How beneficial have vaccines been?

By Marcus_A_Davis @ 2018-05-05T00:08 (+10)

This essay was jointly written by Peter Hurford and Marcus A. Davis.

Previously we estimated how expensive it is to research and develop a vaccine and also how expensive it is to roll-out a vaccine. If vaccines are to be cost-effective, we need to realise significant benefits in return for this expenditure. So let’s take a closer look at the specific and tangible benefits that vaccines have shown to provide, both for our health and the economy. 

Simply put, the benefits to society brought about by vaccines can be categorised into two main types, health benefits and economic benefits, illustrated by the following equation:  

What are the health benefits?

We take health benefits to be any benefit from vaccination which results in improved health, whether it be an illness prevented or a death averted. Looking at vaccination in general, Whitney, et al. (2014) estimates that for children born in the  U.S. from 1994-2013 731,000 deaths would be prevented over the course of their lifetimes from all vaccines, and over 500,000 of those prevented deaths would have been due to diptheria. For developing countries, Ozawa, et al. (2017) estimates that for children vaccinated between 2001 and 2020 in 73 GAVI-supported countries, 34 million deaths will be prevented over the course of their lifetimes, and 1,600 million DALYs will be saved, as a direct result of vaccinations (this excludes the impact of the routinely given diphtheria, pertussis, BCG, and first-dose measles vaccines).

Assessing counterfactual impact

To analyze the impact of additional funding for a given vaccine or the introduction of a new vaccine into an existing vaccine regime, you would typically examine the marginal impact those dollars will make due to additional vaccinations above the status quo (see for example, Podewils, et al., 2005 and Blakely, et al., 2014). This takes the existing levels of treatment of a disease as a given.

However, if you are thinking about comparing the entire counterfactual of what would have happened if there were no vaccine, it may make more sense for the initial baseline to be the treatment levels and rates of the disease at the outset when vaccines were introduced. On the other hand, if it happened that a particular vaccination was never developed for a specific disease, over time additional non-vaccine treatments and preventative measures may have been developed to tackle the disease, which haven’t come to exist or, more modestly, the development of some vaccine-alternative treatments may have come about sooner.

Further, the counterfactual of medical research funds not spent on vaccination may be a shift in budget from vaccines to other health measures. Estimating how effective these hypothetical health measures or spending could theoretically have been might be very challenging[1].

Additionally, it is known that individuals with compromised immune systems are more likely to contract infections, like tuberculosis or pneumococcus, which are vaccine-preventable (Holmes, et al., 2003). However, even if these types of disease were to be completely eliminated by vaccines, individuals with impaired immune systems would nonetheless remain susceptible to other types of illnesses that cannot be treated with a vaccination, and so may suffer from those illnesses instead.

Considering the future

Future technologies could possibly decrease the need for vaccination[2]. In the future, if a treatment is developed, preventative or otherwise, that ameliorates or eliminates the harmful effects of a disease, then the impact of the corresponding vaccine would be significantly diminished if not made completely redundant. To account for this, we need to consider the probability that such technologies will be developed , the cost of developing and distributing such treatments, and the timeline in which we expect them to be created.

To estimate this may prove challenging as you need to disentangle the impact of the technologies that have helped to limit the scope of a particular disease, which may only have come into existence as a result of research into vaccines that provided greater conceptual understanding of the disease.

What are the economic benefits?

What counts as an economic benefit from vaccination?

There are a range of non-health economic outcomes that could directly or indirectly be improved by the administration of vaccinations, from changes in the amount of money an individual can save by not having to pay for healthcare, to national macroeconomic improvement due to having a larger and healthier workforce. Bärnighausen, et al. (2014) argues for the scientific community to evaluate the economic impact of vaccinations on the broader end of this spectrum, rather than the current approach that has a  typically “narrow” focus on health care cost savings and the time-saving of patient’s and caretaker’s time[3].

In addition to these “narrow” concerns, broader economic analysis could take into account productivity improvements due to vaccination, behavior changes (like increased education investment), community health benefits (including herd immunity), community economic benefits (local economic and political performance), welfare improvements due to decreased risk, and the direct “utilitarian” value of health gains (the value of the health gains themselves) (Ibid., Table 1).

