Episode 78 • 8 September 2024

Jacob Trefethen on Global Health R&D

Leave feedback ↗
Contents

Jacob Trefethen oversees Open Philanthropy’s science and science policy programs. He was a Henry Fellow at Harvard University, and has a B.A. from the University of Cambridge.

He tweets as @jacobtref and blogs about science at blog.jacobtrefethen.com.

(Image) Jacob Trefethen

In this episode we talk about:

Resources

Jacob’s recommendations

Let us know if we missed any resources and we’ll add them.

Transcript

Fin: Hey, this is Hear This Idea. In this episode, I spoke with Jacob Trefethen, who oversees Open Philanthropy’s science and science policy programs. Open Philanthropy is a philanthropic funder that I would describe as being unusually serious about researching and identifying causes where funding is likely to help people the most. In this episode, we spoke about global health research and development.

When I think about philanthropy or charity funding in the context of global health, I think about things like humanitarian aid or direct public health initiatives, such as malaria nets, distributing dietary supplements, or vaccines. However, vaccines and life-saving drugs are health technologies. Often, they’re quite advanced technologies. We all saw how it was only possible to spend money on buying and distributing COVID vaccines once we had demonstrably safe vaccines in the first place. The same goes for smallpox; innovations in delivering, transporting, and storing smallpox vaccines contributed to its eradication, sparing hundreds of millions of lives.

If you care about global health, you might also start to think about ways to support research and development, especially for diseases that mostly affect the world’s poorest people. These diseases see much less R&D spending relative to their health burdens compared to diseases with high demand for treatments. That said, it was not clear to me what global health R&D looks like or how to support it, and that’s exactly what Jacob spends most of his time thinking about.

We talked about prospects for TB and Strep A vaccines, monoclonal antibodies for malaria, and even bugs that stop the spread of malaria. We discussed how much money gets spent on R&D for neglected diseases as a share of all health-related R&D, how a drug or vaccine goes from initial development to approval and distribution, policy tools and reforms for the FDA and the drug approval process that could accelerate drugs for neglected diseases without costing taxpayer money, why there isn’t a kind of GiveWell for global health R&D, lessons from Operation Warp Speed, and how global health R&D might compare in cost-effectiveness terms to direct interventions like malaria nets or frontier R&D aimed at economic growth, among other topics.

So, without further ado, here’s Jacob Trefethen. Jacob Trefethen, thanks for joining me.

Jacob: Thanks for having me.

Fin: So we’re going to be talking about global health R&D. But I guess, first of all, how do you describe what you do? If you’re at a party and someone asks you, what’s the answer?

Jacob: I am a program director at Open Philanthropy, and I work on giving away about $100 million a year with half a dozen colleagues to biomedical science. We have a tilt towards global health R&D, focusing on science and product development that might help people around the world, not just in the U.S. where we’re based. We mostly give away the money of Dustin Moskovitz and Cari Tuna.

Fin: Gotcha. A big reason I wanted to talk to you was this blog you recently started, which began with a post on health technologies that probably won’t exist in five years. Maybe a place to start is TB and TB vaccines. So, what should I know about TB as a disease, first of all?

Jacob: TB is a fascinating disease because there’s so much we don’t know about it, even though it’s such an iconic disease that has co-evolved with us for millennia. It still kills over a million people every year. There aren’t that many things that kill over a million people annually; about 60 million people die each year around the world. TB is doing a huge amount of damage on its own. It’s caused by bacteria that can stay dormant in your body for many years. Most of the time, if you’re infected with TB, it doesn’t cause a problem. It’s contested how many people around the world are latently infected, but it’s probably hundreds of millions. For most of those people, it’s going to be okay. However, if you develop pulmonary disease—disease in the lungs—it can be fatal.

And what else is known is that it’s a disease intimately related to nutrition and immune deficiency. We’re going to talk a lot about probably new technologies in this discussion, but it’s worth discussing nutrition, which is a fundamental aspect of food. There’s a recent study in a state in India called the Ration Study, which provided people with TB disease rations—about 1200 calories of food a day for six months, along with some micronutrients. Everyone in the study who had TB received this, but then their household contacts were randomized into a treatment arm and a control arm. The treatment arm, which included kids living with those who had TB but were not yet infected, received 750 calories of food and some micronutrients for six months as well.

The results were striking: in the treatment arm, there was a 40% reduction in the incidence of TB and a 50% reduction in TB disease, specifically pulmonary disease. That’s significant—it’s like a vaccine in the sense that better nutrition in some low- and middle-income countries could greatly help people. I want to emphasize that before we delve into the technology. It’s also striking what we don’t know about TB.

So, how many people are latently infected? It’s contested, so I have to be careful when I say we don’t know because often that means I don’t know. The things I mention are at least contested among researchers, and I’m not a TB researcher. Katherine Wu at The Atlantic had a really good piece on this maybe nine months ago. The statistic you’ll see if you Google it is that a quarter of people around the world—2 billion people—are latently infected.

Fin: That was what I read on the bus.

Jacob: Yes, that’s probably false. That figure is likely based on extrapolation from tests, and it’s probably hundreds of millions, but the exact number is not really well known.

Subclinical TB is a concept where you don’t have any clinical symptoms. There’s a gradation: before you have a persistent cough, coughing up blood, or symptoms like fever and night sweats, you might not know you have TB, but an X-ray would show that your lungs don’t look great.

Fin: Okay.

Jacob: There’s a contested notion of subclinical TB that’s neither latent nor full-blown TB. What really is it? People are debating how to define that and how much it contributes to transmission. Is almost all spread happening from people you would guess, or are there many people without symptoms contributing to the spread?

With modern tools, we now know a lot more about the function of different genes in the bacteria and the proteins they code for. There are 4,000 genes in that bacteria, and we don’t even know what all human genes do yet. There’s still a lot left to discover about what they’re all doing. Most relevant to vaccines are correlates of protection. We don’t really know what in your immune response, which can be measured from a blood sample or something else, correlates with being protected from TB. If we knew that, vaccine development would be so much easier because we could take a sample and say, “Okay, those antibodies are present, so probably that person will be protected.” We could run many more vaccines through clinical trials.

Fin: Are there other diseases where we have a better picture of those things?

Jacob: Yes, absolutely.

I mean, it varies by disease, and it often helps to have had vaccine trials that you can process the samples from until you know, sorry, a vaccine that works. Yeah. So that you can see, okay, well, in people who were protected.

Fin: And I guess that’s kind of rarely discussed. Like, it’s not very headline-making work to narrow down the uncertainties on these important figures, but it does seem actually very important.

Jacob: Oh, absolutely. This is crucial to the work that we do at Open Philanthropy and what the other funders do when trying to decide how to prioritize. You have to know how bad any given problem is before you can decide which you should prioritize with scarce resources. It’s incredibly dire how poorly tracked and estimated a lot of these figures are. There are a couple of main groups that do their best at tabulating different diseases. The IHME, the Institute for Health Metrics and Evaluation, does a global burden of disease study, which sets itself the ambitious target of giving a number for all of the diseases in all the regions and how many deaths they cause, and how instant they are, and all that. For us, that’s incredibly valuable as a starting point. But also, when you go deep into any given number and how it was generated, it can be very disorienting because you realize how much is based on imputation from one study in recent cases. And there’s other content. It’s really rough stuff. So, you know, the WHO also produces some estimates, so you can triangulate from multiple production methods. But it’s rough out there. Yeah. Mhmm.

Fin: Yeah. I guess what I think you see these numbers, right, for, like, TB infections in Equatorial Guinea last year. You think, like, how do you come to know that? Data collection is actually just really hard. It’s very geographically dispersed. So in some sense, it’s kind of impressive that you can get any resolution on these numbers, but more resolution seems good.

Jacob: Oh, completely. And it’s, you know, some of the other things on the list in that blog post you mentioned, it’s amazing how little detail. So with syphilis, there’s—I might get this wrong, but I think until 2016, there were something like 30 genomes that had ever been sequenced globally.

Fin: Wow.

Jacob: That means that you don’t know what kind of bacteria it is. Does it look different in South Africa than it looks in Brazil? Does it look different in Malawi? Now we have more samples for Treponema pallidum, syphilis in particular, and we know, okay, actually, it’s not that much genomic diversity. Although, I don’t think there are any genomes from South Asia yet. So there are still big gaps. We got lucky there that there wasn’t much, but you could imagine someone who’s trying to work on a syphilis vaccine just making the wrong thing because you didn’t have the samples to know which antigens are present on the surface of the bacteria in which countries. You gotta know that kind of stuff, and that’s the real basic. So, yeah, TB is pretty bad, but it’s even worse.

Neglected diseases and vaccine development

Fin: Yeah. Gotcha. Alright. Let’s stick with TB. So, some uncertainty about deaths per year, but it’s, you know, I guess on the order of the total number of people who have died from COVID in the US or something like this. And you mentioned vaccines. So what’s the story with vaccines? Do we have any at all? Are there any currently in trials? What’s going on?

Jacob: Yes. So we do have a vaccine for infants, BCG, that was invented in 1921. So we’ve had it for over a hundred years, and many listeners probably got it as a child or as an infant. So if you got a scar on your arm, it might be from that. That works well around the world for infants, but it wanes during your life, so it doesn’t protect you for life. And there’s geographic variation in how much it protects you in different regions. One meta-analysis tried to sort of aggregate across all and say, okay.

Maybe protects you on average across everywhere, 17% across the lifetime duration of efficacy.

Fin: So it’s not so hot. We don’t have a vaccine that is known to work in adults or adolescents. The majority of deaths occur in adults and adolescents, and also the majority of transmission probably comes from adults and adolescents. I mean, in part for reasons like adults have stronger chest muscles, so you can cough further, stuff like that. And there is a pipeline of vaccine candidates for adults and adolescents. It’s maybe 20 vaccines in the preclinical and clinical pipeline. We have supported at Open Philanthropy some candidates. The Gates Foundation and Wellcome Trust made a really big grant recently to a leading candidate, M72, to run a massive phase 3 trial of 20,000 people. And that grant was $550,000,000. Wow. One of the biggest grants in the sector, possibly the biggest in the sector of R&D global health R&D. And, I mean, we can maybe talk about why that ends up happening and is so prohibitive to the development of other candidates. But the main orientation I’d give you on TB vaccines is they could have an enormous impact if there was one found to work even at all in adults, but you should not get your hopes up for a COVID-like vaccine, in the sense that none of these are gonna be 90% plus efficacious. It’s a way trickier problem than COVID. I think, you know.

Fin: Why is that?

Jacob: Well, with COVID, you have the spike protein, which, you know, the SARS-CoV-2 virus needs to get into your cells. So if you can just block that, you’re doing pretty well. And so there’s lots of different vaccines that can do that. The issue with TB is it’s so complicated. Its relationship with your immune system is much more complicated. You know, for example, if you or I have TB in our system right now, which do you know that? I don’t know if I do.

Fin: I’m not aware if I have TB. Okay.

Jacob: Well, I’m not aware, but I did go for a drink with one of my exes once, and they announced that they had TB, and we were in close proximity. So you never know.

Fin: How do you find out if it’s latent?

Jacob: You can do different diagnostics to determine that, but usually it doesn’t matter. So you wouldn’t. Yeah. In any case, if it’s in your system, your body usually makes this sort of devil’s bargain with the bug where it surrounds the bacteria with immune cells, forming what’s called a granuloma. That is kind of a truce, maybe rather than a devil’s bargain, where the bug is not gonna replicate and take over your body, but at the same time, your body’s not managed to get rid of it. It sort of hangs out there in this globe, and that’s way trickier to, you know, can you get antibodies through that globe? Right. It’s just a different situation than COVID, I’ll tell you.

Fin: Got it. I guess in any case, it seems like it’s taken a very long time to get even candidates which are now in trials for vaccines that work in adults for TB. Yes. Is there some simple reason why that’s the case? Feels a bit surprising.

Jacob: I think there is a mix of technical and economic reasons for that. So we just talked about some of the technical reasons why it’s not an easy disease to create a vaccine against. I don’t think it’s the hardest. So I think HIV is harder. And there is data in the real world from a proof of concept efficacy trial of at least one vaccine, probably working against TB. So that’s M72. So that’s a start. The monetary economic sort of international drivers are very striking, though, where if TB affected people in the US and UK at the rate that it affects people in India and in South Africa, there would be an adult TB vaccine. If you look at the price that vaccine manufacturers or vaccine producers charge in different countries, in the US and UK and Europe, the price is determined in a different way and is often an order of magnitude or larger per dose than the price that you might sell in lower and middle-income countries. So for, let’s see, for shingles, I think that’s a couple dose regimen.

