universal HepB-BD program, it could later be easily reactivated, should resources and capacity improve in STP to allow more directed care of HBsAg-positive pregnant women and their newborns.

Although antenatal screening was found to not be the most cost-effective approach to prevent perinatal hepatitis B transmission, the WHO Health Sector Strategy on Viral Hepatitis recognizes that most people with chronic HBV infection do not know their status and has a target of 90% of people with chronic hepatitis B (CHB) knowing their status.³⁸ Antenatal screening would contribute to this goal by informing mothers with CHB of their status and may be of value to continue in STP for that purpose alone; however, this was not directly evaluated from an economic perspective.

There are several limitations in the economic analysis. First, we did not model the risk of horizontal transmission; although data on the impact of birth dose on horizontal transmission are highly limited, our estimate may underestimate the impact of birth-dose vaccination under a universal vaccination policy by not accounting for reduced horizontal transmission in the first 6 weeks of life. In addition, we did not use a dynamic transmission model to incorporate the protective impact of herd immunity. Many of the model parameters were sourced from secondary data, introducing uncertainty to the findings. For example, cost data for service delivery of vaccination and rapid testing are not available from STP; therefore, we used data from Benin and Cote d’Ivoire, which are West African countries with similar levels of economic development (purchasing power parity-adjusted GDP per capita in 2014: 3,188, STP; 3,496, Cote d’Ivoire; 2,110, Benin).^18,20,32 Our analysis did not account for how cost parameters may change because of changes in availability of external funding to support HepB-BD introduction, such as announced by Gavi, the Vaccine Alliance in 2018, although we expect that any such external subsidies would only increase the cost-effectiveness of the universal HepB-BD strategy by reducing vaccine and scale-up operational costs.³⁹ In addition, the base-case cost parameters were sourced from non–HBV-specific primary data; however, we used a range that included values published in HBV-specific studies.^24,27 Sensitivity analyses to explore a wide range of costs, transition probabilities, and assumptions revealed that our finding of substantial cost savings per HBV-associated death averted under a universal vaccination strategy compared with a selective strategy was robust to this uncertainty in model parameters, demonstrated in Figures 1 and 2. Although we did not account for averted treatment costs of chronic HBV complications and subtract these from the numerator of the ICER, the policy conclusions of our study are unaffected by this decision because the universal vaccination strategy was found to be cost saving even when not considering treatment costs.

Finally, we calculated ICERs in terms of USD per chronic infection prevented or HBV-related death prevented, which cannot be easily compared against annualized economic thresholds such as GDP per capita, commonly used willingness-to-pay thresholds for outcome measures such as disability-adjusted life years (DALYs). Typically, when DALYs are used as the ICER denominator, an intervention is considered highly cost-effective if the ICER is smaller than the country’s GDP per capita. However, as we found that switching to a universal vaccination strategy would result in incremental cost savings per death averted under all but the most conservative cost assumptions, the acceptability of a universal vaccination model is assumed to be greater than that of the base-case model.

CONCLUSION

The current selective HepB-BD strategy in STP could be further strengthened by developing written national policies and training all relevant staff to ensure uniform implementation of the strategy according to accepted standards, including adequate timeliness, documentation, and reporting of HepB-BD. In addition, our assessment and model determined that if implementation matched our model assumptions, adopting a universal vaccination strategy in line with WHO recommendations could be simpler, lower cost, and less likely to miss children at risk for HBV infection than the existing selective vaccination strategy in STP. Future research should examine how the evolving global funding context and HBV elimination strategies—including combined screening and vaccination strategies—influence the cost-effectiveness of policy options in additional country settings and as part of global HBV elimination efforts.⁴⁰

Note: Supplemental appendix A appears at www.ajtmh.org.