Evidence for Broad Benefits

Multiple recent reviews have found many of these broader economic benefits, excluding direct health benefits and herd immunity, as relatively under-examined in the literature (Deogaonkar, et al., 2012; Ozawa, et al., 2012; Njau and Cairns, 2016). Njau and Cairns (2016) found that of the 177 studies they reviewed on economic benefits of vaccines published between 2000 and 2016, only 31 (18%) examined broad economic benefits of vaccines, with only 14 (8%) investigating “outcome-related productivity gains, long-term increased government revenues, behavioral-related benefits of vaccines and costs of [vaccine preventable disease] outbreak investigation, and the direct and indirect benefits of vaccines to the health-system.”

Some of these broad economic benefits may not be considered as actual benefits by all interested parties. For instance, behavior change could theoretically include fertility reduction considered as an economic benefit (Roodman, 2014)[4]. How you view the value of such behavior change depends, at least in part, on your values[5].

A meta-analysis of studies that found positive connections between vaccinations and various broad economic outcomes depicts that some of the studies had limited experimental evidence and/or observational evidence that vaccines are associated with improved cognition and education (Jit, et al. 2015). For example, an observational study in the Philippines that found vaccination to improve cognitive performance in verbal reasoning, math, and reading, had a limited sample size (Bloom, Canning, & Shenoy 2012). Jit, et. al., (2015) also found modeling evidence supported short-term macroeconomic improvement and, outside the field of vaccination, several studies showed a positive relationship between improved population health and long-term macroeconomic benefits.

Estimates for Broad Savings

Ozawa, et al. (2017) estimates predict that between 2001 and 2020, the broad economic benefits of vaccination in 73 GAVI-supported countries will be worth between $1.1 trillion and $2.3 trillion, primarily a result of averting long-term losses to productivity by preventing deaths again, this data excludes the impact of the routinely given diphtheria, pertussis, BCG, and first-dose measles vaccines[6]. Over half of this benefit was estimated to be due to Hib, Hep-B, and PCV vaccinations[7].

Ozawa, et al. (2016a) estimates have that over the decade of 2011-2020, vaccinations in low and middle-income countries will have a net economic benefit of 16 times what it costs to supply and deliver them and when accounting for the impact of vaccinations outside 2011-2020, this net benefit rises to 44 times what will be spent. Another study made similar findings, although conceptualised them slightly differently, estimating that the net economic benefits from GAVI’s program in 75 countries between 2005 and 2020 is the equivalent of a 12.4% return on investment in 2005 rising to 18% by 2020, and that’s excluding any medical costs averted (Bloom, Canning, and Weston, 2005).

In the developed world, Ozawa, et al. (2016b) estimates that the yearly economic burden in the U.S. in 2015 alone due to vaccine-preventable diseases was $9 billion, with 80% of that burden due to unvaccinated individuals. Whitney, et al. (2014) estimates that for U.S. children born from 1994-2013 vaccinations will result in a net saving of $295B from direct medical costs, and $1,380 billion from total costs of reducing rates of vaccine-preventable diseases.

Even given concerns about dearth of direct studies of vaccination on specific economic outcomes, these estimates support the argument that vaccines are potentially being significantly undervalued (Bloom, et al., 2017).

Considering Eradication

For diseases that have been eradicated completely, as happened with smallpox, and could potentially happen with polio and other infections in the future, there would be savings from vaccine production and rollout costs in addition to the “narrow” and “broad” economic benefits discussed above. There also would be additional savings from not having to monitor (as much) for the disease[8]. Additionally, costs for any public health campaigns and non-vaccine preventative treatment for the specific disease should also theoretically approach zero. When considering at the outset the total benefits of creating a vaccine, in order to account for the savings that would be made if the vaccine was to fully eradicate the disease, the chance that eradication could actually be achieved must also be estimated

Conversely, even if a vaccine-preventable disease is completely eliminated, as smallpox has been, there still may be some ongoing costs of containment of vaccine stocks, and some continued monitoring for the disease.  Caution should be used when attempting to estimate what happens to costs when a disease is eradicated, as there is currently only a single case of global eradication of a human disease due to vaccination. 