And each dose is probably what the NHS pays, which is £100 per dose. For HPV, I think that’s about what the US government or US health insurers will pay, which is $280 or $270 a dose. That’s a couple of doses. Whereas if you look at the malaria vaccine that’s just being rolled out in many countries in Sub-Saharan Africa, that’s known as being on the expensive side for a global health vaccine, and it’s $3.90 a dose. So it’s just a totally different situation. The Indian health system would not be able to afford a $280 a dose TB vaccine for the entire adult population. It just wouldn’t work. And if it did work, then we’d have one.

Fin: Gotcha.

Jacob: It wouldn’t work. And if it did work, then we’d have one.

Fin: Okay. Got it. And that is, in some sense, a very obvious and a very big and a very hard to fix problem—the economics. But we will talk about some potential ways to help later. I guess I want to ask as well, is it M72, this Yes. Big shot candidate? You mentioned a very large grant for trials. Why is it so big and what’s it enabling?

Jacob: It’s enabling a particularly large trial in many countries, but the drivers for why it has to be so large are quite interesting. So maybe contrasting with a couple of other things on my vaccine blog list, like Strep A and malaria. Strep A leads to strep throat all the time. Kids, you know, I’ve had it a couple of years ago, and everyone around us gets strep throat. Malaria, in particular regions, has a very high incidence. So if you’re a kid, you might get malaria multiple times in a given season. For those infections, you can design an efficacy trial to determine efficacy much easier. Gotcha. You can enroll, you know, 5,000 people—honestly, fewer. You might be able to do 1,000 for a pharyngitis, a strep throat trial. For TB, happily, TB disease is low incidence in the sense that you’re quite unlikely, in any given region, in any given year, to get lung disease. What an ideal trial would want to look at to determine whether a vaccine works is to see if you give someone a vaccine, will they be less likely to die of TB? There are other things you might also care about, for example, like transmission, but let’s just focus on mortality for now. You know, that is rare enough that no trial is statistically powered to look at that. So you end up looking at, well, what is gonna be, you know, real bad, even though it’s a

Fin: Right. What proxies for the human breast disease? Yeah.

Jacob: And in other fields, other fields get away with dodgy proxies often. You know? In many cancer fields, people debate, well, is tumor growth really a proxy for mortality? Is it? And Alzheimer’s, is, you know, whatever they measure there, is that gonna be really a proxy for what we care about? In TB, we’ve got a pretty good proxy, which is pulmonary disease. So disease of the lung is really dangerous. If you get it, there’s a good chance you will die. But even disease of the lung is quite low incidence. So South Africa is one of the highest incidence countries for pulmonary disease related to HIV infection, where TB can often be opportunistic for people who are infected with HIV. And the incidence in South Africa, across the population, is about—get ready for it—400 per 100,000 person-years cases of pulmonary disease. But 0.4%, it’s not many. So you end up needing to enroll, say, 20,000 people to be really sure. There are different statistical methods for determining how large it has to be for the kind of surety you want. For example, do you just care that it is within the 95% confidence interval on efficacy, it’s more than 0? So you have some effect. Well, that’s easier than if you say it’s gotta be more than 10%. You know, that’s easier if you gotta say it’s gotta be more than 20%. And then most people don’t talk about the lower bound, even though perhaps it’s more useful sometimes to talk about the midpoint estimate. So the phase 2b trial of M72 found that there was about a 50% efficacy midpoint for that vaccine, with a lot of uncertainty around that.

Now maybe I’ll just stop there before I keep rambling because there’s more you could say about what you really demand and who’s demanding what in terms of evidence there. But

Fin: Yeah. I should try sending it back maybe. Yeah. So if you’re doing a clinical trial for a new vaccine, especially in a later phase, phase 2 or 3, where it’s really big and expensive, you need it to be statistically high powered because there’s a high evidential bar to meet that it’s not causing harm and that it’s actually effective before it gets approved. In some diseases where you have a vaccine candidate, it’s relatively less expensive to go into some regions and find participants. In the case of TB, symptoms are rarer than just being affected at all. It’s very dispersed. It’s really hard to get this statistically powered thing without involving a huge number of participants, probably over a very large geographical area, I can imagine, where you’re having to go out to lots of places and administer this trial vaccine. So it’s just logistically a really hard thing to do. Sound right?

Jacob: That sounds right. And you really care about, you know, when you’re screening people on the way in, you’re screening out anyone who currently has TB. Of course. Okay.

Fin: I think I missed that. Yeah.

Jacob: Yeah. So you’re really screwing yourself over there on the statistics. You have to wait for someone to develop TB. TB is a slow-spreading disease. It’s not

Fin: Got it.

Jacob: In the malaria field, you know, you could go outside and get malaria any given day in many regions, and it’s a really risky proposition as a result, and hundreds of millions of people get malaria every year. But with TB, you gotta wait. And, yeah. You know, your geographical variation point is important for other reasons too. Some of them are regulatory and some have to do with engaging policymakers in different affected countries where a country is more likely to trust the findings. So M72 is not just in sub-Saharan Africa; it’s in other countries too. So that makes it logistically harder, but might make the payoff more useful at the end.

Fin: One thing I’m getting from that, and you can tell me if it’s wrong, is it sounds like in general, it is going to be easier to do a clinical trial for a vaccine candidate or just a treatment for a disease where infections are more frequent for a person, but they tend to be less severe on average compared to a disease where, you know, it’s rarer to get infected, but it’s much worse news if you do. Because you just have to wait longer. Yes, if you can just pay attention to the proxies in any case.

Jacob: I believe that’s right, and let me give an asterisk, which is that you have to think that the proxies are, in fact, good proxies. And sometimes the types of people who might—well, let’s switch to Strep A instead of malaria. So pharyngitis, strep throat is usually not a problem. It’s annoying. Sometimes it can cause a huge problem where you can get disseminated invasive infection that causes organ failure because it starts eating your flesh. There are things that can go really wrong. So if you—and really the most common thing that will kill people from Strep A is rheumatic heart disease. If you get repeated infections, there’s a molecular mimicry issue where your body is trying to attack the bacteria that’s infecting you, but in fact starts attacking your own heart valves. That kills maybe 300,000 people every year with rheumatic heart disease. Are you sure that powering a trial against, right, pharyngitis will, in fact, be protecting people who would otherwise get rheumatic heart disease, or is there something different in those people than the people who get protected by your Strep A vaccine? You know, if the Strep A vaccine is 100% effective given the established link between Strep A and rheumatic heart disease, you’re probably good. But there are question marks between the [cases]. [Gap in transcription]

Fin: Maybe we could talk a bit more about Strep A vaccine since you mentioned it. That’s another item on the list of technologies which will likely not exist in 5 years, but tell me more what’s going on with Strep A?

Jacob: Thanks for asking. This is one of my pet obsessions. I sometimes feel crazy because the rest of the world doesn’t talk about Strep A. So hopefully, someone listening to this podcast can tell me why I’m over-fixated on it, and it’s less of a big deal than it currently seems. But, you know, Strep A is a bacteria, group A strep or Streptococcus pyogenes, that’s given us such great gifts as CRISPR Cas9, by the way.

Fin: How sorry?

Jacob: Cas9 is an enzyme from Streptococcus pyogenes.

Fin: Yeah.

Jacob: Yeah. Good to know that. Pretty cool. So it’s given us great gifts. It’s also given us great burdens. It’s very common to get infected globally. The mortality issues come from the two things I mentioned around invasive disease and rheumatic heart disease mostly. Rheumatic heart disease used to be a big problem in the US as well, but with the advent of antibiotics, it has reduced over time. So strep throat is really worth curing. If you might have it, you should get tested and go on antibiotics because it can lead to these longer-term issues with rheumatic acute rheumatic fever and rheumatic heart disease. There’s no Strep A vaccine, and it’s not the easiest. It’s probably harder than COVID, but it is probably easier than TB. You know, it’s the kind of thing where I bash my head against the wall wondering why we don’t have one yet. There are a couple potential drivers for that specific to Strep A, but really, we’re in a nice and positive place right now, I think, where antigens are relatively well characterized. By that, I mean, people have hypotheses for which parts of the bug cause the problems in your body that you might want to attack with a vaccine. There are a few vaccine candidates in the pipeline now, and it’s just a very small scientific community. The people in it are heroes, but there’s just not enough people in it. So I’d love if anyone listening wanted to dabble and learn more about it.

Fin: Yeah. It’s very strange. It just seems, at least if you’re right, a little kind of underrated or under-discussed. Maybe one guess might just be that antibacterials are almost too effective in rich countries. So there’s less demand for a vaccine in the economies that are going to be, you know, demanding the most. Is that right?

Jacob: I think that’s right. And there’s a question for me in high-income countries whether if there was a vaccine, would it be integrated into childhood schedules simply because strep throat is what it is.

It would be, you know, that probably parents would be pretty happy if their kids got a strep vaccine just for that. Or is the mortality threat the thing that would really get things integrated into childhood schedules? There are other things in childhood schedules that really can kill you with more likelihood.

Fin: I was gonna ask where in the world are people dying from Strep A?

Jacob: It seems that after COVID, for some reason, the invasive form of strep is going around at higher prevalence and is killing probably 100 kids a year in Japan, for example. So it’s not at COVID levels of crisis or anything, but it seems to be on the increase, which is quite disturbing. But then if you look at the preponderance of deaths, it’s in lower and middle-income countries. This is a good example of how measurement is really tricky, though. The Global Burden of Disease estimates that maybe 40% of deaths are happening in India. Estimates that maybe 40% of deaths are happening in India, but that doesn’t necessarily match up with clinicians’ work. Cardiologists working on heart disease in India sometimes say, well, I’m not so sure the number is as high as that because in India, antibiotics are prescribed more easily often than in the US. So it’s possible that that number is not quite right.

And then on the flip side, you have practitioners in Uganda who are saying, you know, most people coming to my clinic are showing up with rheumatic heart disease. Are you sure that the numbers aren’t much higher? Right. It’s not well known, I think, where most of the deaths are occurring, apart from the fact that, as you might guess, it is in lower and middle-income countries more than in high-income countries.

Fin: Got it. Same question I asked with TB. Are there already vaccines? Are there vaccine candidates? What’s the vaccine story?

Jacob: There are no approved candidates and there are no candidates near the finish line. There are about eight candidates in mostly preclinical studies right now. That’s in animals and cell lines rather than humans. We have funded some work on one leading candidate to take it into phase one human trials, and we have funded some preclinical work.

Fin: We, as in?

Jacob: Sorry. Open Philanthropy, where I work. And I’m struck by how far away we are when that doesn’t necessarily need to be true. There was an issue for a while that might have contributed to why there are so few candidates, which was that the FDA essentially didn’t allow for testing of Strep A vaccine candidates from the seventies through to 2006. And that was lifted in 2006, but should not have really been on the books as long as it was, in my opinion. And that has sort of led to a chilling of that scientific field. Now, what do we need to do to make progress from here? I think that, honestly, a lot of it is happening, but just not quickly enough, really. So here are some things you can do and probably should do. Epidemiologically, figure out what we are sure about how the strain varies in different locations. Sure enough, that if we take forward one particular vaccine candidate, we’re confident it will work both in Japan and that it will work in Uganda.

You really don’t want to be taking one forward unless that’s true, in my opinion, given where most of the mortality occurs. Number two, do we have a good sense of which antigens are most productive to go after and which combinations of antigens might be useful to pursue? So there’s some debate, for example, in the community of the Strep A research community about whether you should include in a vaccine the group A carbohydrate, which might be protective or might not, or whether you should just go for proteins and target the seven or eight different antigens that people suspect are good from a protein point of view. There’s also debate about whether you should go for the M protein, which might be implicated in autoimmunity issues. Some of those debates you can generate evidence on in mice and non-human primates. Nowadays, it’s more accessible to do so in humans too because some people have set up a human challenge model in Australia, Josh Osoewigee and others, to test. If you purposefully infect someone with Strep and then you cure them to ensure that nothing bad happens, what does their response look like? And if you give them a vaccine, are they less likely to be infected? All that stuff. The toolkit is getting better. I can be optimistic, I guess, but we’re still far away.

Fin: Yep. Okay. But many things to do. Maybe some listeners can get kind of similarly obsessed about it.

Jacob: Oh, gosh. Yeah. Microbiologists who are listening, please get involved. I mean, even just health modelers and economics people, I guess that’s too broad a category. But, you know, I would love to know, and I don’t know, under what conditions would the US government and health insurers in the US pay for a vaccine for all kids in the US? If you had a clearer answer to that, you’d know whether we’re dealing with a global health vaccine in the sense of

Fin: Interesting.

