Elie Hassenfeld: [00:00:00] Hey everyone, this is Elie Hassenfeld, GiveWell's co-founder and CEO, and today I'm going to talk to Dan Brown, who leads GiveWell's New Areas team in our research department.

In general, GiveWell has teams that make grants in areas that we've done a lot of work on in the past, so those are areas like water, nutrition, livelihoods, malaria, and more. The grantmaking we do in all of these areas is large-scale and it's impactful, and altogether, grantmaking in those areas totaled $350 million in 2025. But there are many important areas beyond these problems where people are suffering and children are dying. And so the New Areas team is searching for programs that address problems that fall outside of the ones that we have focused on to date. We hope that by searching more widely, we can find more cost-effective ways to address some of the largest problems in the world.

I think there are a couple things that are really important about this [00:01:00] work that I want to highlight. So first is just that I hope that this conversation will give some insight into the increased scope and breadth of what we're doing. And I think many of these are different from what you might imagine is the stereotypical kind of program that GiveWell has supported in the past. And so in one way, we're doing things differently because we think that it will enable us to help more people. And then at the same time, I think the core of what GiveWell is and how we approach questions is going to sound very similar. It's the same principles of looking for evidence or supporting evidence, bringing rigor and truth-seeking to what we do, focusing on cost-effectiveness, and then learning over time, that is the foundation of all of this work, including in New Areas.

So today we're going to talk to Dan Brown, who leads this team, and we're going to go through three of the grants his team has made. So there's first an electronic supply chain management system in Laos. We're then going to talk about research and learning about rolling out a newly developed approach [00:02:00] to medical oxygen, and then finally a randomized control trial of Noora Health, a nonprofit that's existed for a long time, and this trial will help us determine how cost-effective its program is and whether we should support it at scale.

So Dan, thanks for having this conversation. Could you just introduce yourself and some of the work that you and the New Areas team are doing?

Dan Brown: Yeah. Hey, Elie. So I'm Dan Brown. I'm a senior program officer, and I lead GiveWell's New Areas team. So this is the sub-team within the research team that's kind of responsible for investigating and finding programs in cause areas that we've not really spent much time on historically.

So we have a separate team for malaria, nutrition, safe water, vaccines, livelihoods, and we're kind of responsible for anything and everything outside of that. And really our role for GiveWell, I think, is to explore new stuff, learn about it, and tee up areas that we think could be really promising, could have lots of highly cost-effective interventions that could absorb a lot of funding in the future.

We're a relatively small team, six people. That's three researchers, [00:03:00] two project managers, and myself. So on the research side, we have Meika Ball, Dilhan Perera, and Daniel Issing, and they cover our work across family planning, maternal and newborn health, TB, HIV, medical oxygen, health system strengthening, all kinds of different areas.

And then our two project managers, Rachel Mitchell and Kim Vidal, are supporting those researchers with kind of all of our planning, and resourcing and capacity, and engagement with grantees. And so yeah, it's a tight team, but I think a really productive one and covering a lot of breadth.

Elie Hassenfeld: And it's one of the teams—I mean, we wanna grow across the board, but it's one of the teams that we know we really need to increase our size. We're hiring for, because there's just so much ground that we need to cover and that we could cover. It's pretty amazing the ground we are covering, you are covering with this small team. So, yeah, hopefully we'll be able to bring some folks on to help us even do more.

Dan Brown: Yeah, very excited about that.

Elie Hassenfeld: Let's just like dive in. I want to talk about a few different grants that the team has made and is considering. So I want to start with one, the  [00:04:00] technical term is an electronic logistics management information system. Basically, an electronic system for the supply chain. So, you know, just tell me a little bit about this grant, where it is, what it does, what problem it's trying to address.

Dan Brown: Yeah. So this is a grant to the Clinton Health Access Initiative, CHAI, to implement this, as you say, a bit of a mouthful, electronic logistics management information system.