Smallpox

Shortly after eradication there may still be extensive monitoring for a disease (Ježek, Khodakevich, & Wickett, 1987), and some countries may continue to vaccinate against an eradicated disease for at least a couple years post-eradication (Ibid.; WHO, 2010). Fenner, et al. (1988) found that globally from 1967 until 1979, the year after the last case of smallpox, there was an additional $23M a year spent to eliminate smallpox totalling $299M but that in 1967 the yearly costs, economic and otherwise, due to smallpox was many times that amount per year at about $1,500M (Chapter 31, pg 1363-1368)[9].

Polio

There has been a sustained global effort to eradicate polio since 1988, since which time there has been a 99% reduction in incidence (GPEI, 2018). Nevertheless, 30 years later, polio is still endemic in 3 countries—Afghanistan, Nigeria, and Pakistan—and there were outbreaks in two other countries in 2017, the Democratic Republic of the Congo and Syria (Ibid., WHO, 2017a; WHO, 2017b; Zaffran, et al., 2018). Tebbens, et al. (2010) estimated that between 1988 and 2035, the polio eradication program will result in savings somewhere between $45.5 and $56.8 billion (in 2017 dollars, original estimate given in 2008 dollars as $40-50 billion) when compared to the cost of continued routine immunization. It should be noted that this study assumed eradication would be achieved by 2012 and routine immunization would end in 2016, however,it also noted was that the net benefit over this period only decreased 5-6% if the eradication was delayed 3 years.

Prior studies have assumed polio vaccination would have stopped even earlier (for example, Bart, Foulds, & Patriarca, 1996 and Khan & Ehreth, 2003) and thus even though modeling of the potential economic benefits of eradication continues (Thompson, et al., 2015), there will be considerable uncertainty about the ultimate benefits of eradication efforts unless and until polio is actually eradicated.

Others

There is ongoing discussion of whether it is feasible and cost-effective to eradicate other diseases (at least in part) via vaccination including malaria (Liu, et al., 2013; Shretta, Avanceña, & Hatefi 2016), measles (Moss & Strebel, 2011; Levin, et al., 2011; Roberts, 2015), rubella (Plotkin, 2001; Goodson, et al., 2012; Babigumira, Morgan, & Levin, 2013), and HPV (Crosignani, et al., 2013). Additionally, there was a failed campaign to eradicate malaria from 1955-1969 (Nájera, González-Silva, & Alonso, 2011).

Additional Expenditures

As vaccination decreases or ceases due to the elimination or prevention of a disease, money may still be being spent  on reducing the risk of an outbreak due to bioterrorism or from an accident. This general risk of a renewed outbreak in an unvaccinated (or largely unvaccinated) population means that some money and resources will continue to be spent on that disease, even after it has been completely eradicated. Indeed, the risk of a post-eradication smallpox outbreak is a topic that has been long studied by academics (Capps, Vermund, & Johnsen, 1986; Working Group on Civilian Biodefense, 1999; Meltzer, et al., 2001; Ferguson, et al., 2003) and at least some government resources have been spent preparing for such an outbreak (Nature, 2011; New York Times, 2002; Sell and Watson, 2013), and likely more spent analyzing whether such a plan is a worthwhile investment[10].

It’s worth noting that it is not clear if post-eradication spending is as tightly correlated with the real risk from disease as spending on traditional monitoring and direct efforts to combat the disease. Thus, the actual spending on post-eradication preparedness measures may be too high or too low relative to the actual ongoing risk of a new outbreak (Posner, 2008).

On the other hand, if a vaccine for a disease is never created, the specific bioterrorism or accidental risk from that disease may not exist but there would still be ongoing monitoring for the outbreak of a disease (see, for example, pre-vaccine Ebola monitoring in Tambo, et al., 2014). Thus, post-eradication spending on monitoring for a disease outbreak isn’t completely counterfactual.