Jacob: Philanthropy has to support it or whether you could make loads of money as a vaccine company making the first or the best vaccine. And I’m still not sure.

Fin: Surprised that’s still up in the air!

Jacob: I mean, totally. There’s some modeling on that question from this group, Sabac, and it is useful to look at that modeling. But, you know, it’s one paper as far as I’ve seen. I haven’t seen all.

Fin: Right. Another paper. There we go.

Okay. I hope we will get to talk about ways to generally speed up these global health R&D processes, like developing vaccines and running them through trials. We’ve spoken about two of these technologies you mentioned in your blog post.

Jacob: Right.

Fin: TB vaccine, strep vaccine. There are eight others.

Jacob: Oh gosh.

Monoclonal antibodies for malaria

Fin: We should not talk about the eight others in as much detail, but I would love to rattle through some of them just to get a sense of the kind of scope of technologies that we could get in global health. One is this idea of monoclonal antibodies for malaria. I don’t know what that means.

Jacob: If you think of the history of medicines and vaccines, the real classic of early 20th-century medicine development was chemists making small chemicals, small molecule drugs, and, as medicines, and then vaccinologists making attenuated versions of a given bug that you can give to someone without causing them much harm and will probably protect them in the future. If you go back to the classical approach, you’re talking about ingesting herbs and plants that contain the small molecules, but let’s get past that. In the last 50 years, we’ve had an opening up of many different new biotechnologies that are not described by those two. There are really two main ones. I’ll give you the spoiler: they’re recombinant DNA and monoclonal antibodies. Those are both biological products in a sense. They’re not something you can easily symbolize, but the gimmick is that you’re using biology to do the work for you.

So with recombinant DNA, you’re saying, “I want this string of DNA that encodes for a human protein.” It’s a human gene of some sort. I’m going to take that splice and put it into some bacterial DNA, or it could be a yeast cell, or it could be anything. You’re going to get that cell’s machinery to produce loads of these proteins. This absolutely opens the aperture for what you can make as a biological product because proteins do a huge variety of different functions in the body and in biology, but they’re very hard to chemically synthesize. We just got this cell to do it for us. Great.

Fin: Yep.

Jacob: Now monoclonal antibodies are sort of the other half of this double punch in the last 50 years. Antibodies are very useful, and lots of your immune system uses them in all sorts of ways. Now we are able to produce antibodies at scale to serve biological functions that small molecule drugs might have failed to do. There are many other new platforms that are very exciting that you’ll hear about in the news, like CAR T cells and mRNA. mRNA is a new biotech platform, but sticking with recombinant DNA and antibodies for now, those are two biggies.

You may ask, and I’m sure what’s on your mind is, have monoclonal antibodies made it as a global health technology?

Fin: I need to know!

Jacob: The people are crying out to know!

And the answer is absolutely not. There is a perception, which is, I guess, accurate so far, that it is still too expensive to produce monoclonal antibodies for various global health applications. Lots of the time, administering a monoclonal antibody will be in a hospital setting with a specialist, and if you don’t have as much health infrastructure in a given country, you’re not going to be able to administer intravenous drugs as often or with the right monitoring. Fair enough. But it’s very striking in malaria where the use case for a preventive monoclonal antibody is really quite enticing, and we don’t have one yet. Let me tell you about that use case. I mean, there are a few of them. We now finally have a malaria vaccine, but it’s not that efficacious. You have to have four doses, which is really prohibitive.

We do have preventive drugs as well, which you give to kids in West Africa and other regions where malaria is prevalent. However, you have to take lots of drugs over the course of many months, and kids don’t really like them, which poses problems. Now imagine if you could take the best antibodies that those vaccines are triggering. You would take the very best, right?

Fin: Right. And then cut it out of the vaccine.

Jacob: Totally. And then, yeah, give that as a dose to someone who might need it. The downside of that versus a vaccine should be obvious: you aren’t prepared; it washes out of your system eventually. You’re not going to be protected for life, whereas the hope with an efficacious vaccine in the future would be that you do get protected for a long time. But there are many benefits. You would have what’s called passive immunity, and you’d have it immediately. So if there’s a malaria outbreak, you could have a vaccination that would take a while to prompt your immune response to protect you. Whereas this…

Fin: You kind of rush out ahead of the frontier and make sure you can distribute.

Jacob: Absolutely. You can imagine a world where kids who currently get preventive drugs that aren’t perfect could, in addition or instead, get a protective antibody that lasts for many months and could take them through a whole season. And it’s only one injection instead of constantly having to take medication, so it could be really wonderful. The economics, in my underinformed opinion, could work out. I think it could get there, but we’re not quite there yet. There is initial evidence from, I believe, a monoclonal antibody given intravenously at a very high dose that does protect at about an 88% level, which is really good—higher than…

Fin: To be clear, than the current leading vaccine.

Jacob: That’s right. And it’s not a fair comparison in a sense; it’s not apples to apples, as the protection is at a really high dose.

Fin: The protection.

Jacob: There’s a question of how long that lasts. In that study, the last thing is a question of if you engineer monoclonal antibodies with a specific goal in mind, how long could you get it to last? I’m not an expert in either of those, so maybe I’ll just not answer, but I think the hope is that you could get something that covers a season. You could get four months out of just one injection. It might also be better for pregnant women because your immune system changes when you’re pregnant, and this would be a way to provide passive immunity during that period.

Fin: Yeah. That was an extremely worthwhile detour. I appreciated the historical context. Okay, so while we’re on the topic of malaria, another technology you mentioned in the same blog post is the idea of bugs that stop malaria from spreading. What does that mean?

Jacob: So malaria is a fun one for a product developer relative to some other infectious diseases. The classic infectious disease toolkit includes treatments, diagnostics, and vaccines. Malaria has a mosquito involved, or a vector, as you might call it, which opens up a whole set of product development avenues where you can go after the mosquito instead of just focusing on the human body. For example, bed nets intercept mosquitoes and have insecticide on them. They’re in some sense a medical product, and in some sense, they’re just something you put around your bed. Chemical vector control is common in several forms. There’s insecticide on bed nets, and you might spray the walls in your hut to cause problems for the female mosquito after she takes a blood meal. There’s this new product class people are experimenting with called attractive targeted sugar baits, where you might put something outside your hut that has wonderful scents to lure a mosquito to their demise. So that’s chemical vector control. The thing I’m talking about with bugs that stop malaria from spreading is biological vector control, which could take a few forms. In dengue, there’s been this really intriguing and successful attempt over the last 20 years. There’s a microorganism called Wolbachia that can intercept the transmission of dengue and malaria. Does that…

Fin: Remind me how dengue spreads?

Jacob: Oh, gosh. I mean, via Aedes mosquitoes.

So it’s another mosquito one. For malaria, there are a couple of attempts to find microorganisms that might interrupt transmission. They’re sometimes called symbionts for the mosquito. We support work at Open Philanthropy on some microorganism research related to malaria, trying to interrupt malaria transmission. There’s this group in Kenya, Isepe, that published a really interesting paper a few years ago, discovering this kind of fungus, Microsporidia MB, which seemed to block transmission of the malaria parasite Plasmodium falciparum. It spreads vertically and horizontally, so if you give birth to mosquito progeny, you might pass it on. If you mate with a mosquito mate, you might pass it on. My colleagues, Chris Somerville and Heather Young, saw that paper and thought, “Well, that’s intriguing. We should follow up to see if this could work as a tool. Will this really inhibit transmission?” They called up the investigators and ended up giving them a couple of really big grants to aggressively pursue this. That’s incidentally one of the cool things about philanthropy; you can pour over what’s the most interesting thing in the literature during a six-month period and then just call people up to try and help them. So, that was fun.

But with bugs that stop malaria, you can also think about the mosquitoes themselves. That technology has already been used in many places around the world, mostly made by this company, Oxitec, but others too, creating gene-edited mosquitoes—green gene-edited mosquitoes that disrupt the normal mating patterns of a mosquito population, potentially crashing the population.

Fin: Is this the same as gene drive?

Jacob: It is not the same as gene drive. Yes. You can just edit a mosquito to have certain properties so that when you release a large number of them, they might disrupt the population. Gene drive is an even more advanced concept that we have been supporting alongside the Gates Foundation investigation. It involves editing a mosquito so that their progeny preferentially inherit whatever trait you edit into them. The idea is that if you edit a mosquito to not be able to carry Plasmodium, for example, that trait gets passed down to the next generation and then to the next generation. After tens of generations, you might have no mosquitoes left that can carry Plasmodium. That could be a very effective tool because you wouldn’t have to constantly do releases in theory if that worked. However, that’s something that’s never been tried in the wild before and comes with many considerations about ecological implications.

Fin: Yep.

Jacob: I guess the irreversibility is both a pro and a con.

Fin: Yes.

Jacob: Well, you would actually be able to reverse it. There are technologies you can try to develop to reverse a gene drive or to have a gene drive that peters out over time. You can tweak some of the different aspects there.

Fin: But yeah. I’ll mention one last technology from your post, which was the idea of multipurpose diagnostic testing. Normally, if I suspect I have COVID, I have a little COVID test I can take.

Jacob: Yes. In fact, I don’t know the current state in England. Is there anything like this in England?

Fin: If you walk into a chemist, you will not see one of these tests on the shelf.

Jacob: Okay, cool. In America, there is a test called Lucira, which I can’t remember if it’s made by Pfizer or another company, which costs $50 and allows you to test for flu and COVID. So that’s two tests in one. It’s not nothing. I ended up with egg on my face regarding this on the list of ten looks. Apparently, in some European countries, there are lateral flow tests that do meet the definition that I provided for this.

So, I have to do a little mea culpa on this. The definition I’m thinking of isn’t very widespread in many countries. I’m thinking of a test that takes about 15 minutes and can be done at home to detect whether you have, for example, COVID, flu, or strep throat. If I have strep throat, I really want to get antibiotics, but for the other two, I don’t. If I have COVID, I might not go to work or I might wait if I’m testing negative. That kind of thing. Not just for respiratory diseases, I think it would be pretty interesting if there was a fever panel or an STI panel.

For example, if you’re in Brazil and you get a fever, when you go to a clinic, you want to know if you have dengue, malaria, or typhoid. Each of those requires different treatments. An STI panel could theoretically be done at home. Why aren’t we in a world where this is already possible? You wouldn’t need a blood sample; you could use a urine sample or a swab to see whether you have syphilis, chlamydia, or gonorrhea. That could increase the number of tests people do and might be pretty useful. So, that’s the vision. I have some ideas why this doesn’t exist more widely than it does, but it’d be nice if it did.

Fin: It sounds like there’s no breakthrough needed. It sounds more like a matter of getting it cheap and combining all these existing tests together. Is that right?

Jacob: I have mixed feelings about what is needed. People with more expertise in diagnostics than I have could convince me either way. I think there’s a mixture of technical problems, regulatory problems, and market problems. For each of these different use cases, that might be a different constellation.

Technically, I should look into this more since I got it wrong, but I believe that the tests in Europe are lateral flow tests. Those can be produced very cheaply. There are many companies in various countries that can produce lateral flow tests cheaply, but they’re often less sensitive than some other diagnostic platforms. You might not be able to get the chemistry right to create a sensitive enough at-home test for cases where you really care about that. There are a couple of companies working on that. I should disclose that Good Ventures Foundation is an investor in Sherlock Biosciences, which is a diagnostics company working on CRISPR diagnostics and other relevant things.

Then there’s the regulatory question of what you’re allowed to sell and what claims you’re allowed to make about what it does. Again, this isn’t really my area of expertise, but in the US, you often have to run a trial that costs a couple of million dollars before selling a diagnostic. That’s obviously prohibitive for smaller companies and smaller diagnostics. The benefit to the public is that they know how the diagnostics work. Some people argue, however, that you should just let a thousand flowers bloom, and you’ll figure out what works based on what ends up selling. Finally, there’s the question of whether there’s enough demand for this product, depending on what it is.

For example, the Lucira flu COVID test costs $50. I bought it once and never again. I’m not going to spend $50 on that investment; it’s a one-off. It’s ridiculous. You feel like such an idiot because you have to throw it away since it’s disposable, but then you feel bad about throwing away the plastic.

Fin: Well, I didn’t know about the at-home flu test. But I guess as long as it’s much more expensive than just buying both of them individually, the sell isn’t quite there yet.