It's a specific version of that system called Open mSupply. And the idea is to implement it in primary health facilities across four provinces in Laos and conduct an RCT, a randomized controlled trial, to evaluate the impact of this system on stockout rates and patient treatment access. So this system, this electronic logistics management information system, people sometimes refer to it as eLMIS, which I might do during this podcast because it's such a mouthful. It's basically a piece of software that tracks health commodities across the supply chain right from the sort of central warehouse level through the regional warehouses down to individual [00:05:00] health facilities. And the idea is it allows facilities themselves and Ministry of Health officials at all levels to see real-time data on stock levels for key health commodities, all the way down to the facility level. And hopefully, that enables them to basically improve planning and reduce the chance of stockouts in public health facilities. So this system in Laos is already in place at the warehouse level but not at the facility level, so the idea here is to expand it to the facilities, and then test whether there is an impact on stockouts.

Elie Hassenfeld: So tell me a little bit more about what exists today and what problems this new system would resolve by being present, you know, not only in warehouses, but in facilities too. Like what are the concrete challenges that the health system faces without this system in place that it could address?

Dan Brown: So right now the system is pretty much paper-based in all health facilities, and by that I mean that staff at health facilities are counting stock for commodities by hand, [00:06:00] typically around once a month. They're filling this out on paper forms and then entering some of that data into existing sort of national data systems. And I think they're typically doing that for only a subset of health commodities, and there's about 300 commodities per health facility.

So right now, that means there's kind of no real-time visibility on stock levels. You're getting this kind of snapshot once a month for a subset of products. And we think that can lead to both stockouts, or on the flip side, it could lead to some commodities being overstocked. Essentially the reason here is that facilities may only realize that they've stocked out of a commodity when, you know, they go to pick some up from the ward or the storeroom, and there's nothing left. And then they need to reorder, and that takes some time, and as a result, patients don't have access to treatment in the meantime.

Or on the flip side, if facilities are worried about the possibility that they're going to stock out but don't have that real-time visibility on what's there, they may sort of over-order preemptively and in turn that can lead to like expiry and wastage of products. So we think this system can kind of improve on that basically by [00:07:00] allowing facilities to order in plenty of time because they have that visibility on whether stocks are running low, and flip side and not over-order stock.

It also allows better planning at like higher levels beyond the facilities themselves. So district managers at the Ministry of Health can see the stock levels across facilities, so if they see that one facility is relatively low in stocks, the other is overstocked, they can kind of move stocks across facilities. And even then at the national level, if you have this visibility on stock levels across facilities you can hopefully better plan and make procurement decisions that are kind of meeting demand.

I should also say one other benefit is, I mentioned there's like 300 health commodities I think on average at facilities, and one other benefit is this system generates automatic kind of reorder alerts, again to let the staff at the facility know that they're running low and then in turn order more.

So yeah, we think in terms of the size of the problem of stockouts, I think it varies a lot by countries and facilities, and actually kind of part of the point is in some cases there's not good data on this, because you're only getting this snapshot once a month. But we have heard [00:08:00] it be a really serious problem in a lot of countries.

I think there was a Lancet study recently that suggested there was this sort of bundle of essential medicines, I think mostly antibiotics and maternal newborn health medicines, and this study found that, you know, only around 50% of these are stocked on average in facilities in a sample of low-income countries. So, yeah, we think it could be a real problem that patients are not able to access treatment because of stockouts.

Elie Hassenfeld: And just so to make it like really concrete, is there an example of an essential medicine that you imagine is periodically stocked out and you know, what would happen? So what's the case where the caregiver, the family shows up and they need, you know, some particular medicine because of some problem, and then it's—I mean, can you just walk me through an example that you think would be representative of what this would address?