Assessing the Benefits of Individual Vaccines

To get a better breakdown of vaccine benefits, we return to the vaccines we have examined  in previous articles to highlight as a case study, and we look at the DALY burden prevented by that vaccine over time compared to the number of vaccinations.

An early problem here is that it is hard to know what time period to count as benefits for a vaccine, as each vaccine is in a different stage of rollout (from already having outright eradication of smallpox to not yet even starting to rollout malaria and ebola vaccines). To try to better compare apples to apples, we look at two different hypothetical scenarios for a vaccine: rolling it out to 60% of the relevant population in Sub-saharan Africa (SSA), where most vaccines are targeted[11] and which could be thought of as a realistically achievable yet full-scale rollout of a vaccine, versus hypothetically pushing rollout until the entire disease is eradicated, like smallpox. Eradication is defined as a single roll-out where enough people are vaccinated such that ongoing vaccination campaigns are no longer necessary and the disease no longer comes back for a period of multiple decades (except maybe in a few very isolated cases).

Note that all DALY estimates exclude the potential impact of herd immunity[12] or other non-DALY economic benefits[13], and that we don’t calculate eradication benefits for rotavirus[14] as it is not possible to eradicate that disease.

Also note, the estimates for smallpox and measles are counterfactual estimates we calculated, and hence aren’t about the real burden from these diseases today, while for other vaccines our estimate of the total burden if eradicated is more closely attached to the present DALY burden given these vaccines have only recently been introduced or haven’t yet been introduced.

Fenner, et. al. (1988) estimates that before the intensification of the eradication campaign in 1967, between 1.5 and 2 million people died from smallpox, and between 10 and 15 million contracted the disease (Ibid., Chapter 31, pg 1363). They also cited pre-eradication campaign estimates of $0.83 ($0.10 in 1960) per person per vaccination in developing countries and  approximately $48 ($6.50 in 1960) per person per vaccination in developed countries (Ibid., Chapter 31, pg 1364; Chapter 9, pg 369, 413). At these costs, smallpox was costing approximately $10.4 billion per year and it’s elimination results in a net benefit of ~$8.2 billion per year.

However, while the economic benefit is relatively clear, calculating the DALY benefit from smallpox is difficult, as smallpox was eradicated before DALYs as a concept was invented. Instead, we need to estimate and infer from other sources.

Using these incidence and death estimates, in the early 1960s every year smallpox vaccination was preventing somewhere between 430,000 and 1,000,000 deaths and approximately 13-31M death DALYs in the developed world per year and between 150,000-600,000 deaths and  ~4.5-18M death DALYs in the developing world, given only 20% of the developing world was being vaccinated at the time.

Using the same estimates as Fenner, et. al. (1988) and assuming the same percentages of the global population had continued to acquire and die from smallpox then without eradication, ~714-1,071M people would have acquired smallpox between 1968 and 2016 and ~109-138M people would have died from it during that period[15]. If it assumed that the death rate and percentage of cases would have fallen to 20% of their 1967 rates, those estimates would be ~143-214M cases and ~22-28M deaths over the same period[16].

At 30 DALYs per death[17], that means eradication prevented ~3,288-4,140M DALYs at 100% of the death rate of 1967, and ~658-828M DALYs at 20% of the death rate of 1967[18]. If smallpox still was endemic and modern SSA had a 60% vaccination rate with a case percentage and death rate similar to those of the developed world in 1967, this would imply ~7.4M DALYs would be prevented via vaccination in 2016. Finally, if between 1968 and 2016 the case rate had held steady at what it was in endemic countries in 1967 but the global death rate fell to half of of the death rate of the 1967 rate of ~10%, this would imply eradication benefit estimates of 0.006 DALYs saved per vaccination during the initial rollout and ~44.7M DALYs saved in the year 2016.

Measles Vaccine

To understand the impact of the measles vaccine, it is not sufficient to look at the current DALY burden of measles, as this rate is already the result of over half a century of vaccination lowering the DALY burden. To truly understand the impact of the measles vaccine, we would have to rewind the clock back to the very first rollout and understand the costs and benefits as they happened then, not now.