Jacob: That’s a good point. And even, I mean, not to rag on the US, but the last time I got a COVID test, it was still $20 for two tests.

Fin: Really?

Jacob:

It’s crazy.

Fin: That’s very strange. It’s much more expensive than the UK.

Jacob: Maybe I’m not going to the right place.

Fin: So lots of reasons, lots of potential barriers, but it seems totally doable and very useful.

Jacob: Yep.

How global health R&D happens

Fin: Okay. So I’m gonna back up now. We have talked about a bunch of global health technologies which don’t exist and could exist. So Strep A vaccine, TB vaccine, these multipurpose diagnostics, a bunch of things to do with malaria. I would like to talk now about how global health R&D just happens at all. Where does the demand come from? What do the processes look like? How long does it take? So maybe a natural first question is just how much money gets spent on, let’s say, R&D across the world for things you might call neglected diseases?

Jacob: The quickest answer to this is $4,000,000,000 a year. That number comes from G-Finder, which is a report compiled by an organization, Policy Cures Research, that’s been written since 2007. That tries to collate industry, philanthropy, and government spending on neglected diseases related to global poverty. That number is more likely an undercount than an overcount for reasons I can go into if you’re interested. But just running through who contributes that $4,000,000,000: about 40%, according to that report, is from the NIH, the US National Institutes of Health, and the US is providing a big global service by funding science. A lot of that science is relevant to people around the world, not just in the US. That’s the real bedrock that a lot of the rest of what is done in product development is based on.

Fin: Right.

Jacob: The second real standout is the Gates Foundation. So 15 to 20% of that money comes from the Gates Foundation. It tends to be not just focused on science like the NIH, but also on product development. Without that funding, we would not have a lot of the tools from the last 10 years that we do have. Then there are a bunch of other contributors who aren’t at that NIH or Gates level, but that’s the Wellcome Trust, EDCTP in Europe, which funds clinical trials, UNITAID, Open Philanthropy, and then some other smaller funders too. USAID in the US also does some R&D funding. So that’s all on the supply side of R&D funding. Oh, I should say industry contributes some amount, but surprisingly little for a lot of these neglected diseases. Then on the demand side, often these products are selling into public markets, not to individuals or private buyers. At the global level for the lowest-income countries, that often means Gavi for vaccines and the Global Fund for other medical products related to TB, malaria, and HIV.

Fin: Can you say a bit more about Gavi and how it works?

Jacob: Absolutely. Gavi is a very impressive institution set up about 20 years ago, a little bit more, that pays for the purchases of vaccines or pays a slice of the purchases of vaccines for use in lower-income countries. Vaccines are one of the product categories that are really well known to save a lot of children’s lives, especially. There’s been a move over the last 20 years to make sure that if you’re a child born basically anywhere, you’re going to get the main set of vaccines that will protect you, and that’s been a really successful push overall. Gavi is responsible for a lot of the money behind different countries being able to purchase vaccines.

Fin: And that money, so it’s supplementing the public health funders of recipient countries, presumably.

Jacob: Exactly.

Fin: Is that money that Gavi is spending coming from the aid budgets of high-income countries?

Jacob: Yes, exactly. Gotcha. I guess I kind of want to just hear about the nuts and bolts of what drug development looks like for a neglected disease. So, you know, maybe you start off developing some candidate and then you need to run it through trials, which are very expensive and take a long time. What kinds of institutional arrangements are normal? Who often funds different stages? Just like, what should I be picturing in terms of the actual process of getting one of these drugs actually distributed?

There’s a lot of variety depending on the given drug, so I’m hesitant to give you one model that you should picture, but I’ll give you a couple, maybe.

Fin: Great.

Jacob: One thing that’s fairly unified, though not always the case, is that the initial research towards these drugs and vaccines happens in an academic setting. You’ll have a lab focused on microbiology, which is working on how to culture this bacteria, how to make more of it so I can run tests on it, or how to make more of it to look at its genome in-depth. In the pharmaceutical industry, sometimes that happens in companies, but for global health, usually that happens in academia. Once you have something that you think might be a useful product—after you’ve tested it on a plate, in cells, and maybe in mice or non-human primates, or whatever else is relevant to your disease—you start thinking about translation. You begin to consider how to make this into a product that people might be able to use and how to take this through human clinical trials.

Then you get more variety. Some universities do take things through clinical trials, which are often academic-led. However, sometimes you have a company that does that, and sometimes a nonprofit. There are relatively few end-to-end nonprofit drug developers, and in fact, only one comes to mind: MDGH in Australia, who I have a lot of respect for. They are really taking moxidactin and other drugs all the way through to regulatory approval and beyond, sometimes doing studies that are requested after approval. More often, the way our world currently works is that you have companies that handle the sales at the end.

Fin: That’s usually a company. Yep.

Jacob: And you need someone on the hook for if something goes wrong, so you can sue them. You also need someone responsible for actually making and distributing it, and that’s usually a company. So by the end of the day, you’re usually dealing with companies.

Fin: Okay. So I guess even if the end buyers are public buyers like governments and presumably big nonprofit organizations like Gavi, they are typically buying from a for-profit pharmaceutical company because they have the engine to actually manufacture and distribute the drugs, and also you can sue them.

Jacob: Yep. That’s right. It’s a very global industry in the sense that the eventual manufacturers and sellers of a lot of these products are based in India, Indonesia, China, and Brazil. It’s quite rare to have vaccines, for example, made in San Francisco. If you look at where the Gavi vaccines come from, a lot come from the Serum Institute in India, Bharat Biotech in India, and Biological E in India. They are making a lot of the drugs that are saving lives around the world.

Fin: Because they’re cheaper to manufacture elsewhere or?

Jacob: Yep. That’s a lot of it.

Fin: Given that for-profit pharma companies are selling vaccines in volume to Gavi, I would imagine they have some real incentive to do some of the earlier stage drug discovery and development as well in anticipation of that big reward. Is there a reason they’re not doing that more, or are they just in fact doing that much?

Jacob: They’re not doing it more, and there are a couple of reasons. The overarching reason is that while the payoff exists, which is wonderful, it’s not large enough to support enough R&D. If you look at the R&D budget of a Pfizer or a Novartis or a really big multinational pharmaceutical company, we’re talking like $10 billion a year. It’s really very large. It depends on what they’re counting towards R&D, and you probably have a tax incentive to count more or less.

Fin: It’s hard to know what…

Jacob: That really means, but they have really big budgets.

If you’re looking at the total amount of R&D going towards malaria drug development, you’re way lower. Probably around $100,000,000. It’s just so much lower across the whole sector—maybe a little bit more than that, around $100 million. There is a type of institution that steps in to bridge that gap, which I haven’t mentioned yet, called product development partnerships. There are about a dozen of these with budgets usually between $30 million to $100 million a year. They are sometimes designated as responsible for a clinical pipeline of a drug, diagnostic, or vaccine where there is not large pharmaceutical interest. Some pharmaceutical companies may have assets that are useful, but they’re not going to prioritize them or push them ahead because the market’s weak.

So, you get organizations like the International Vaccine Institute (IVI), the Malaria Medicines Venture (MMV), and others like the International AIDS Vaccine Initiative (IAVI), along with Drugs for Neglected Diseases Initiative (DNDi), that support pharmaceutical industries. Sometimes, pharmaceutical companies have their own labs. For example, IAVI will sometimes take full responsibility for the development of the recent typhoid conjugate vaccine. That funding usually comes from rich country government aid budgets, but also from the Gates Foundation.

Fin: Gotcha. So product development partnerships are cases where public or nonprofit funders will kind of knock on the door of a lab, including the lab of a for-profit pharma company, and say, “Hey, do you want to team up?” Or “Can we borrow your equipment effectively to work on this early-stage work?” And then that can become something that they could distribute and sell.

Jacob: Exactly.

Fin: Very cool. If you talk about the FDA, so…

Jacob: In terms of…

Fin: Yeah, in terms of instruments to get some drugs for neglected diseases developed sooner. One of the cooler ideas I’ve heard from your blog is this priority review voucher program. Tell me about how that works.

Jacob: So I think that economists listening might find this interesting. It’s pretty neat. The priority review voucher is something first proposed by David Ridley at Duke, an economist. When was that? Maybe it was 2007? It’s been around for a little bit now.

Fin: Trust me.

Jacob: So there are known market failures in health products, many of which we’ve talked about. There’s neglected tropical diseases where a need in a given country or population may be high, but they’re not able to pay a large amount. There’s also medical countermeasures for pandemics where there’s not stable market demand for a future product that might be helpful in a pandemic, but it would be useful if we had that product ahead of time. And then there are rare pediatric diseases or rare diseases in general where there aren’t many patients, so there’s not a lot of economic demand in that sense, but there are some patients who could benefit from drugs.

The priority review voucher that the FDA has says, “Okay, if you’re in one of those classes of products where there’s a known market failure, we’re going to help you out.” If you get approved for a vaccine against chikungunya—this is a recent one, it’s an infectious disease that’s pretty rare in the US but more common elsewhere—then you are going to get a voucher as soon as you get approved. That voucher, you can either use yourself to speed up the approval process of the next drug you bring to us. So, you know, if you submit a drug application to the FDA, they try to get back to you in 10 months. This track says we’ll get back to you in 6 months.

Gotcha. So that’s a 4-month speed-up. Or you could sell it. Now, if you’re developing a chikungunya vaccine, you might not be Pfizer. You can sell it to Pfizer, who also has all these other products that they might want to sell to the US market. They will pay a lot for a 4-month speed-up of that review because that means they might get, if they get approved, 4 extra months of patent life on whatever their other drug is. So by making them tradable, you can get money for them.

And that means that when they were first released, one of them sold for over $200,000,000, maybe $290,000,000. They’re now trading at about $100,000,000 for this voucher. That gives you an incentive to develop drugs that fit into these neglected categories.

Fin: Very cool. One reason I like this, if I’m understanding it right, is that it kind of feels like free money. Taxpayers are not paying for this scheme.

Jacob: That’s correct.

Fin: The FDA has a queue of things to review and approve or not approve. They can move around the order of the queue without paying more or less money. When you can trade vouchers to be brought earlier in the queue, why not use that effectively as the money you raise from giving that option to these products where market failures exist, including for neglected diseases?

Jacob: Yes, I think it is very neat. I could talk your ear off about ways it’s not perfect, but I don’t know if you want to get into that.

Fin: Well, if there’s a brief… Well, let’s see.

Jacob: A couple of ways it’s not perfect are that there’s no obligation, once you get the voucher, to actually sell that drug in the countries that need it. It’s meant to incentivize approvals, but those are FDA approvals. There are plenty of cases now, or at least a few cases, where someone has gotten approval. You fast forward 10 years, and are people in the countries affected by what that’s meant to fight actually getting that drug? Not often, not all the time. That’s not perfect. You know, not everything’s perfect. There’s also an incentive for pharma to game the system; if you create a new category of disease, like breaking up some disease into sub-disease 1 and sub-disease 2, and you’re the first in sub-disease 2, then you’re going to get $100,000,000 for something that you know. So it’s a little bit gamable. Yeah, but it’s definitely good. If I were an academic or a small biotech, I would be looking at it. It is a way to work on more impactful products that you might actually get paid for in the end. It’s wonderful.

Improving how the FDA approves life-saving medicine

Fin: Okay. Speaking of the FDA, I don’t really understand exactly how it works, but what I’m picturing for a typical drug review process is that it has some drug that’s being asked to approve or not approve, and it’ll look to clinical trials within the US as evidence of whether to approve. When we’re talking about neglected diseases, I’m imagining that most of the clinical trials are outside of the US, right? Because that’s where most infections are. How does that work? Can the FDA use the results of those trials from outside of the US to then approve some drug domestically? One reason I ask is that I’m imagining that FDA approval is a really big boost for a drug in terms of how likely it is that other governments approve it in the countries that matter.

Jacob: The FDA often looks at out-of-country data for all sorts of drug approvals. In fact, the latest report I read on inspections of clinical research indicates that the FDA sends someone to interview the scientist responsible for a given clinical trial. Apparently, two-thirds of those scientists are now in non-US countries. That doesn’t mean that two-thirds of the data is coming from non-US countries, but the FDA definitely has the ability to accept data from outside the US and often does. The mandate of the FDA is to figure out what is good for Americans. That means they are not, in terms of their mandate, interested in trying to regulate other countries, which is a good thing. Sometimes that means that neglected tropical diseases are in a tricky middle zone. Sometimes, though, if you take cholera, cholera is a disease that, happily, we do not have much of in the US. There’s a vaccine that recently got approved for US travelers called Vaxxtera, and that was based on a bunch of different data, but some of the contributing data was from a human challenge model where you purposely infect people to see if the vaccine works.