Dan Brown: Yeah. So I think one we've heard about a bit in Laos and elsewhere is amoxicillin. So this is like a first-line treatment for bacterial pneumonia. Pneumonia is a major killer of under-fives, and it's accountable for more than 10% of under-five deaths globally. So I think in [00:09:00] practice this could be like, you know, child is diagnosed, they need to be put on amoxicillin, it's not available. In terms of what the caregiver does next, I think is a great question, and this is actually something that we're hoping to learn a lot more about through the trial that we're funding as part of this work.

So it is possible that caregivers go and seek care for their child elsewhere. So that might be that they go to other public health facilities to see if the drug they need is stocked there. It could be that they go to private pharmacies, and if the treatment again is available in private pharmacies, they may pay out of pocket. And so one nice feature of the trial that we're funding is that, as well as directly measuring the impact on stocks in the facilities in this trial, the enumerators, those doing the study, are going to follow up with a subset of patients who weren't able to access the treatment. About, I think, like two weeks later they'll make phone calls and ask, you know, "What did you do next? Did you go and seek care elsewhere? Were you able to access treatment elsewhere?"

Elie Hassenfeld: Yeah, I'm always particularly struck by amoxicillin as an example. You know, I have young children, you do too. I mean, I [00:10:00] don't know if you've done this, but I've certainly gone around the corner to pick up amoxicillin.

It's like one of the most common drugs that I need to pick up for my kids when they have something wrong, and it never even crosses my mind that it wouldn't be immediately available when I need it, and I'm always grateful. I mean, I think about, I mean, GiveWell has done some work to fund amoxicillin-focused programs in the past, and it's just really frightening to imagine being in a position where you're a caregiver bringing your child for treatment and the drug is this like very basic antibiotic is literally not in stock.

But I'm just, I'm curious, like why are we doing this in Laos as opposed to elsewhere? Like how did we land there for this work?

Dan Brown: It's a great question. We actually don't think, I mean, we think the program will be impactful, but the direct impact of the program in Laos probably isn't going to be as cost-effective as some of our outside options or other funding opportunities.

But in Laos, I think we found partners, you know, in the Ministry of Health who were open to testing the effect of the program on stockouts and patient treatment access. So in part, I think it's kind of identifying an opportunity where we really do have high [00:11:00] learning value, and it may be that if we go to sort of fund the program at scale, we would do it elsewhere, perhaps in a context where morbidity and mortality in health facilities is much higher, and the ultimate effect of the program and cost-effectiveness of the program may be higher. But yeah, we really saw this as a great opportunity to learn.

Elie Hassenfeld: Yeah. So let's turn to the evidence part of this, and I want to talk about the trial that we're supporting and what we plan to learn. But before that, like what's the state of the existing evidence on this type of program? I don't know how broadly you should define that, but when you think about the trial that we're supporting that's going to add to the amount of knowledge we have about the potential success of this program, you know, what exists before this trial?

Dan Brown: My understanding is there's no randomized control trials that are trying to assess the impact of an electronic logistics management information system. I think there's just very few studies out there full stop, whether it's a randomized control trial or not. I think in our estimates of the cost-effectiveness of the program, we were relying heavily on a modeling [00:12:00] study. So this is, you know, a study that's making estimates without going and collecting a load of primary data on this. So I think the evidence at the moment is very, very limited, and I think that's something that maybe is sort of generalizable across health systems strengthening interventions, for which this is a sort of a subset focused on the supply chain. I think that's an area that we're really excited about potentially funding more, not least given that the US government before funding cuts was funding a lot of work in that space.

But often a lot of interventions in that area are not things that you can have a randomized control trial or even a sort of controlled trial for because, you know, it may be like a national-level intervention, and there's just no way of having a control if you're making changes in like procurement or forecasting work at a kind of a national level. So yeah, we think health systems more generally, that sort of work is, yeah, an area where there's relatively few rigorous evaluations.