Prior to widespread vaccination, in the early 1960s, globally there were around 135M cases and 7-8M deaths from measles each year (Clements & Hussey, 2004, Chapter 4). Given a population of ~3 billion at the time (World Bank, 2017), that means ~4.5% of the world population got measles each year, ~0.2% of the total population died each year from it, and the case fatality rate was ~5-6%. Nearly the entirety of the DALY burden of measles comes from death (WHO, 2018), so we will only look at that. Since the measles vaccine is 85% effective at one dose and essentially universally effective with two doses (CDC, 2014, p250), this would roughly imply that widespread vaccination would have averted a measles death for every 510 single-dose vaccinations or every 438 two-dose vaccinations. At 30 DALYs per death[17], this would be 15-17 vaccinations per DALY or ~0.059-0.069 DALYs per vaccinated person during the initial rollout.

In the modern world, the death rate from measles is quite different. In developing countries, the rate ranges from ~0.7-1.3% for the unvaccinated population in India (Murhekar, et al., 2014; Raoot, et al., 2016), to 5-10% in endemic regions of SSA, to as low as 0.1% in developed countries (Moss, 2007; Naniche, 2009). In the U.S., for example, the death rate is 0.2% (CDC, 2017), and measles has little impact on DALY otherwise (WHO, 2016). These figures imply a death prevented from measles per 264-545 single-dose vaccinations or 227-468 per two dose vaccinations in modern SSA with 16-18 vaccinations per DALY, and figures of 2,034-3,894 single-dose and 1,746-3,343 two-dose vaccinations per death prevented with 111-130 vaccinations per DALY for modern India. Using the case rate of the pre-vaccine world and the current death rate in SSA from measles as baselines, if modern SSA had a 60% vaccination rate this would imply ~34.1M DALYs prevented in 2016. Again assuming the case rate of the pre-vaccine world and a death rate of 1% would imply 2016 global estimates of ~0.011 DALYs saved per vaccination and ~99.3M DALYs saved if measles were eradicated.

Stack et al., (2011) estimates (an inflation adjusted) ~$1.1B per year estimate in economic benefits from expanded measles vaccination in GAVI-supported countries from 2011-2020.  However, Ozawa et al. (2017) estimates actual and projected economic benefits from second dose and supplementary vaccinations from measles, excluding the first dose of measles vaccination at a far higher (inflation-adjusted) rate of ~$32B a year from 2001-2020 in 75 GAVI-supported countries.

Rotavirus Vaccine

The rotavirus vaccine is 39-85% effective (RotaCouncil, 2016, p11) and has reduced diarrhea by 30-54% in SSA (Madhi et al., 2010, Msimang, et al., 2013)[19]. As of 2016, diarrhea is responsible for 37.5M DALYs in SSA, down from 53.2M DALYs in 2005 before the rotavirus was introduced[20] (Global Burden of Disease, Results Tool, 2016a). For children under five, the DALY burden was 41.7M DALYs in 2005 and 25.5M DALYs in 2016 (Ibid.).

If we assume we can achieve a 60% vaccination rate of the under-five population in SSA, by vaccinating 96.3M children under five (see Population Pyramid), and reducing diarrhea by 30-54% among those vaccinated, we would avert ~7.5-13M DALYs per year using the 2005 DALY rate or 4.6-8.3M DALYs using the 2016 rate, with a DALYs per vaccination of ~0.180-0.307.

Unlike for measles, Stack, et al., (2011), which analyzed 2011-2020, and Ozawa, et al. ( 2017), which covered 2001-2020, have similar yearly estimated benefits from rotavirus vaccination at ~$4B and ~$3B a year in GAVI-supported countries respectively.

HPV Vaccine

Rolling out the HPV vaccine to all girls aged 5-15 in SSA with a 60% vaccination rate would involve 130M vaccinations (see Population Pyramid). Cervical cancer currently accounts for 7.4M DALYs for all ages, and 3.1M DALYs in women aged 15-49 (Global Burden of Disease, Results Tool, 2016b), with ~1.8M DALYs in all ages and ~780,000 DALYs in ages 15-49 in SSA. HPV vaccines provide 76-98% protection against 70% of cervical cancers (Agosti & Goldie, 2007; Lu et al., 2011)[21]. Given these estimates, the HPV vaccine could prevent ~249,000 to 321,000 DALYs per year in SSA, at a rate of ~0.002 DALYs prevented per vaccination. Were HPV eradicated completely, this would save ~5.2M DALYs per year assuming a 70% reduction in cervical cancer.