And I forget where that trial took place. I wouldn’t be surprised if it was in the US because, you know, you can collect that kind of data in the US. If you look at an example of a neglected trouble, the chikungunya vaccine, it’s probably a bad example. But, basically, sometimes you’re going to be looking at data where it’s not as robust as if you had a trial in the US, but it’s not possible to run a trial in the US. So the evidence is better than nothing. And, you know, it’s over some safety threshold. Whatever vaccine you’re talking about is going to be safe enough, and it’s going to help probably Americans who want it too, even if the trials are elsewhere. You can’t be as sure as if the trials were in the US.

Fin: My poorly disguised intention with asking that question, I guess, was like, should the FDA pay more attention to these out-of-country studies?

Jacob: Yes. I believe in my heart the answer to this is yes. I would love if I had more on-the-ground experience to be sure of this because we do already look at a lot of out-of-country data. The reason though I think it’s just gotta be a yes is that people do vary in how they respond to different drugs, devices, diagnostics, all of that. But we, as humans, don’t vary that much in the grand scheme of things. The reason that clinical trials work at all is that you have a subsample of the human species in a trial, and you’re going to extrapolate from that trial and generalize. There are a lot of questions for any given generalization about how sure you can be that it’s going to generalize. But most of the time, we find a way to generalize from the US population to other parts of the US population. I think we could do more generalizing from high-quality clinical trial sites in other countries. That would help people in those countries too, I think, because often, as a company, you might not market a drug in a country that was not involved in a clinical trial. You might not have that oomph. Whereas if that country did have medical research contributing to that result, you’re much more likely to market that drug. So I think it can be more of a win-win if the FDA looked at data more often from other countries. That said, it already does some.

Is global health R&D as cost-effective as bednets?

Fin: I want to put on my kind of EA class effectiveness hat for a minute. And you’re going to skew on me. That’s right, time has come. Let’s say I’m a philanthropist and I want to save lives with my philanthropic dollar. One thing I can do is give money to direct public health interventions, like handing out bed nets, like you mentioned, distributing medicine, even the vaccines we’ve talked about. Another thing I can do is fund research and development of drugs or vaccines for neglected diseases. I think historically when I thought about global health, I’ve mostly thought about those direct public health interventions. But how should I, how might one compare the cost-effectiveness in terms of just saving lives between the really direct, really well-evidenced things and R&D, whatever parts of R&D are most promising cost-effectiveness-wise?

Jacob: That’s the kind of very difficult comparison answer that we have come to. We expanded our global health R&D grant making in 2023 because we thought there were opportunities at the margin that would help people more per dollar than giving that same amount of money to cost-effective public health interventions. So we currently believe there are things that we’re not doing that would be more cost-effective, with the following asterisks.

Fin: Wait, sorry. You currently believe there are things you’re not doing that will be more cost-effective than…

Jacob: Than the margin of the most cost-effective public health interventions.

Fin: Gotcha. And that stuff is R&D?

Jacob: Yes. Sorry. Stuff within global health R&D.

Fin: R&D.

Jacob: Absolutely. Yep. There are driving assumptions there that are really meaty that even different donors might think of differently. So one is around risk. One is around duration, or rather timeline. With the work that we do on the science team, a lot of it will not lead to any new product.

You know, we’ll look back in 20 years and say, “Oh, well, that one generated some knowledge in a couple of scientific papers, but nothing came out of it that directly impacted global health.” That’s most of the grant making we do. In addition, there’s the length of time between the activities we’re funding and actual health impact. We fund some things that we don’t expect to have meaningful health impact in the next 20 years. Science takes a really long time to actually play out, and plenty of what we fund is more in a sort of 10 to 15-year window where we’re hoping it’ll have an impact.

So, under various assumptions that effective altruists discuss a lot regarding what’s going to happen in the future, the stuff that we’re doing looks more debatable the more you think the future might radically change in the next 10 years. We won’t get around to having an actual effect on people’s lives in the next 10 years on our team. Whereas, if you donate to distribute more bed nets or seasonal malaria prevention drugs, that’s probably going to affect children in the next 2 years.

So, you put all those assumptions into a soup, and we debate all the different values those assumptions could take. Then we look at what opportunities we are in fact finding within each given area. We are finding ways to fund academic scientists and ways to fund nonprofits developing drugs and vaccines that seem quantitatively to be a higher return on investment to us currently at this margin.

Now, how true would that be? We give away about $100,000,000 a year in global health R&D and biomedical science, roughly. It’s been more in 7 years, less than us. Would that be true at $1,000,000,000 a year? Would it be true at $200,000,000 a year? Those are all different questions about different margins, and I’ve got my own takes on all of those, but I don’t know because we’ve never tried at that scale. So, it gets a bit complicated.

Fin: Gotcha. I wonder if one reason that there are apparently many great opportunities in global health R&D is that it’s just harder to find them. I’m thinking about for-profit investing. There are certain kinds of investments that are hard to make just as a small retail investor. You need to be some kind of accredited venture capitalist to access opportunities and also to find them in the first place. That might be a reason to expect that the returns of that kind of VC investment are high. Does that sound roughly right as one of the explanations for why there are, in fact, very good opportunities in R&D?

Jacob: I don’t believe the analogy is tight, but I do agree with the conclusion. I think the analogy is not tight between VC and philanthropy in R&D because once you are a VC, the best deals are sometimes obvious.

Fin: Right.

Jacob: It’s actually not as intellectually thrilling as some people think before they get into it. Just going off work friends telling me. And that can, to some extent, be true for people who are in philanthropy too. But I think there is inherently more debate about what is the most impactful way to fund science and what kinds of science are relatively underfunded and neglected. I think there’s more meat on that bone. I do agree with the punchline that it is harder for a retail donor to know where to give to high-impact science or high-impact global health R&D.

If you look at our prototypical grant, it will be $1,000,000 or $2,000,000 over 3 years to an academic group based somewhere around the world. They might be in Cincinnati, they might be in Oxford, or they might be in Kenya. That type of grant does require perspective on the scientific literature and on what is interesting and underfunded within it, which is not possible to gain without having done some work beforehand.

Well, just before I close on that point, I can tell you want to ask something, but I won’t let you because I will emphasize that I think generalists are able to make high-impact scientific grants. If you look at how we are staffed, for example… Mhmm.

You know, I work with half a dozen people. This should have been my answer up top. What do you do? You don’t often tell what someone does by who they chat to most, and here’s who I chat to most: Chris Somerville, Heather Youngs, Alex Bowles, Ray Kennedy, Catherine Collins, and Matt Clancy. And, you know, a bunch of other people open, Phil, but those are the half dozen people I talk to most. Their backgrounds include biochemistry professors, malariologists, global health practitioners, and in Matt Clancy’s case, an economist of science and innovation. There’s a lot of generalism in that group. It’s not like everyone has spent, you know, for example, we don’t have a TB expert. We don’t have someone who’s spent their career in TB or someone who’s worked at a TB lab for 20 years, but we do have people who are very comfortable assessing the quality of evidence in a TB paper. You know? So, you don’t have to necessarily be 20 years deep in something, but I do think you have to have a bit more time set aside and skill and scientific background than a retail donor. That’s right.

Fin: The thing that comes to mind is, again, as a retail donor, I want to know where to spend, you know, $500 on charity this year. It’s also really hard as an individual, as a generous individual, to know what the most cost-effective nonprofits are doing in public health work as well as R&D. But in that case, things like GiveWell exist, where they centralize and institutionalize that research, and then they can disseminate it. I can just go and check their website and learn what to give. So the question that comes to mind is: why is there not a, you know, GiveWell for R&D where I can go and figure out, you know, which labs are doing the best work and then go and wire them $500?

Jacob: I don’t think it would be impossible to create a GiveWell for global health R&D. I think that it would not satisfy the average, well, I don’t know the average GiveWell donor, but it would not satisfy a lot of donors in the same way. So you might end up appealing to a different donor set, and you might end up making different design choices as an organization as a result. One striking difference is the relatively larger number of grantees in our science database, which you can check online. All our grants are public compared to top recommended GiveWell charities. There are a couple of reasons I think that there’s a difference there. But I’m almost curious for your answer as someone who looks at GiveWell and other donation opportunities. You know, why are there, what is it, four top recommended charities, or why are there relatively fewer top recommended charities? And should we be trying to get to a world in science where there are some standout scalable nonprofits like that? What’s going on here?

Fin: So I guess if you imagine that there is a bunch of candidate grantees or organizations that can absorb money, and they vary in cost-effectiveness on the margin. I’m imagining that in the case of R&D, the ordering does move around a bunch because the marginal effectiveness changes. Because science is this dynamic process, right? You just don’t keep doing the exact same kind of science.

Jacob: Yeah. I think another way to say what you just said is that we’re not changing something in the domain of knowledge or technologies like you’re trying to do with scientific ground-making. You’re creating knowledge that, once it’s created or discovered, will be around forever. And you’re creating technologies that hopefully will be around for a long time. Of course, that’s not inevitable, and that’s a different structure than an annual distribution of a health commodity, which has benefits in that year or in the coming few years. One other thing that strikes me as I think about this is that there are inherent scalability limits to academic labs that don’t apply to larger goods and services nonprofits or to clinical trials. You know, I mean, in clinical trials, you can really spend tens of millions of dollars or even 100 quickly if you want to generate evidence. But within academic labs, you’re often limited by the time of the principal investigator and how many students and postdocs they can take on. Yep.

And, you know, there are some exceptions where someone gets to take on a huge number of people. So, you know, George Church and David Baker probably have hundreds of people they collaborate with, in some sense, who are their students in the labs. But usually, you’re dealing with single digits or maybe low double digits of people working around a specific set of problems or tools in the lab. So that means you might get stronger diminishing returns in individual academic labs.

Fin: One thing that comes to mind is that retail donors may have a preference for really reliable outcomes. I like to be confident that I’m actually helping people. I’m not spending money on this lottery, which could maybe pay off, but probably won’t. One thing you could do about that is create a fund and agglomerate lots of grants to derisk it that way. I guess you still have the problem that there’s a long delay from my donation to any outcome that could be attributed to that donation. Because you just said that science takes a really long time, so that can be part of it. And it’s quite hard to run a retail fund compared to other kinds of funding.

Jacob: Yeah. Running a retail fund sounds horrible. There are so many more people who can get mad at you. Yeah. Only a couple of people can get mad at me. Yeah. It’s a problem in philanthropy. You don’t get enough external input. I love when people post on the effective altruism forum, and you have actual good detailed critiques of open philanthropy because that gives you something. But, I mean, most philanthropists don’t even get that because they don’t have a community around them.

Fin: So totally.

Jacob: It’s rough out there.

Global health R&D from a Progress Studies perspective

Fin: I guess I want to take off my EA hat and put on my progress studies hat. I want to know how you think about comparing the economic returns to global health R&D writ large, compared to what I might call frontier R&D, like really high technology. “Sexy” inventions—Streppa is not sexy to you. I take it back. I take it back.

Jacob: Come on. Streppa has got a kind of unconventional profile, right?

Fin: Okay, right. So, you know, if you were king of the world and could shift spending on global health R&D, you could maybe take some away from high technology. How do you think about how these two things stack up in terms of making the world economy better off?

Jacob: So you’ve framed it as if you could do one and not the other. I think one of the really striking things about research and development across the economy is how undersupplied it is in general relative to health benefit and economic benefit. It’s a very common phenomenon that happens with, you know, if I look at my economist friends, something that often occurs is they get into economics because they want to understand how the social world works and use some analytical tools for it. Okay. Then they try not to have too many normative beliefs about what the world should look like. But one of the beliefs they sometimes allow themselves is that GDP going up is probably better than GDP going down. Then, if you look at what drives GDP going up, it turns out that over the long run, it’s R&D. I mean, not exclusively, but largely, R&D. And so you end up with this funny phenomenon socially where all the economists then start getting really interested in technology development and science kind of from the back. I find this quite socially striking because it’s almost like, oh, it’s too late for you now. You’re in the wrong community. But I think it’s correct economically that frontier R&D is undersupplied, and global health R&D is undersupplied. Okay, that’s my dispute number one. Do you want to…

Fin: Just quickly say why that is in the most general terms?

Jacob: Oh, yeah. Totally. Totally. So the standard stories you all hear using economic language are that there are large positive spillovers from R&D that can’t get captured or don’t get captured by the inventor.