Elie Hassenfeld: So it's really interesting. There's extremely limited randomized evidence, maybe no randomized evidence for a program like this. And so then, yeah, tell me a little bit more about what will be measured here, what the outcome indicators are going to be, and especially, [00:13:00] like, what are the most important questions? Like, the key questions on your mind where, if this kind of intervention isn't as effective as you think it might be, like, why would that be, and how will the trial help answer that really critical question?

Dan Brown: So I think in terms of the design and the outcome measures—well, maybe I start with the design. It's a randomized controlled trial with about a hundred and eighty facilities, a hundred and seventy-six facilities, I think. So at random, they'll be allocated into, you know, half of them will receive the eLMIS and the other half won't. I think there'll be about three months of baseline data collection and then some follow-up for about twelve months afterwards.

And there's two main outcomes that we're looking at here. One is actual stockouts of the commodities themselves. I think the design will both look at understocking, so actual stockout rates, but also this other point I mentioned earlier that sometimes there can be over-ordering and you have some expiry and wastage of products.

Exactly how that will be measured, I think is still being worked out. We'll receive a kind of full plan from the researchers very shortly. So, you know, for the kind of understock side of things, do you just count cases where [00:14:00] a commodity is completely out and there's like none of it left at all? Or what about if there's like one bottle of amoxicillin left, does that count as adequately stocked? So I think there's still a little bit of working out exactly what the measure will be. But yeah, that's one piece is understanding the stockouts themselves.

And then the second piece is these patient exit surveys. So when patients leave a facility, asking them, "Were you able to receive the treatment that you needed? If not, why not?" And then picking, I think it's about three to five patients per facility to follow up with calls a couple of weeks later and ask them, "What did you do next? Were you able to access this treatment through public health facilities or private sector?"

That second piece, again, I think is really important. We could find that there is an effect of these systems on stockout rates at facilities, but patients are able to go and pay out of pocket for the treatment in private pharmacies instead. If that was the case, whilst the program would still have an impact, the impact is primarily sort of reducing out-of-pocket costs as opposed to necessarily impacting health outcomes, because people find a way to substitute and access the treatment elsewhere.

So yeah, I think we're trying to [00:15:00]  understand that second step, and obviously that's not all the way down to the final health outcomes. We're not measuring morbidity or mortality directly in this study, but I think we're going sort of one step further than just looking at the impact on stockouts.

And to your point of, why would this program not work? I think there's a lot of explanations. The sort of theory of change here is that stockouts are being caused by this sort of informational or like planning problem. But there are lots of other reasons why commodities might stock out. You know, one is just there may be no supply of the commodity at a national level for whatever reason. Maybe the government is sort of budget constrained. It could be there's, like, global supply shortages. Another explanation is there could just be issues in logistics. You know, commodities maybe, like, stored in improper conditions and degrading and that's leading to shortages. It could be sort of a staff motivation question. You know, you set up these systems, but it still does require effort to track what commodities are being issued and so on. So there's lots of other explanations, and I think we'll start to get a bit of a sense of that by seeing whether this system in itself does have an impact on stockouts.

Elie Hassenfeld: Got it. So we have good [00:16:00] reason to believe, I mean, you have good reason to believe that stockouts are a major problem, and this electronic supply chain system will enable us and, or enable facilities to have better supplies of the drugs they need.

And then there's just this question about, well, you know, do we know that this system is actually making a difference? So, it might, other things equal, reduce stockouts, but there could be other challenges in the supply chain that are not informational that cause stockouts. Or, it might address stockouts, but patients might be able to get drugs elsewhere, and so therefore it's not having an effect. So those are the kinds of questions this trial will help us answer in Laos. You know, why aren't we trying to study directly morbidity and mortality? Like, why not go that far in the trial?

Dan Brown: One of the main reasons is just that fortunately, mortality events particularly are relatively rare, and that's particularly true in Laos relative to some of the other countries that we might fund these programs.