HIV Vaccine

If an HIV vaccine comes to exist and is 50% effective (or more), and is given to 60% of the eligible population, it would avert 30% of all HIV[22]. Given that HIV in SSA among those 15-49 years old accounts for 29.5M-35M DALYs (Global Burden of Disease,  Results Tool, 2016c), this would imply the HIV vaccine rollout could save 9.7M using the mean DALY estimate at a rate of ~0.35 DALYs per vaccine. Were HIV to be completely eradicated this would prevent ~57.6M DALYs per year.

Malaria Vaccine

While insecticide-treated nets, anti-malarial medicines, and indoor residual spraying have likely reduced global malaria mortality by 50% over the past fifteen years, they are still unable to drive down malaria in some areas despite good coverage (GAVI, 2016,  p1). Emerging issues about insecticide resistance also challenge these efforts (Ibid.). According to WHO, Malaria still causes over 438,000 deaths per year (Ibid.). Based on this, WHO, GAVI, and other organizations have pushed for the development of a vaccine for malaria.

Alternatively, the Global Burden of Disease estimates show that malaria is responsible for 56.2M DALYs, with 44.3M DALYs for those under five (Global Burden of Disease,  Results Tool, 2016d). Given that the malaria vaccine is currently 39% effective (GAVI, 2016, p1) and assuming a 60% vaccination rate, that would mean vaccinating 96.3M children (see Population Pyramid) would save ~10.3M DALYs at a rate of ~0.406 DALYs per vaccination. Were malaria to be completely eradicated this would prevent ~56.2M DALYs per year at the current DALY burden.

Conclusion

By our calculations, ignoring ebola, the benefits of the vaccines we have analyzed ranges from ~250,000 to 34M DALYs per year in modern SSA or ~9-634 vaccinations per DALY prevented for a vaccine, with averages of 11.6M and 132, respectively.

As mentioned above, there is more limited data from which to estimate the broad economic benefits of vaccination. However, by our estimates, using Fenner et al. (1988), smallpox vaccination had a net benefit of ~$8.2B per year in the early 1960s. This aligns fairly well with the modern academic literature on the broad economic impact of vaccines. For instance, for the bundle of vaccines we examined, Ozawa, et al., 2017 estimates that between 2001 and 2020, the average yearly benefit in developing countries would be ~$8B while Stack et al., 2011 estimates the benefit at ~$3B per year. For the average vaccine in each study (which includes some we don’t examine here), Ozawa, et. al. (2017) had a much higher estimated benefit of ~$17.4B while Stack et al. (2011) estimated a ~$2.3B per year per vaccine.

Making a full accounting of the economic benefit of vaccinations may mean further updating our perception of the total benefits of vaccination.

Our cost-effectiveness analysis of the development of vaccines is in three parts on the EA Forum: Part 1, Part 2, and Part 3. To see all three parts together, you can view the article on our research site.

Thanks to Anna Mulcahy, Katherine Savoie, Joey Savoie, and Baxter Bullock for their help reviewing and editing this piece. 

Endnotes

[1]: Finding real world historical examples where countries do or don’t implement an available vaccine, or where vaccination isn’t given at high rates, could theoretically provide some insight here. However, in the real world if an available vaccine isn’t given in a particular country for a long time there may be factors like anti-vaccination support or a lack of resources that are likely to be reasons for the difference in vaccination. See Gangarosa, et al. (1998) for a historical analysis of neighboring countries that had vastly different vaccine delivery for a prolonged period.

[2]: For example, CRISPR-Cas9 potentially has the ability to do this for certain viruses. See Singh, Braddick, and Dhar (2017) for an overview of CRISPR’s potential in this area and Kaminski, et al. (2016) for a specific application on HIV in human cells.

