Evaluating the potential cost-effectiveness of microarray patches to expand access to hepatitis B birth dose vaccination in low-and middle-income countries: A modelling study

Timely birth dose vaccination is key for achieving elimination of hepatitis B, however, programmatic requirements for delivering current vaccine presentations to births outside of health facilities inhibits coverage within many low-and middle-income countries (LMICs). Vaccine technologies in development such as microarray patches (MAPs) could assist in overcoming these barriers, but procurement could incur higher per-dose commodity costs than current ten-dose (US$0.34) and single-dose (US$0.62) vial presentations, necessitating an evaluation of the economic value proposition for MAPs. Within 80 LMICs offering universal hepatitis B birth dose vaccination, the cost-effectiveness of using MAPs to expand coverage was evaluated using a mathematical model. We considered three potential per dose MAP prices (US$1.65, US$3.30, and US$5.00), and two potential MAP use-cases: (1) MAPs are used by lay-health workers to expand birth dose coverage outside of health facility settings, and (2) MAPs are also preferred by qualified health workers, replacing a proportion of existing coverage from vaccine vials. Analysis took the health system perspective, was costed in 2020 US$, and discounted at 3% annually. Across minimal (1% additional coverage) and maximal (10% additional and 10% replacement coverage) MAP usage scenarios, between 2.5 (interquartile range [IQR]: 1.9, 3.1) and 38 (IQR: 28,44) thousand DALYs were averted over the estimated 2020 birth cohort lifetime in 80 LMICs. Efficiency of MAPs was greatest when used to provide additional coverage (scenario 1), on average saving US$88.65 ($15.44, $171.22) per DALY averted at a price of US$5.00 per MAP. Efficiency was reduced when used to replace existing coverage (scenario 2); however, at prices up to US$5.00 per MAP, we estimate this use-case could remain cost-effective in at least 73 (91%) modelled LMICs. Our findings suggest even at higher procurement costs, MAPs are likely to represent a highly cost-effective or cost-saving mechanism to expand reach of birth dose vaccination in LMICs.


Appendix 1: Regional aggregation of low and middleincome countries within the model
Analysis focused on low-and middle-income countries (LMICs) providing hepatitis B birth dose vaccination for all newborns in the WHO/UNICEF estimates of National Immunization Coverage (WUENIC) [1]. Coverage estimates for LMICs could be from either 2019 or 2020, with the most recent in each LMIC used for modelling. From these countries, we modelled the six WHO regions, plus a composite "All LMICs" region encompassing all 77 LMICs. Population-weighted parameter averages (using 2020 births) [2] from relevant country subsets were used for regional analysis, excluding those with missing data.
Each modelled LMIC also needed to be aggregated into a Global Burden of Disease (GBD) world region for estimation of HBeAg prevalence among 15-49y females (reproductive age) [3], GBD Super Regions for estimates of costs associated with vaccination outreach [4], and World Bank income classification for estimation of human resource costs [5]. Missing data for modelled LMICs was imputed using the relevant WHO regional average (Table 2, Main Text).  [6] GBD World Region [7] GBD Super Region [7] World Bank Income Classification# [8] Hepatitis

Baseline distribution of hepatitis B birth dose coverage: facilities vs community
To account for reported coverage inequities amongst births occurring outside of health facilities (i.e., in the community) within the model [9,10], baseline estimates of hepatitis B birth dose vaccination coverage were weighted according to the below formula. This was required as national estimates of hepatitis B birth dose coverage are reflective of coverage across an entire annual birth cohort, and hence do not capture any heterogeneity across birth locations.
A weighting factor (w.f.) of 2 was used within the model, consistent with observational data indicating odds of vaccination of births at home are approximately half that of those occurring within health facilities [11,12]. Facility: Community: BD: Baseline hepatitis B birth dose vaccine coverage, wf: Weighting Factor; odds of facilitated vaccination against community, Facility/Community: Proportion of facility/community births

Baseline distribution of hepatitis B birth dose coverage: timing of delivery
It was assumed that birth dose vaccination timing would be prompter for facility births, as compared to those in the community, given vaccine availability and vaccine access considerations. At baseline, we assumed that 80% vaccinations given in a health facility would be timely (<24 hours); a midpoint average between observed values of 64% and 90% [13,14]. Remaining doses were uniformly distributed among the non-timely vaccination strata (i.e., 6.7% on each of day 2, days 3-7 and days 8-41).
Within the community (births outside of health facilities), barriers currently inhibiting timely vaccination coverage include suboptimal qualified health worker attendance and/or travel distance from a health facility where birth dose vaccines are kept [15][16][17]. To capture these potential latencies, we assumed that 30% of baseline coverage in the community was timely (<24 hours), 40% delivered on day 2, and 15% on each of the days 3-7 and days 8-41-time strata.