And so I think in order to essentially be able to detect effects on those kinds of outcomes, you would have to sample a much, much larger set of facilities, and with [00:17:00] that comes, you know, a lot of additional costs, and it's also practically a lot more difficult. So I think, part of it really is that there's sort of a bit of a trade-off between like paying for a larger, better-powered trial that can detect effects on those kind of outcomes versus, you know, do we really need that level of evidence to feel comfortable moving forward with a program like this?

So I think, you know, the current plan would be to try to look at the effects on patient treatment access, and then you're making still some assumptions about the impact of that on health outcomes. But we think we feel some level of confidence on those kind of assumptions and would still be willing to move forward with scaling up the implementation itself based on the level of evidence we get from here.

Elie Hassenfeld: Right. And I think it's like something that's so worth emphasizing that even randomized control trials, which in some ways are the, you know, they're like the gold standard of evidence. There's still this art to the science, which is what outcomes of interest do you look at? Where do you trade off between more information and less information? And so we're making this judgment call about where, you know, most of the uncertainty lies and trying to get [00:18:00] reasonable return on the research investment to answer the questions that we're most likely to need to answer to know how to proceed, and then letting other things go.

And there's certainly no definitive or true answer there. That is just a judgment call that we're trying to make, and is presumably a judgment call your team has to make a lot in the research that it supports.

Dan Brown: Exactly. And every dollar we spend on research is a dollar we're not spending funding the program itself, right? So there's a real balance there, and you want to make sure the opportunity costs make sense.

Elie Hassenfeld: Another place that this judgment call comes in is the question of, okay, so let's imagine that we run this trial, and the results are really positive. And then, presumably, we would then want to try and support this kind of program elsewhere. Because like you said, it's unlikely that Laos is the place where this program's going to have the biggest impact.

And so tell me a little bit about what comes next, and then in particular, how you think about this question of generalizability. You know, we'll have learned something about the way in which the electronic supply chain [00:19:00] system helps Laos, but, you know, what does that mean for some other place we go? And, you know, just like how do you think about that question and what comes next?

Dan Brown: It's a great question. I think it's our main uncertainty with making this grant. The sort of primary value of this grant is its learning benefit, and if the learning we get is not generalizable to the countries that we would be most excited about scaling up, then, you know, obviously that kind of undermines the value of the program.

I think there will be lots of similarities in the way that these systems work in countries, and the core, like, theory of change that I described earlier makes a lot of sense if informational barriers and planning barriers are the main barriers. Maybe the two aspects of generalizability that we're most concerned about, if we're sort of taking the results in Laos and applying them elsewhere, are firstly the role of the private sector. So to that point, around, you know, can patients get access to treatment elsewhere, and particularly, you know, paying out of pocket at private pharmacies. Our understanding is that private pharmacies are a much smaller share of the market in Laos compared to some other contexts we might consider funding. And so it's [00:20:00] possible that, you know, we get the results in the trial, we see actually there's, you know, a large effect on stockouts, and in turn that has affected patients' treatment access because they can't access commodities through the private sector, and we kind of incorrectly conclude that that would be the case in a country that we scale up. And actually in that other country, people do have much more access to commodities through the private sector. So that's like one question that we would need to consider. And, you know, part of that question also is people's ability and willingness to pay in different contexts. So I think that really in practice means, like, trying to better understand what the private sector looks like in other contexts, people's willingness to pay, and treatment availability through those pharmacies, and think a bit more carefully about whether the results would generalize.

Probably the second biggest piece I think is the supply chain complexity and how that compares between Laos and other contexts. And I think in some countries, the system is maybe a bit more fragmented. There's kind of separate supply chain logistics management information systems for each disease vertical, whether they're digital or on paper. And so, like, practically implementing a system like this [00:21:00] that cuts across lots of different commodities might be harder in other contexts where there's this sort of more fragmented system.