[3]: Also see Bloom (2015) for a case against the narrow economic analysis of vaccines.

[4]: Driessen (2011) seems to include a similar analysis of this topic but we couldn’t locate the full manuscript. We attempted to contact the author to ask for a copy but never heard back.

[5]: Whether or not you take a reduction in fertility due to vaccination to be a benefit may depend on how that reduction in childbearing impacts a family’s wealth, happiness, desire fulfillment, and whether the children who would have been born would have had net positive lives. However, it may also depend on what you think of population ethics, and in particular whether, all else being equal, you would  prefer there to be more people born now rather than later, and the value of possible people versus actually existing people. For a discussion of these topics see Part Four of Derek Parfit’s Reasons and Persons (Parfit, 1984), and Chapters 4 and 5 of Nick Beckstead’s On the Overwhelming Importance of the Far Future (Beckstead, 2013).

[6]: These figures have been adjusted for inflation to 2017 USD.the original range given in 2010 USD is $560 and $1,200B for 2001-2010 and $420B and $870B for 2011-2020.

[7]: The PCV death estimates were similar in Tasslimi, et al. (2011) for PCV 7, which Sinha et al. (2007) analyzed, but we’ve not attempted to verify that these estimates are within the current consensus. Our purpose here is not the reliability of the death estimates, but the estimated economic impact per death prevented.

[8]: Whether or not a country will attempt to eradicate a disease at all, via vaccination and containment, within its borders will depend on whether or not that particular country views it as a good investment (Sicuri, Evans, & Tediosi, 2015), and may also depend on how they view the likely they view the prospects of other countries eradicating a disease within their borders (Barrett, 2013).

[9]: It’s somewhat unclear what year these dollar figures are in. Chapter 10, for example, makes it clear that 1967 US dollars are being used, but there’s isn’t as clear a statement in Chapter 31, where these figures originate. Assuming a baseline of 1967, total spending on eradication was approximately $2.2B in 2017 dollars while the yearly averted costs were approximately $10.4B. Naturally, the ratio of spending to benefits holds with or without an inflation adjustment.

[10]: For example, in 1997 the U.S. government spent ~$33.5M ($22M inflation unadjusted) for 300,000 doses of smallpox vaccines (New York Times, 2001). Sell and Watson (2013) estimate the US federal government spent approximately $262M on smallpox prevention between 2000 and 2005 but, along with Watson, Watson, and Sell (2017), only $3.5M total since then. Some of this is spent vaccinating against the specific threat of the use of smallpox as a biological weapon in high risk areas (Grabenstein, et al., 2006). Interestingly, in the face of a potential threat from North Korea, South Korea has chosen not to immunize their troops against smallpox but U.S. troops deployed to South Korea are vaccinated (Nuclear Threat Initiative, 2012).

[11]: Note however, that some vaccines may be better targeted in regions other than SSA, like India. For example, malaria is more prevalent in SSA than India (Global Burden of Disease,  Results Tool, 2016f) and the largest burden on rotavirus is in India (Tate, et al., 2016), so this scenario may unfairly penalize analysis of the cost-effectiveness of rotavirus.

[12]: In brief, herd immunity is the phenomenon of protection from infection for those not immune to an infection which is created by having some other portion of the population that is immune. What percentage of the population needs to be vaccinated before the benefits of herd immunity come to exist in a population depends, among other things, on the efficacy of the vaccine, the basic reproduction number of the infection, how random (or not) the interactions of the population are, and selection effects of what portion of the population is vaccinated. For a more thorough view of how herd immunity impacts the effectiveness vaccination campaigns see Fine, Eames, and Heymann (2011).

[13]: Economic benefits were impossible to calculate on a per-vaccine basis as we don't know the number of vaccines given in any of the economic analyses other than for smallpox. None of the economic benefit studies we cite give this information for the developing world.

[14]: Rotavirus is spreadable to humans via nonhuman animals (Steyer, et al., 2008) and thus is unlikely to be eradicated via vaccination (Anderson, 2014), particularly under our definition of a one-time effort. Whether HIV can ever be eradicated via vaccination is unclear, as it too is believed to have originated via interspecies transmission of simian immunodeficiency viruses, the primate relative of HIV (Faria, et al., 2014, Sharp and Hahn, 2011). Thus, discussion of the eradication of HIV should be considered speculative and theoretical.