So you might invent, I mean, let’s take my colleague, Catherine Collins. When she was a PhD student, she invented the R21 malaria vaccine, which is now being rolled out 13 or 14 years later.

Fin: Yep. So, decently successful.

Jacob: You know, we all had successful PhDs.

And, you know, in that PhD, she got the antigen for malaria to express more frequently on this little particle so that it might end up being an effective vaccine. And how much is she able to capture the value of that invention? Well, even if it were a product that is useful in the rich world, there’s a long enough delay there, and there are enough intermediaries involved along the way that the inventor is not going to capture that much. You know, you look at Kathleen Carico, who worked early on mRNA. How much of the enormous economic benefits, which probably measure in the trillions from COVID vaccines, did she capture? I assume a very small portion. So, that’s a common phenomenon.

Fin: That feels to me like the big headline here as an explanation of why R&D in general is undersupplied. It’s this kind of spillover. It’s very hard to internalize all of the enormous benefits of some, especially in kind of fundamental science. I think that answers my question. Okay. But so you’re disputing my original question, which is how do you compare Frontier and Global Health R&D? Okay. I buy that both of them are under the radar, and they’re not trading off against each other, at least not strictly, but all the same. So, you know, a marginal dollar to high technology R&D compared to global health. Which one just goes further in terms of making people better off and making the world richer?

Jacob: We have tried to look in quite a lot of depth at Open Philanthropy at this question and at this level of abstraction, which I think is quite difficult because it’s so abstract. In particular, my colleague Tom Davidson wrote a report. My colleague, Matt Clancy Moore, recently also wrote a report on the returns to science in the presence of technological risk. What Tom did was adapt a paper structure from Larry Summers and Ben Jones that looked at the social returns of R&D. What Matt did used a sort of different model structure but was also looking at science in particular. What would you expect the economic returns and health returns to be from that over the long run? The quantitative punchline from both of their reports is that R&D as a sector of the economy looks like it generates very high social returns that we in our quantifications would call 50 or 100 times better than just giving cash to someone at the average income in the US, which at the time we set the benchmark was about $50,000 a year. Yep. So in our incredibly zoomed-out quantifications, contributing to R&D per dollar looks 50 or 100 times better than that already decently good thing of just giving someone some money. And that’s astonishing. So that’s really very large. And it’s so large that it makes you wonder, can this really be right? This is something that we’re saying is 2% of the economy, you know, hundreds of billions a year around the world is really contributing a huge social return over the long run. If you run the same calculations 200 years ago, the returns according to Tom’s model would be even larger because you’re getting at some of the most valuable technologies that have now been invested in, and there’s a diminishing returns sort of dynamic going on. If you run his numbers, 200 years ago, it would have actually been more effective to invest in R&D at the frontier.

Fin: Mhmm. In terms of social returns?

Jacob: In terms of social returns. Absolutely. So if you had, you know, whoever invented the cotton gin or whoever… Yeah. Yeah. Yeah.

All those inert technologies that would have, over the long run, transitioned us from a society where we were mostly employed in agriculture to a society where you and I can talk in front of these microphones. You probably got to my apartment where we’re recording this on—well, actually, you didn’t take an Uber; you took public transport. It’s better.

Fin: I took a 60-year-old technology!

Jacob: But, you know, the listener of this podcast is probably listening on a phone, and maybe they’re using AirPods. If that’s you, then hi. We’ve reached this state, and we wouldn’t have been here 200 years ago. Actually, per dollar or per pound back then, the frontier R&D would have looked more effective than just direct charity. So I think that is a totally possible conclusion. That is not the conclusion I would reach now as a punchline. I want to be clear, though, that when we give to global health R&D, the driving quantification there is not economic returns. So you asked me a question about economic returns, and that’s not the main reason global R&D looks so good.

Fin: Which was, to be clear, totally tilting the scale in favor of R&D, which produces products that high-income countries buy because they’re spending more. So the dollar amount comes out as higher. But spending consumption doesn’t necessarily mean true social benefit. Indeed.

Jacob: I would really probe for people who think that frontier R&D is where they should tilt their energies and attention. That could well be the right conclusion for them, but I think it is worth introspecting where that intuition comes from. Something I’ve found valuable for myself is that a lot of the time when people think about the benefits of frontier energy, they’re thinking about productivity benefits that lead to, over the long run, a fantastic, glorious, large economy, etc., where you can do anything. You don’t have to do things you don’t want to, blah blah blah. That’s not what most biotech, even frontier biotech, is going to help with that much. A lot of frontier biotech in the US, for instance, is working on health problems like heart disease and cancer that disproportionately benefit older people, and a lot of those people aren’t in the workforce. So you’re not going to be affecting productivity that much.

Fin: That’s interesting.

Jacob: That’s not to say it’s a bad thing. I think that’s something to scrutinize. It’s like, do I not care about preventing cancer in people outside the workforce? Well, probably you do care about preventing cancer in people outside the workforce. As soon as you start caring about preventing cancer in people outside the workforce, the opportunity to do a huge amount of good in global health R&D starts rearing its head more substantially.

Fin: I see.

Jacob: It really is the case that some of the problems we’ve talked about in this discussion, you might actually get to a solution if you spend some of your career working on them. Whereas it’s a horrible situation now for a lot of people working in biotech where they don’t know. They could spend a whole career and not actually work on something that gets approved. A lot of these areas involve pouring over the same target, and you just might not succeed because many people have tried a lot already. Whereas with strep A, I’m pretty sure we can get a vaccine, and it was so fun to work on that together. So, anyway, I think it’s more of a personal aside there, but that’s my answer.

Fin: Yeah. That’s super interesting. I guess, in some way, it’s a bit unfair to kind of gerrymander all of R&D this way and to talk about this one big blob of R&D that I’m calling frontier or high technology and another blob called global health. There’s a lot of things you can research and develop in the world.

Jacob: Yes. Yes. And a lot of ways you can categorize them.

But I hadn’t considered this point that much of what we would call frontier R&D is not this kind of fantastic productivity boosting, you know, inventing new things like the Internet or iPhones. It’s often not economic benefits; it’s health benefits for people in rich countries. Absolutely. And that’s good, of course.

Jacob: That’s great. But it…

Fin: doesn’t mean that the question boils down to how we should compare economic growth to health benefits in low and middle-income countries. Yeah, which is a very nice framing.

Jacob: For those productivity lovers out there, I love productivity. I get up, I start typing, I start reading, I start making calls. And I just want to make sure that that’s on the record. I love it.

Fin: We wouldn’t want people to think…

Jacob: Oh, it’s one of my productivity sets.

Learning from Operation Warp Speed

Fin: Okay. I am going to step back again. So we’ve talked some about global health technologies, like particular vaccines that we could develop or that could be developed. We’ve talked a bit about how global health R&D works in general. There are terms to global health R&D. It might be worth getting a bit more opinionated about ways to speed this stuff up because it’s really important and it saves lots of lives. And luckily, you also have a blog post about policy ideas for accelerating global health R&D. I’m going to ask about those ideas, but the first thing I want to ask about is Operation Warp Speed. So, during COVID, the US launched a campaign to develop vaccine candidates and eventually distribute vaccines for COVID. It got a marketable COVID vaccine within about a year, ultimately, I guess, two. You’ve been talking about vaccines, which from candidate through trials to approval and distribution can take more than a decade. Doesn’t Operation Warp Speed just show that if you really care, you can make this happen ten times faster? Are there really lessons there, or is it just a very different case?

Jacob: Very different case with real lessons. I mean, huge success, Operation Warp Speed. And I think it does show if you have political will, if you have creative contracting, if you have a lot of demand from society for a particular scientific or technical outcome, you can get there more quickly. You can get there with more redundancy, where you throw money at a problem, where if the first, second, or third way you might try to solve it don’t work out, you have a fourth way. And that’s wonderful and applies in part, I think, to a lot of these problems. Of course, you can’t drag it across exactly. We were in a moment where everyone was focused on COVID, where a lot of the companies that made the vaccines, a lot of the staff there viewed this as their life’s work, as their moment. This is not something you can generate out of nowhere for hepatitis C vaccines or for a lot of the other problems on the list, but you can do a lot of the same tools. So you can, for example, use creative contracting to assure inventors and manufacturers that there is demand at the end.

This is something that’s been discussed a lot by Nan Ranzahoff at Stripe Frontier. She wrote a great piece on advanced market commitments in works in progress recently that I highly recommend to anyone interested in this topic. Also covered by Rachel Glennerster at CGD and the University of Chicago Market Shaping Accelerator. There’s definitely more that could be done there, and that’s mostly on governments. That’s basically on the demand side, just making demand clearer and making demand larger in some cases. There’s also stuff you can do just strategically. One thing that Operation Warp Speed did was take uncorrelated shots on goal. So you take Strep A. One thing we could do is look at the different technological platforms that you could make a Strep A vaccine on—so mRNA, virus-like particles, just standard protein vaccine—and take a bunch of shots that are their successes relatively uncorrelated on those platforms. Then also look at the antigens you would go after. You can go after SLO, the carbohydrate, the spy, pepper, spice—I forget what they all get called these days—but, you know, go after combinations of those, try and get uncorrelated risk there. You can also go after different adjuvants that get used to boost the immune system.

So you can definitely steal a lot of the ideas there about uncorrelated technical risk. You’re more likely to achieve a success at the end, even if you don’t know which success. That’s the kind of thing you can invest in for high social value problems that aren’t just bottlenecked on scientific understanding.

Fin: Gotcha. One of the many things you mentioned was advanced market commitments, AMCs. So, embarrassingly, I don’t actually remember how Operation Warp Speed works, but I gather that public health agencies in the US committed to buying vaccines if they were developed. That gave the pharma companies a very obvious incentive, or at least it gave them more confidence that developing these vaccines would indeed be profitable. You’re suggesting, I guess, we could do something similar—just have a buyer who commits to buying something on spec. Who would the buyers be there? Could it just be governments? Could it be some kind of nonprofit funder?

Jacob: I think nonprofit funders tend not to have enough money for that sort of thing. I say that in a defeatist way. There are people in the world, because we live in such a strange world, who do have $100 billion. Of course, there are. It’s a lot of money. But I think mostly it’s a government thing. If you look at successful versions of this in the past, it’s not just Operation Warp Speed. The GAVI administered one of these for the pneumococcal vaccine 15 to 20 years ago, and that had five donor countries or maybe four, including the Gates Foundation. You can get these things together either as a group of countries or one country if there is something that affects people in the US. The US government could easily do more of this. The US government is a rather big place. I’m not a DC person, so I don’t know the exact administrative ways to set that up. The CDC recommends and purchases a lot of vaccines. Even health insurers in the US—the few really biggest ones—could they get one of these going? I don’t know. This is sort of out of my realm of expertise.

Fin: Yeah. Interesting. I feel like another policy idea you mentioned in this blog post kind of rhymes in my head with advanced market commitments. It’s also about installing some kind of certainty and confidence about what happens if you were to successfully develop some kind of drug for a neglected disease. This is the idea of ACIP work groups.

Jacob: So what it does is it’s a part of the CDC in the US that recommends which vaccines should be used in what populations to the US government. The FDA gets a lot of attention in discussions of R&D and health technology development, as it should, because it’s illegal to sell various health products unless the FDA approves them. However, once you have a health product that you think is going to help people, there’s a separate question about whether people should pay for it, whether insurers should pay for it, or whether the US government should pay for it, and for which people and conditions. Across different governments around the world, you have what’s often called health technology assessments, which are a different step in the process than the regulatory assessment. One of the bodies that does this in the US is ACIP, which we both know what it stands for.

Fin: Yeah.

Jacob: That looks at vaccines in particular. They have a meeting three times a year that discusses in-depth if you have a vaccine near the finish line or that just got approved by the FDA. Well, what should we do with it? Should we use it? How cost-effective is it? And which populations should we prioritize?

Fin: That’s what it does. Yes. The idea you suggest is about taking those meetings that normally happen once a drug is developed and then bringing them forward in time. How would that work?

Jacob: Yes.

So this is something that the FDA encourages drug makers and vaccine makers to do for their own products, often creating something called a Target Product Profile (TPP) that outlines how effective a vaccine needs to be for it to be worth submitting a drug application at the end. Does it need to be 50% better than the thing on the market? Does it need to be 10% better? How many doses does it need to be in? Do I need a booster dose? The FDA encourages this process.

One thing I think would be useful, though, is to not just make that about a specific candidate, but to create a framework for scientists and product developers to know what they are really aiming at. If they meet those targets, they can be fairly confident that there will be a significant payoff at the end because the US government has indicated interest in that vaccine if it can be developed.