Elie Hassenfeld: I mean, one way that I'm thinking about this, just based on what you're saying, is that it's probably more realistic to think about this as a first rigorous study that's going to teach us a lot about this kind of approach, but it's far from, you know, the answer to how this approach would look and what its success would be in a country that's very different than Laos. And it's not clear yet, like, what we will do next, but perhaps it's reasonable to think about it as one important step towards getting more information that we then need to carry forward in the future.

Dan Brown: Absolutely. And I think it's possible we'd fund trials again in other contexts, or other types of evidence generation alongside, you know, a rollout elsewhere, specifically to try to get signals on these questions.

Elie Hassenfeld: Yeah, that makes a lot of sense. Okay, so let's move on to something else. Let's talk about medical oxygen, and maybe just to start, help us understand what it is, and what problem it addresses, and then we can dive into the grant itself.

Dan Brown: Yeah, so medical oxygen is an essential part of treatment for a lot of severe [00:22:00] illness. So, people describe it as like horizontal, it's cutting across treatment for lots of different things. Yeah, I think people probably intuitively understand the importance of oxygen post-COVID, but to give an example of some other diseases that can cause people to become hypoxemic, so that is they don't have sufficiently high levels of oxygen in their blood for sort of core bodily functions.

Pneumonia is a key one, and accounts for a little over 10% of all under-five deaths globally. But there are severe forms of a lot of other diseases that leave people requiring oxygen. I think malaria is a primary example. But then if you go to a hospital, oxygen is used in kind of all sorts of other ways, like, you know, newborn care, if a newborn is experiencing respiratory distress, it's used in surgery. It just has a lot of applications and ultimately there's just no substitute for oxygen.

The reason I think I'm so excited about medical oxygen as an area for us is, it does seem like there are quite big gaps in oxygen access in low-income countries. There's a study in The Lancet recently, I haven't dug into this in that much detail, but I think that the claim was that only around 30% of patients who require [00:23:00] oxygen in low- and middle-income countries receive it. So we think it is an area where there may well be gaps. It could be neglected by other funders if, you know, I think in some cases funders sort of maybe miss areas that are horizontal and cutting across a lot of different disease verticals. And the interventions that we've looked at in this space so far have looked potentially cost-effective, and we think may be able to absorb a lot of funding as well. So yeah, it's an overall area that I think we're quite excited about and are just starting to make some grants.

Elie Hassenfeld: So then say a little bit more about what we're actually supporting to get more needed oxygen to people.

Dan Brown: So, this specific grant, the first grant that we've made in this space is actually a, it's quite an unusual grant for GiveWell. It's a very small grant to an organization called PATH to basically scope what support or interventions will be needed to help bring a new oxygen production device called MCOG, which it stands for Medical Ceramic Oxygen Generating, to bring it to market in low- and middle-income countries.

So let me maybe [00:24:00] give a bit of context on, like, how oxygen is produced, and what this device is and why we think it could be important or, like, additive to what's out there already. So there's three main ways that oxygen is produced for use in health facilities. The first is liquid oxygen. This is basically what's used everywhere in high-income countries, it's sort of the gold standard. In low-income countries, this isn't available in most places. It's generally produced in large industrial sites called air separation units that are primarily providing for industry, so for, like, the mining industry or something like that, not really for health facilities. So it's not generally an economically viable option for health facilities unless they're very close to these large industrial sites, basically because it costs a lot to transport liquid oxygen.

So the second and probably main method of oxygen supply in low- and middle-income countries is what's called a pressure swing absorption plant, so a PSA plant, and this is something that's housed at health facilities. It's like a classroom-sized building that has four or five different machines in it. The downside of these PSA plants is that they require a lot of [00:25:00] maintenance. As I say, you have these, like, multiple machines. If any of the parts breaks, then the whole plant is down, and it's no longer producing oxygen. And so they're kind of more widespread, but I think require a lot more maintenance and less reliable.