Appendix 3: Additional details on vaccination costing Vaccine Supply Chain Costs
A supply chain cost component was used to account for the costs associated with transporting vaccines from a national store to a health facility, and storage of vaccines in a cold chain. Economic costs were taken from a review by Portnoy and colleagues, which estimated the non-commodity costs of introducing a new vaccine into a routine immunization program [18]. Costs were available for 78 (98%) modelled LMICs and assumed constant for both vial and MAP presentation of vaccine.
Source costs were presented in 2018 US$ and modelled in 2020 US$ (CPI adjusted). Impacts of alternate assumptions for MAPs were quantified in a one-way sensitivity analysis (Main Text, Fig 3).

Vaccine Commodity Costs
Within the model, per dose commodity costs considered three individual components: vaccines, needle and syringe, disposal boxes; plus, an allowance for wastage where required. Across all modelled LMICs, commodity costs were assumed equal.    Delivery times for a CPAD within the PATH analysis were used as a proxy for MAPs, consistent with application times in a childhood MAP vaccination study [24]. However, target product profiles for MAPs indicate this time could be much longer and remain acceptable for use [25], hence the impact was evaluated in one-way sensitivity analysis (Main Text, Fig 3). Valuation of time was linked to per capita GDP, using multipliers from an econometric analysis by Serje and colleagues according to World Bank Income Classification status [5]. Health workers were assumed to work 37.5-hour weeks for 48 weeks per year and that doctors, nurses and midwives would vaccinate births with an equal probability; however, costs were weighted proportionally to the number of each cadre within a LMIC [26,27]. Valuation of trained lay-health workers to administer MAPs to births in the community used the "other health workers" multiplier as a proxy.
Regional analysis used population weighted (2020 births) averages across relevant subsets of LMICs.

Outreach Costs
Outreach aimed to capture costs associated with providing a birth dose vaccination in the community, including transport costs and travel time. Costs were taken from an analysis by Nayagam and colleagues [4]; however, costs in the South Asia GBD Super Region were deemed unrealistic ($32 per dose, 2020 USD). For LMICs within this region, a population weighted (2020 births) average of outreach costs elsewhere was used in lieu.
As the study only provided point estimate costs, uncertainty was modelled as a uniform ±5% from point estimates.

Appendix 5: Supplemental Analyses Sensitivity analysis: larger coverage gains due to MAPs
Incremental gains in additional coverage only (scenario 1) were associated with greater health benefits (Table L); however, did not impact cost-effectiveness of MAPs (i.e., ICERs did not change).
Cost-effectiveness of replacement coverage (scenario 2) was enhanced with higher levels of additional coverage, but remained less cost-effective (i.e., higher ICERs) compared to additional coverage only (scenario 1).

Fig C:
Outcomes of assumptions on incremental coverage gains from MAPs within the model. Shaded region represents Interquartile Range (IQR) of 1000 model simulations.

Sensitivity analysis: MAPs do not create new coverage, but are used to replace existing needle and syringe coverage (and hence improve timing of delivery only)
Cost-effectiveness of using MAPs to replace existing coverage only (Supplemental Figure 3, right panel) was equal across all analyzed increments (1%, 5% and 10%). While less cost-effective than additional coverage from MAPs (i.e., higher ICERs for a given MAP price), this sensitivity analysis indicates use of MAPs to only achieve gains in birth dose timeliness may present some value.

Sensitivity analysis: MAPs also provide additional, new coverage in facilities (rather than just additional coverage in the community).
For this supplemental analysis, we assumed that additional coverage from MAPs in facilities was half of that in the community (i.e. 0.5% additional coverage). However, as the majority of births remain in health facilities and vaccination does not incur an outreach cost, even constrained use in this setting improves cost-effectiveness.

Appendix 6: Sensitivity Analysiscost-effectiveness of MAPs if implemented with a baseline CTC approach
The hepatitis B birth dose is deemed a CTC priority vaccine by the WHO CTC Working Group [29]. In this sensitivity analysis, we investigated how a theoretical baseline scenario where the CTC approach was already being used would impact the cost-effectiveness of MAPs to deliver the birth dose.
We assumed that all vial vaccines used under a CTC approach would be single-dose, and that each vial would be fitted with a combined vaccine vial monitor (VVM) and threshold temperature indicator (TTI). As vaccines were in vials, administration remained a task for qualified health workers only (i.e., required a needle and syringe). Consistent with previous CTC hepatitis B birth dose modelling studies [30,31], at baseline, CTC provided an additional 5% timely (day 1) coverage of births in facilities and 10% timely coverage of births in the community. Additionally, CTC improves timeliness of 5% of facility and 10% of community vaccinations by replacing existing cold chain coverage, modelled as a left shift (i.e., days 8-41 to days 3-7, days 3-7 to day 2, and day 2 to day 1) in vaccine timing.
Results from this analysis suggest introduction of MAPs under a baseline scenario with a CTC approach would have negligible impact on their cost-effectiveness, as compared to a baseline scenario entirely reliant upon the cold chain.