For example, for hepatitis C, various people are working on vaccine candidates, but there is currently no available vaccine. If vaccine developers knew under what conditions and for which populations the CDC would recommend the use of that vaccine, it would influence how they design the trials and how much they invest in that particular line of inquiry. Right now, there’s a lot of demand uncertainty. It’s not even about the actual dollar figure at the end; it’s just that uncertainty makes it hard to convince investors and boards to invest in a product line.

Fin: Got it. And I guess just to say it back, if you really knew that, conditional on some outcomes of the trials, the CDC would or would not recommend your products, then you could go to your investors and board and say, “Here’s exactly what it’ll look like if we succeed here.” There’s something about the confidence and the narrowed uncertainty that matters in itself beyond just the expected returns. That speeds things up.

Yeah. Another policy idea you mentioned had to do with the way that the FDA shares its assessments of new drugs, right? The FDA takes a look at the data from trials on a drug, decides whether to approve it or not, and then releases some public information, which is often quite sparse.

Jacob: Well, you can check on the website. They do release their drug reviews, which are often 200 pages long. The issue, as you say, is that they are sparse in the sense that they often block out key information you might need to know if you are another regulator approving the drug, like where it was made and what type of bioreactor was used.

Fin: First of all, why does the FDA not discuss a bunch of things it knows about why it approved or did not approve a drug? Why is that a bad thing, or why could it be a bad thing in some cases?

Jacob: I published a particular version and posted on my blog an idea that I thought was very watered down, and I was sort of annoyed that no one got mad at me in the other direction. You could imagine a world where you say, “Oh, you, pharma company, want to access the US market? Absolutely fine. You gotta apply through the FDA to make sure that your drug doesn’t cause harm and that it does what it says it does.” But guess what? One of the prices of doing business, where you’re gonna make hundreds of millions of dollars in our market, is that we get to tell the public how you made the drug. By the way, you can keep your patent.

Fin: Right. Fair enough.

Jacob: This isn’t threatening the patent.

So this is not what I recommended because I’m not as strong-willed as I used to be. But no.

Fin: One even milk toast version.

Jacob: Of the thing. I had a much more milk toast version which said, okay, well, the FDA should be more proactive about signing confidentiality commitments with other government agencies, saying, look, you’ve got to keep this secret because you can take a look; you obviously need to read this full application before you approve it for your population, but just don’t share it. And, you know, that is something that absolutely could happen at a much quicker pace than it currently does. The current agreements don’t even usually cover trade secrets. They cover sharing other bits of information that might otherwise get screened out. So just making it possible to share trade secrets with more countries, even when they’re private beyond that regulator, would be wonderful too.

Fin: So if I am a government regulator in a country that is not the US and there is some new drug, which I would like to know whether to approve to distribute in my country, it’d be really useful to look at the approval decision from the FDA and learn the precise reasons why it made that decision.

Jacob: Absolutely right. Yeah.

Fin: Currently, by default, there’s a bunch of missing information from whatever the FDA publishes. A minimal and kind of obviously good-seeming thing is for the FDA to selectively share that useful information with me, the other government regulator. I’m sure I’ll sign a confidentiality agreement. But, man, that would be helpful because it would just massively speed up and maybe make it more likely that I approve this drug, which is in fact a good thing to approve.

Jacob: Yes. Sign me up. Sounds great. I mean, there’s also a broader question of redundancy in the system where currently a lot of countries rely on-site inspections of a manufacturing plant done by the European Medical Agency, the FDA, or the WHO. Or the UK MHRA. And that is overall a good thing because you don’t want 200 countries to send 200 inspection groups to each manufacturing site. It’s okay to rely on each other. And, you know, it’s okay for the US to rely on the EMA more than it does. That would be a happy world if we had a little bit more regulatory reliance. Yeah, you can check each other’s work, and you don’t have to accept everything. But if you shared more documents, that would be a great first step, and it would not be that hard, I think.

PEPFAR and the importance of stable funding

Fin: Mhmm. Let’s do it. Since we’re rattling through policy ideas, the next one I want to mention is, so you talk about PEPFAR in this press, and you talk about the budget cycles that PEPFAR runs on. But first of all, what is PEPFAR?

Jacob: PEPFAR is a global health program set up under the George W. Bush administration that started at the height of the HIV/AIDS epidemic in terms of the number of deaths that HIV/AIDS was causing at that time. So I think maybe 1,500,000 or so people were dying of the disease in 2003 when I think it was set up. So what was it? I mean, it was set up to help provide funding and develop health infrastructure in lots of countries that were most affected by HIV/AIDS. A lot in Southern Africa in particular, but elsewhere too.

And, you know, it’s a really huge success story in that a lot of people have spent a lot of their career trying to work on bringing HIV under control, and we now live in a world where it is more under control. I’m not sure off the top of my head exactly how many deaths there are per year now, but it’s probably 600,000 or something. So we’re really in a vastly better situation than we were, and it’s something the US government can be very proud of because PEPFAR was a very substantial contributor to that really enormous public health win.

Fin: I would be interested in some color on what this really looks like. Like, what kinds of programs were enabled under PEPFAR?

Jacob: Yeah. I mean, just to say up front, it’s not really my…

Fin: Sure.

Jacob: Sure. Sure.

Expertise, but I think what you know, for example, much better, antenatal health screening for pregnant women, where the rate of transmission from mother to child of HIV has just gone way down in Botswana and South Africa. That is kind of how these other positive benefits emerge. If you set up better antenatal care for HIV/AIDS, you know, for those reasons or because the funding is coming for that purpose, it means that more women get at least one antenatal care visit anyway. So, there might be other stuff that you’re doing there. So, that really changes the infrastructure in a country in some ways. So, does that kind of work? There’s also other stuff related to having better trial sites for vaccines; that kind of work has also been helped indirectly by.

Fin: PEPFAR. Gotcha. And PEPFAR very much still exists, still operates. The idea you want us to suggest had to do with the way it’s funded. What’s the problem there?

Jacob: Yeah. So, it’s been a program that’s had bipartisan support since being founded 20 years ago and has been on sort of 5-year cycles of reauthorization. This year is the first year that has changed, where the budget is staying flat. So, it is funded this year, but it’s only been extended for one more year. So, there’s not a stability around it that it previously enjoyed. That causes a bit of an issue in the confidence that people working in different countries have regarding what they can commit to in future years. It’s really depressing to reflect on because it’s one of the real things to be proud of, I think, that we as a country have collectively done. I think getting back to 5-year cycles would be the obvious next thing you could do with PEPFAR, and hopefully, something we can achieve collectively.

Fin: Yeah. Maybe this is just stating the obvious, but what’s so bad about shorter funding cycles if the budget itself is staying flat year on year?

Jacob: Yeah. The type of work that’s being done is often providing healthcare or support to very vulnerable people. So, it is not ideal in that context to have a level of uncertainty where you’re not sure that that support will come in the future. Right. There’s almost an extra responsibility you take on, I think, once you’ve built something up that is playing a large role in people’s lives who won’t have great options if you pull back. When you inaugurate a program, it’s not something that has to last for the next 500 years.

But giving a duration of visibility is kind of a responsible thing to do. And, of course, it makes people worry that you might cut funding, and it means that you pull back on new programs. It means that you pull back on addressing the problem. Right. Right. Right. And still, HIV/AIDS is one of the biggest killers globally. Sure.

Fin: For sure, now.

Jacob: is not the time to pull back. So, if you’re making people wonder, “Wait, are they going to pull back now?” Which probably the US won’t, but then it’s just a really spooky thing to do.

Fin: Sounds right. And I find it interesting that one theme through some of these policy ideas we mentioned is finding ways, independently of just total funding going to something, to install more confidence about how some programs will look in some years down the line. It just seems like, in general, committing to things and installing confidence is a really neat policy direction because, in itself, it doesn’t cost money.

Jacob: That’s right. I think more money is good too, but it’s so easy to just say, “Well, if we had more money, we could do more.” And, you know, I should say one of the ideas was just to make sure that Gavi and the Global Fund get sufficient support in the upcoming replenishments. As I look at Gavi now, I am.

Fin: sort.

Jacob: of concerned by how cost-effective it looks. You know, you shouldn’t have an institution that’s as big as Gavi look that cost-effective at the margin.

Fin: That it looks very cost-effective. It looks.

very cost effective because it looks right now that, you know, one way you could look at it is that the cost effectiveness at the margin in the next 5 years will be partly driven by the R21 malaria vaccine. They got more money. They might be able to support countries at higher levels of ambition with that vaccine, and there’s going to be relatively high demand for it. And that vaccine looks very cost effective. If that ends up being slightly too low ambition level, you know, that’s leaving so much undone. So, anyway, we’re in a particularly cost-effective period potentially.

Fin: Yeah. I like this thought that at least these really big, obvious, scalable, international programs shouldn’t be crazily cost effective.

Vigilante accountability projects

Fin: Let’s talk about things people can do in general. People meaning maybe people listening to this. You wrote this—I don’t even know if it was a blog post. It might have just been a Google Doc—about vigilante accountability projects. I think maybe you coined this idea. So what is it?

Jacob: Oh, yeah. I wrote that a few years ago, partly for fun, but I think there is a dearth of a particular kind of written or filmed piece that I personally would love to read or watch more of. So maybe it’s a personal need, but where you take a big institution that is performing some important function in society and critique it from the outside, but not to try and take it down necessarily. You know, maybe it’s a good institution. Maybe it’s the Gates Foundation, and they’re doing a lot of good work, and there’s something they’re not doing that they should be doing. And you really try and dissect how GAVI actually makes decisions. You know, how does it decide which vaccines to support and which not to support? How does it interface with different country health ministries? What level of constraint are they operating under? You know, all of those questions are so important for so many kids who will end up either getting or not getting lifesaving vaccines. But GAVI is so boring that no one knows the answer to those questions.

I bet you some people who work at GAVI don’t know the answers to those questions. And I think what would be wonderful is if someone who has an instinct for the institutional really takes a focus on these institutions and writes something that is interesting enough that it provides analytical value to an outsider and can indeed be opinionated and used by insiders at that organization to win certain debates internally. That’s absolutely a theory of change. One piece that comes to mind that I think I really liked when it came out, again, maybe this was a couple of years ago, was Matt Fahidi wrote a piece on NIH for New Science.

Fin: And I can’t remember seeing that. I definitely didn’t think of it.

Jacob: Yeah. I wrote it. The thing is that I did read it. That’s how fun it was. And I was like, wow. Okay. The NIH is huge. It’s almost too big to write one piece about. Maybe you could peel off the National Cancer Institute or something like that. But it was really valuable to me as a reader to get a sense of, okay, how does this work? You really can piece together a lot as an outsider if you just do a few interviews with people who have experienced it. The Center for Global Development does related things on GAVI and on multilateral banks and all that. So I don’t want to imply that this never happens. Sure, think tanks do related things, and investigative journalists do related things.

Fin: Yeah. One thing this makes me think is, you know, often you’ll get some potentially quite charged discussions about, you know, are we pro or anti this program or this agency? Like, should the FDA be more funded or less funded? I wonder if often those conversations are kind of guided by pretty hazy pictures of what the agency or program actually does.

Jacob: Oh my gosh. I mean, this is like the bane of my existence.

It that it’s, you know, if we could have a debate about the FDA where people knew what the FDA did, I’d hang up my spurs. I think we’d be done.

Fin: But yeah, it’s just so boring to know things. It’s just like a real problem that it’s way more fun to get mad about something.

Fin: Yeah. And, like, you know, one thing people say to this kind of problem is, well, we should hear both sides of some debate. But there is this independent thing where you should just understand the processes and the mechanisms and the boring stuff. You just need to read lots of words. I know. I know. It’s…

Jacob: I’ve been wanting to write, actually. Maybe I’ll write a blog post about how to find it fun to read boring things because I think that is a limitation on our society. A lot of the time, the information that you need is public. It just is in an FDA document. It is in an IPO filing. It is in… and they’re just…

Fin: So anyway, there’s an incredible amount of juice you can get as a journalist just from being prepared to read the 100-page documents. All the other journalists just cannibalize the one person who actually read it and had opinions about it.

Jacob: Totally. And I, you know, I keep waiting for the world where large language models will mean I can stop reading. And they have helped when I do read, but we haven’t quite got to a place where I could stop reading.

Fin: And I remember one time, sort of an apocryphal quote from someone at Open Philanthropy, who was asked, what’s your career advice for people who want to have the kind of success you’ve had? This person, who kept anonymous, said, my advice is to read a lot of nonfiction from a young age. I was like, oh no. You don’t want to hear that.