And then the third method is what's called an oxygen concentrator, which kind of works in the same way as a PSA plant, but it's like much smaller scale. It's portable, it's, you can kind of take it to the ward and directly connect patients.

So this new device that we are supporting PATH to scope out, we've been told it's basically the like first major innovation in oxygen production in decades. And the main benefit of this and potentially why we're excited about it is that it allegedly requires very little maintenance, unlike the PSA plants that I mentioned before, because it has very few moving parts. And so the idea is that hopefully it has less need for biomedical engineers to come and fix things, spare parts to come and sort out one of the problems, and essentially there's fewer disruptions to the oxygen supply to the health facilities and to patients.

The device itself, just to maybe give you a bit of a feel for like what this is and what it looks like, and [00:26:00] I'll give you the proper kind of lay version that's been given to me. I've got like no engineering knowledge, and I like that people were able to sort of dumb it down for me, so I'll give you the same explanation I've got. But it's basically like a small metal box that essentially works like a very hot sieve, this is what I've been told. So it's heated up to about 700 degrees, it pulls in ambient air from the surroundings. There are these things called ceramic wafers, so sort of blocks of ceramic that are about the size of your phone. There's about 20 of them stacked together within the box. And if you apply an electrical voltage to the air as it passes through these ceramic wafers, you're able to in some way apparently extract basically pure oxygen.

So yeah, we think this is a very exciting, low maintenance, new oxygen device that could have applications in lots of health facilities in low-income countries, potentially to replace PSA plants, or alternatively to extend oxygen access to facilities that don't currently have any.

Elie Hassenfeld: Yeah. So this is—I'm sure to anyone listening sounds very different than the kinds of programs that GiveWell has supported the most in our history. [00:27:00] And I'm curious just to maybe dig in a little bit more about the nature of the product right now and what our support is going to. Where do we see our funding and this grant moving the needle on this idea or device getting closer to, you know, being present in the health facilities where people need it?

Dan Brown: The device exists, or a version of it exists. It was actually, I think, originally sort of invented by NASA for applications in space and is produced by a manufacturer called NOVAIR and another called American Oxygen. So the product exists, although I think there's some appetite to potentially design versions of it that can produce a higher quantity of oxygen. But it's not currently being used in facilities. There's some research, implementation research that's about to begin that's been funded by Unitaid, there's a large program called INNOVATE. It's doing lots of separate things within medical oxygen, but one of the things this program wants to do is test whether this product really works in low-income country settings. So, you know, how does it handle intermittent electricity supply, that kind of thing.

And where we fit in is that we're funding [00:28:00] PATH to start doing some thinking around essentially the kind of business case for this product if you're going to bring it to market. So, can this product be produced at a scale and a price that makes it competitive with existing oxygen production technologies, such that facilities and ministries of health would be able to and want to pay for it?

More specifically, the two activities that PATH will be doing using the funding we've provided are firstly, they'll be doing a kind of detailed bottom-up costing to understand how much manufacturing costs of this product could come down as volumes of production scale. Part of that is the usual story of economies of scale and spreading fixed costs over more units. But there are some, like, technical aspects of raw ceramics production that really come into that. I maybe won't go into that now, but essentially if you produce larger batches, you are able to produce a higher proportion of these ceramic blocks that I mentioned earlier that are kind of usable and sufficiently high quality to be put into the product. So one piece is, this sort of, yeah, really understanding how much can cost come down if this was a product that came to market at [00:29:00] scale.

And then the second part of this is trying to estimate the size of the market for this product. So essentially building out, like, what are the use cases, what kind of health facilities would have a sensible case for an MCOG device. So, you know, that might be like mapping out if facilities have access to liquid oxygen, they're probably not going to want to use this. Probably you're looking at places where electricity is reasonably reliable. You know, perhaps the best use cases, places where, like, maintenance is currently very difficult, so maintenance of the PSA plants if you're thinking about, like, replacing PSA plants, there's one option here. So if there's like a lack of biomedical engineers, spare parts, that kind of thing.