[15]: The assumption that smallpox would have continued to kill the same percentage of the population as it did in 1967 is unrealistic for several reasons, most prominently among them that the proportion of the world’s population living in smallpox endemic countries in 1967 (those driving the case and fatality rates) did not remain constant from 1968 to 2017. Additionally, medical advances would likely have meant some decline in the death rate from smallpox if there had been no eradication campaign. However, one possible alternative assumption for a “flat” trendline, that the total number of deaths would have remained constant, also seems misguided because the world population has more than doubled between 1968 and 2017. Simply put, we do not make this estimate with great certainty.

[16]: Ehreth (2003) has a fairly different estimate than the figures we use here. In (Table 2) estimates the study estimates 5,000,000 years of life saved (YLS) per year from smallpox eradication. At 35 years per person that’s 142,857 prevented deaths a year (by 1999). The study uses death estimates from times before widespread vaccination (1900-1904 for US, pre-1971 for Africa) but it appears the original deaths per year estimates used here aren't adjusted for increased population as it states the current per year US YLS estimate is “based upon 48,164 average deaths between 1900–1904 multiplied by 35-year life expectancy” and the same method was applied to the African death estimate of 26,659.

[17]: It’s unclear what the correct DALYs per death figure should be. We follow Karnofsky (2007) which estimated that a “life saved” in SSA for a child under five gives them a 50% chance of living another 60 years, or an expected value of 30 years (see more here). However, reverse engineering a DALY/death figure from GBD (2016) suggests 84 DALYs/death, which seems intuitively high.

[18]: Smallpox came in two strains, Variola minor and Variola major. Variola major is significantly worse and carries a case-fatality rate of 20% or more in the unvaccinated, whereas the case-fatality rate of Variola minor is under 1% (Fenner, et. al., 1988, p5-6). Additionally, Variola major appears to have symptoms roughly as bad as malaria for those who contract it but don’t die from it (Ibid., p6). We estimate that about 7% of the DALY burden of Variola major would come from the morbidity and 93% from the mortality, though this would vary depending on how you weight the badness of death[17]. All outbreaks of smallpox in Asia and most of those in Africa were due to Variola major, wheares Variola minor was endemic in some countries of Europe, North America, South America, and many parts of Africa. Given the preponderance and severity of Variola major in Africa, we chose to model Africa as only being affected by Variola major.

[19]: It may not make sense to extrapolate the current efficacy of the rotavirus vaccine (or other very new vaccines like HPV and malaria) as anything other than the initial efficacy. Vaccines are improved and refined in part by being actually implemented and manipulating the timing of the shots, the number of shots, and by learning from impact in a real population. For example, the pneumococcal vaccine, PCV, has gone through several iterations with PCV 7 being replaced by PCV 10, which is being replaced by PCV 13, each of which covers more strains of pneumococcal than the last (see, for example, Miller, et al., 2011). Accounting for this would change our estimate of rollout DALYs prevented since it may be preferable, in such a modeling exercise, to use a longer time horizon and some probability distributions on the potential efficacy of the given vaccine over time.

[20]: Of course, the rotavirus vaccine is not responsible for this entire decline.

[21]: These are heavily simplifying assumptions. HPV vaccines accounting for 70% of cervical cancers does not imply that these cancers are responsible for 70% of the DALY burden.

[22]: 60% vaccination rate at 50% effectiveness implies 30% (60% * 50%) for the overall burden. A vaccine of this efficacy and at this scale would possibly have a far greater proportional impact on new infections (Stover et al., 2007). This estimate assumes high-risk behavior stays roughly the same. This may not be a realistic assumption of real behaviour in response to an HIV vaccine (Newman, et al., 2010). A low-efficacy HIV vaccine given to a mid to low percentage of the population, combined with an increase in risky behavior could actually increase the burden of HIV (Blower, Schwartz, and Mills, 2003).