Fin: You don’t…

Jacob: …want to hear that. Yeah.

Fin: That’s very funny. How do I read more boring stuff and find it fun? What’s the tip?

Jacob: One thing I find kind of fun is viewing our society through the lens of someone 100 or 200 years in the future and looking back on it. You might have some friends. I’ve got a friend who looks back on documents from the bureaucracy of Napoleon, like counting up different cavalry and doing this, that, and the other as a sort of, wow, this is how things used to work. So sometimes I look at an IND filing, and I go, okay, what a strange artifact I’m looking at right now. Like, who is it designed to persuade? Who is it designed to calm? You know, just getting a bit more into the texture of, oh, what is this? Like, why is this the way we as a society have landed on conveying this information? That way, you can read a couple of them and then give up, and you can at least have learned something about the way information is…

Fin: I like it. Yeah. I guess I’m picturing, like, for a Martian, and I wanted to understand how some country works. And I was able to read all the documents, but I couldn’t have access to that country’s Twitter, right? I couldn’t learn what are the current hot topics, what are the battle lines and the political controversies.

Jacob: Yes.

Fin: I just have to start from ground zero, and I might form quite a different picture from the picture that is presented in, you know, the kind of surface-level political conversations.

Jacob: And I mean, in terms of learning, one other thing that is definitely useful is integrating the Martian perspective from reading those documents with having just one or two conversations with people who have produced the documents or who understand what’s really going on when people say XYZ. And that can also make it more fun. You can be like, oh, wow, I’m sort of getting an insight into this societal production, what’s really going on. So I definitely would pair the social with the reading.

way that maybe would be even better than the motion.

Fin: I like that—cracking the code. Here’s another question. If someone listening to this just wanted to contribute in some way, maybe they have some economics background. They could do some research. Maybe they even have a public health or bio background. What work needs doing? I can narrow that down if that’s an unwieldy question, but I’m just going to leave that.

Jacob: Okay, I’ll take a stab. I mean, there’s a lot of scientific work, which is almost obvious to say. If you have a background in biology, biochemistry, pharmacology, epidemiology, structural biology, or biomanufacturing, there’s a huge amount of work related to the problems we’ve discussed. But leaving that aside and thinking about social science research, I think there are still ways to contribute to these problems as a non-scientist. At Open Philanthropy, in February, we posted a blog post about research topics in social science that would be useful for us, and maybe you can link to that. We are massive consumers of social science research. There are things around what drives the diffusion of medical technologies across countries. Why do certain medical technologies make it to different countries sooner than others? A lot of social science tools can apply to that. Also, what burden of disease estimates are most likely to be wrong? That’s something you could tackle as an outsider without needing as much scientific background. If the NIH says no to a proposal, what is the chance it gets funded by another funder? That’s something we have really tried to scrutinize, but I haven’t seen many papers on.

For investigative journalists or journalists generally, I think there’s a lot you could cover. For example, there’s a lot you could write about the delay between drugs getting invented and actually making it to people who need them. For small molecules, where it’s quite easy to make generic versions, you could look at different companies and say, “This company got this small molecule approved in the US in this year. Has it marketed that product in middle-income countries yet?” You can go through and ask, “Why not?” If not, they should at least not be defending their patents in that country after a few years. It’s absurd if there’s an invention that they’re not even marketing. You could really run through different pharmaceutical companies and look for this. They could form deals with the Medicines Patent Pool. There are ways to do this that don’t harm the interest of the company in question that much, but would be wonderful to write about.

Fin: Yeah. Go ahead. Okay. These are some nice examples, mostly social science questions. What else can people do? I’m sure that this is just indefinitely long, right? But tell me some other things.

Jacob: Some other things. In any country that has an aid budget, you can contribute just by advocating for GAVI, for GoFund, or, if you’re in Europe, for EDCTP as an institution. You can volunteer for medical research. I am trying to get into a trial right now, and I haven’t succeeded yet, but I’ll get paid a grand if I get in, so that’ll be nice. On the writing front, if you’re not a journalist, I think you can still write long-form pieces that I would love to read. So I’ll just selfishly plug that I would love to read a detailed piece on HHMI and the Genelia Research Campus. It’s been around long enough that there are some lessons to learn there, but I haven’t seen that. HHMI? HHMI, the Howard Hughes Medical Institute. Gotcha. I’d love to read about the Global Funding, Gary, as we talked about. I’d love to read about the really successful vaccine companies in Hyderabad and in Suriname and Pune. I’d love to read about Gates, Wellcome, and the other funders.

I’d love to read about us. Maybe one day, I will get around to writing something about open philanthropy. Yeah. Biometrical decision-making, but I haven’t gotten around to it yet. So someone look at our grants database and tell us what we’re missing, and that’d be great.

Fin: I’m actually realizing that there is a question that I meant to ask, or at least some people would notice that I didn’t ask if I didn’t, which is about AI. So, you know, in some sense, another division of the organization that you work for, I believe, is spending a lot of time worrying about, or at least anticipating worlds where artificial intelligence ends up, among other things, automating a bunch of R&D. Maybe quite soon, maybe within, you know, 15 years or something, maybe even less. Presumably including a bunch of life sciences research and the kinds of R&D that you’re funding. So I should at least ask, how do you think about that? Wouldn’t it make sense to do the really quick payoff public health stuff now, as far as you care about global health, and wait for the AI to do the R&D for you? Isn’t this in some kind of awkward sweet spot?

Jacob: What a question. I think that, well, I have a lot of thoughts on this cluster of topics. I’m wondering what to go for first. So one thing implicit in your question maybe is that AI would help life sciences R&D or change that sector of the economy quicker than it changes other sectors of the economy. Do you feel like that’s key to your question or not?

Fin: Let’s see. I think it’s kind of halfway there. So you might just think AI takes care of tasks across the economy at some point, in which case you want to do as much immediate good as you can in terms of health outcomes before that happens. Maybe it differentially speeds up life sciences R&D, in which case that’s like an extra reason to not focus on or fund the kind of long timeline R&D projects. So you’re…

Jacob: Not interested enough. Okay. But so focusing more on the first one, then I guess, I think that is a reason. If you believe that AI will sweep across the economy in the near term, then that is a reason to tilt more towards public health and away from the kinds of R&D that we support. I think that’s true. We live in a world of much uncertainty. And how you integrate that uncertainty into your decisions as a grant maker, into your decisions of what you work on, is tricky. And I don’t think everyone, even at Open Philanthropy, certainly not everyone at Open Philanthropy lands in the same place on that. But I bet if you polled my colleagues on the global catastrophic risk side of the organization that think about AI transformation a lot, their numbers would be maybe 2x or 3x more towards GiveWell-type interventions than the stuff that we work on in the science side, would be my guess. I am not imposing on any particular person.

Fin: Yep.

Jacob: But for the reasons you’re pointing to.

Fin: Gotcha. Any of that? I feel like you have lots of things, the same thing you want to say in your world.

Jacob: I do. I mean, we’re sitting right now in San Francisco, and many of my friends work on AI, totally unrelated to the topics that we’re talking about now. So, you know, I can’t really escape many of these discussions from all sorts of different angles. And I think maybe my more detailed thoughts are for another podcast.

Fin: Gotcha. I look forward to having that podcast. Yeah, maybe one question is just, what is a book that you are really surprised to love from the last year or two?

Jacob: You know, the last nonfiction book I remember loving that I sort of had no opinion going in was, I was like, “Wow,” was The Journalist and the Murderer by Janet Malcolm. So my friend, who’s a journalist, recommended it to me. So it’s about journalistic ethics and is about the relationship between a nonfiction writer and the subject they’re writing about.

Sounds quite dry, but it’s actually fascinating and gets into some cool stuff about truth, how you can get truth out of someone, and what constitutes an ethical versus an unethical way to do that. It discusses narrative and the presentation of people in nonfiction. There are wonderful quotes about how the best people to interview are those who’ve already done the work on themselves to become literary characters. If you interview someone who can present in a zany and intriguing way, you’re much more likely to write about them. I find that fascinating.

Fin: Yeah.

Jacob: This was not in the last year; it was, I don’t know, a couple of years ago, but I finally got around to reading Anna Karenina. I assumed that would be, you know, sort of eating my broccoli, and it would be pretentious. But it was amazing. It was so good, and I have to thank a different friend, Anya, for really getting me started on it. I found it very funny and a sort of calming experience because it was written so long ago, but it seems so familiar. All the drives of the characters, the jealousies, the petty grievances, the joys, the sorrows—it’s just really calming about the human condition.

Fin: That is so great. I do feel like lots of classical literature gets bucketed into this category of high art, like literary quinoa. It’s healthy, but it’s kind of boring. There’s just a ton of variety, right? So much—like, 19th-century novels are serialized; they’re really pulpy, really melodramatic. We have all these fun, relatable characters.

Jacob: Well, now I want a recommendation from you. What’s a 19th or early 20th-century novel I should read?

Fin: When I was saying that, I had in mind Dickens and The Brothers Karamazov.

Jacob: Nice. Well, I’ve done very little Dickens and no Brothers Karamazov.

Fin: The other one that I haven’t read and would love to read, because it fits in this category of just being engrossing and fun, is The Count of Monte Cristo.

Jacob: Oh, yes. Yes. Yes. I think I’ve been to the island where that character is—is it fiction?

Fin: It’s fiction.

Jacob: Is there someone locked up on that island?

Fin: That’s right.

Jacob: Very cool. Couldn’t tell you where it was.

Fin: Maybe when I was… yeah, that sounds good. Great. Good stuff. Alright, let’s do some closing questions. Something we ask everyone: just three resources, meaning books or papers or whatever, that listeners can read if they want to find out more about what you’ve been talking about.

Jacob: There’s a podcast called This Podcast Will Kill You.

Fin: Oh, cool. Have you ever listened to it?

Jacob: I have not listened to it, and Luca keeps telling me I should listen to it.

Fin: You should; Luca’s a wise man. Yes, This Podcast Will Kill You is a lot of fun because each episode is about a different infectious disease. The two co-hosts go into historical and scientific detail about that disease. You can really scroll through the old episodes and think, “Oh, what’s the disease I’m interested in or have had before?” You know, and really learn about it. It’s great fun, and they manage to weave in the more technical details while making the boring interesting. I definitely recommend them.

Then, in terms of writers, I wouldn’t recommend academic papers per se, but there’s a lot of long-form writing that’s pretty good. Some of the newer science publications come to mind, like Asimov Press. They publish loads of detailed write-ups on what I find to be interesting bits of history and technology. There are particular writers as well. I guess I meant to stick to three, but I’ll say Saloni Datani’s writing is absolutely wonderful. She is at the nexus of global health and science in a way that I find very valuable to read.

So you’d subscribe to her Substack, or if she has both a Substack and writes pieces.

Fin: That’s right. Yeah. Friend of the podcast as well, actually.

Jacob: Friend of the pod. Okay. Very good. Now I’m feeling guilty. There are many other good blogs out there.

Fin: We will link to all those things. Asimov Press, I will sing about that. They are great, and it’s interesting to me how rare that kind of model is for a technically detailed magazine about important technical subjects.

Jacob: Yeah. The closest I can think of is Quanta Magazine. Yeah. They get too much. You know, there used to be this magazine that the Wellcome Trust funded. I don’t know if you’d call it a magazine, but it was called Mosaic. They no longer exist, and that’s a great shame. They had a great piece on Strapay, and they wrote some good pieces. Sometimes these come in waves, but we got a new way of building it, it seems. Yeah.

Fin: Let’s skim on magazines. Alright, last question is, where can people find you and find your work?

Jacob: They can find me at blog.jacobdefeathen.com if you want to read some of the blog posts that we talked about. And you can find me and my colleagues in the Open Philanthropy grants database where every grant we make is public, and you can poke around and filter on science and global health R&D and see what you think.

Fin: Alright. Jacob Trefethen, thank you very much.

Jacob: Thank you, Fin.

Fin: That was Jacob Trefethen on Global Health R&D. If you’re looking for links or transcripts, you can go to hearthisidea.com/episodes/trefethen. That is T-R-E-F-E-T-H-E-N. If you find this podcast valuable in some way, then probably the most effective way to help is just to write an honest review wherever you’re listening to this, so Apple Podcasts or Spotify or whatever.

Jacob: And you can also follow us on Twitter. We are just @hearthisidea, all one word. As always, a big thanks to our producer, Jason, for editing these episodes, and thank you very much for listening.