Cross-posted from the EA Forum.
We previously estimated the cost of developing a vaccine from scratch to be $460M to $1.9B with a mean of $960M. However, this still does not tell us the full cost of a vaccine, because developing a vaccine but then not ever using it accomplishes nothing. Instead, you need to roll out the vaccine to people, which costs more money. Thus, vaccine R&D could be thought of as “unlocking” the opportunity to roll-out a vaccine, and the hope is that the high cost-effectiveness of rolling out a vaccine will help offset the high cost of vaccine R&D.
Understanding Roll-out Costs
Rolling out a vaccine costs more than just the price of manufacturing the vaccine. As Wolfson, et. al. (2008) notes (see Table 3), there is a significant logistics component to the costs – you have to ship the vaccine to its destination; keep it stored correctly (e.g., using a “cold chain” to make sure the vaccine is consistently stored at the correct temperature at all times); pay for waste management; train, supervise, and compensate volunteers and support staff; and monitor and evaluate vaccine performance. You may also need to pay for media to promote the vaccine and help the populace understand and accept it.
The price per dose of a vaccine is aimed not just at recovering variable manufacturing costs, but also aimed at recovering fixed costs including R&D (Elder & Cohn, 2013) and, possibly, the cost of trials (Weinberg et al., 2012). If we accounted for R&D spending both separately and in the per dose costs of vaccines during roll-out, we could be overestimating vaccine costs.
When thinking of costs, it is useful to separate “fixed costs” (one-time costs incurred from setting up a vaccine project) from “variable costs” (costs that will be incurred on every vaccine regardless of scale). However, with economies of scale, it is difficult to tell where fixed costs end and variable costs begin. For example, many modern vaccines today are cheaper to roll out because they can rely on infrastructure and technology built during previous rollouts. Even simple and apparently unrelated infrastructure like building roads and bridges could be conceived as an investment in reducing the rollout costs of vaccines. Estimating roll-out costs of vaccines is very tricky because it is unclear to what degree should we include these investments. Bishai, Johns, Lefevre, and Nair (2010) found that distributing a vaccine could be up to 38x as expensive in highly remote areas compared to core, easily accessible areas with strong pre-existing infrastructure.
At this time, I no longer include the ebola vaccine, because it is too early stage to know how much it will cost to roll out. For the remaining vaccines, I find the following rollout costs and elaborate more below:
Smallpox - $0.73 - $47.62 per child
Measles - $1 - $38 per child
Rotavirus - $3 - $28 per child
HPV - $2.55 - $22.71 per child
HIV - $50 - $160 per child
Malaria - $22 per child
Smallpox represents the pinnacle of rolling out a vaccine – continuing to deliver it until the disease itself is eliminated. Prior to the eradication campaign, however, the smallpox vaccine had to be rolled out like any vaccine. This rollout cost $0.10 per child per vaccination in the developing world and $6.50 per vaccination in the US (Fenner, 1988, p1363–1368). Costs for developed world countries outside the US were not available. Adjusting for inflation, this suggests a comparable rollout cost of $0.73 to $47.62 per vaccination.
Rolling out single-dose measles vaccinations costs marginally $1 per child in areas with pre-existing strong vaccine infrastructure, $18-$28 in core areas of low income countries, and $27-$38 in outlying satellite areas that are hard to reach (Bishai, Johns, Lefevre, & Nair, 2010, p3). For one example, Janusz, et. al. (2015) found a cost of $1.60 per dose of the MMR (measles, mumps, and rubella) vaccine in Honduras.
As of 2007, the Brazilian government struck a deal with GSK to pay $7 per dose for a two-dose rotavirus vaccine (Saxenian, 2007, p10), or $14/child. With delivery and other costs, this would be $28/child assuming vaccine dosage costs are 50% of costs (Brenzel, 2015). Later, GAVI negotiated a price of ~$3 per dose (RotaCouncil, 2016, p28), but paying up to $7 per dose in the developing world is still common ( Drug Price Search). For one example, Janusz, et. al. (2015) found $5 per dose in Honduras.
The HPV vaccine is assumed to cost about $48 for developing countries (MSH Drug Price Search), though in 2013, GAVI was able to negotiate $4.50 per dose (UICC, 2013 ; GAVI, 2013), down from the prior price of $13 per dose (GAVI, 2013). However the manufacturing cost of Gardasil for developing countries is currently between $0.48 and $0.59 per dose (Clendinen, Zhang, Warburton, & Light, 2016), which is much lower than the current price per dose, suggesting the price has room to decrease significantly over the future.
These prices and manufacturing costs also do not include delivery costs. Delivery costs between $2.40 and $3.37 per three dose schedule according to one estimate, between $3.10 and $9.21 according to a second estimate, and between $1.11 and $2.74 according to a third estimate (LaMontagne, et. al., 2017).
Based on this, the total cost may range between $2.55 ($0.48 per dose for three doses plus $1.11 to deliver) and $22.71 ($4.50 per dose for three doses plus $9.21 to deliver) per child.
Hecht and Jameson (2011) estimate the HIV vaccine would cost $60-$150 per person to administer. However, as Forsythe (2011), also writing for the Copenhagen Consensus, points out, there may be even more costs to create demand for the vaccine and to manage the supply chain, and suggests that the vaccine could cost up to $465 per person or more. On the other hand, Hecht and Jameson (2011) includes costs to the manufacturer to recoup R&D spending in the vaccination costs, which risks double counting. Adjusting for this could result in a lower cost per person.
The pilot of the malaria vaccine is supposed to roll out to 360,000 children by 2022 (GAVI, 2016, p2). The long-term plan is to set a price per dose of $5 (Ibid., p6) – with four doses per child (Ibid., p1), that would be $20/child. There would be an additional cost of $2 per child (Ibid., p22), for $22/child total.
For the pilot itself, there is an expectation of $101M (Ibid., p2) in non-vaccine costs and $200M in future vaccine development costs (Ibid., p10). At the long-term $22/child rate, vaccinating 360,000 children would cost $7.92M. Thus the other $93M in non-vaccine costs must be transitional costs to set up the pipeline for the vaccine. With $605M in R&D spent, $200M in future R&D, and $93M in transition costs, that is $898M in fixed spending to unlock the ability to roll out the vaccine for $22/child.
In our historical sample, the rollout cost of a vaccine ranged between $0.73 to $160 per child vaccinated, and this is even before taking into account other large sources of variation, such as level of pre-existing infrastructure and remoteness of region. Furthermore, the price, particularly in lower and middle income countries, depends somewhat on the negotiation of large groups, NGOs and governments and as such it’s difficult to generalize these estimate to future prices. Additionally, it’s not clear that our six ranges can be directly compared, given that they were collected at different times in history with different methodologies, taking into account different factors.
Disregarding that and taking the mean of our six ranges produces a mean range of $13.21 to $53.05 and a median range of $2.78 to $33 (though again it’s unclear how meaningful these numbers are). Brenzel (2015) finds that routine immunization costs on average $27–42 per fully immunized child including all costs to deliver all routine vaccines, with significant regional variation, though this would be biased downwards compared to our estimate, as it would not include the more expensive HIV and HPV vaccines.
Thus it is difficult to figure out what it would mean to estimate the “typical” roll-out cost for a new vaccine, should one exist. For example, if we were to roll out the ebola or malaria vaccines widely, how expensive would we expect them to be? Understanding this number, along with the price of research and development, is important for understanding the marginal costs and benefits of new vaccination.
This essay was jointly written by Peter Hurford and Marcus A. Davis.