So they're going to sort of map the archetypes of facilities that could use a device like this, try to understand how many of those facilities exist in countries that the product might be introduced in, and then potentially work out the willingness to pay for a device like this.

And then you kind of bring those two bits together, so that this is all sort of part of what people often refer to as market-shaping work. So trying to see is there a case here for a kind of win-win between the manufacturer and patients or health facilities, where you can [00:30:00] produce at a scale that brings the cost down sufficiently that this becomes competitive and is enough demand to kind of justify producing at that scale in the first place. So is there, like, an actual kind of use case in the market for this at scale? And essentially by funding this now, alongside the implementation research I mentioned earlier, we think that we could speed up the introduction of this product, a few years down the line when it's ready to come to market.

Elie Hassenfeld: So the place that this stands is the device exists. So it works, in a sense, and then the questions are, there are these different questions, there's a million of them, but two main ones are how does it actually work in real-world conditions? Well, that's being funded by someone else, and then these questions about, can it reach a scale and then a price point where it's actually competitive and facilities will purchase it?

I'm really curious, you know, what happens next after this grant is over? You know, there's obviously, like, failure scenarios, but in a positive scenario, and then what is the timeline to this thing really, like, potentially moving the needle from 30% of patients who require medical oxygen and receiving it to some larger share? You know, how does that happen? What's going to take place on [00:31:00] what timeframe as far as you know?

Dan Brown: I think the implementation research is going to take several years, and I think the aim is for this device to be coming to market at scale potentially, like, four or five years down the road. This is very much subjective guesses, and people who know more than me might disagree with that. But I think that's the sort of mental model I have, it's like several years' time. The kind of immediate next steps of this work are basically to work out, like, what interventions would then be needed to reach that point that I mentioned before of the large-scale production and demand from health facilities and this becoming like a widely used oxygen production device. And that's essentially the next step that PATH would take after they've done those two activities that I mentioned before.

So the kind of thing I have in mind here is, there's a range of different, people refer to as, like, financial instruments in the market-shaping world, but essentially ways of intervening to try to get to a point where a product is produced at scale. So one example of this might be something called a volume guarantee. So this is where an organization, could be GiveWell, could be others, essentially sort of [00:32:00] pre-commit to making a manufacturer whole and covering their costs if the demand didn't turn out as high as we hope and expect it would be. So if they fall short of a sales target that we think would enable them to go down that economies of scale cost curve and produce at a low cost. And in exchange for that, manufacturers would then agree to a lower price that makes it kind of accessible for health facilities.

So those kind of—I mean, that's just a completely hypothetical example, but those kind of interventions, working out how they might work would probably be the next step of a piece of work like this. But I think we're quite far away from that, and there's all sorts of questions on the implementation research side and the regulatory side and so on.

Elie Hassenfeld: One question that I've been thinking about as we've been having this conversation, is why does this require GiveWell support and why isn't anyone else there? And you mentioned that for some of these programs that cut across disease areas, there often isn't the same kind of clear funding stream.

I also wonder whether it is true that like largely there is a gap, from the time, you know, there's a material amount of funding that goes to [00:33:00] earlier stage research and development, and this is a case where I guess the earliest stages of developing a product that works in the lab, so to speak, has already happened, and now it's this like slog of work of, you know, getting this thing out.

And I think we have seen that in some cases in the past where there's been a bit of a delay from the end of the randomized trials to, you know, the program really reaching scale. Like I think we saw some of this with seasonal malaria chemoprevention for malaria that we've supported a lot of via Malaria Consortium, where there was like a pretty long delay from the trials being completed to that very cost-effective intervention reaching scale.

And so I'm just like curious how you think about that in this case, especially, where does GiveWell fit in? Why is GiveWell needed? And maybe, to the extent that this is right, like how common is this sort of gap in the funding ecosystem to get effective products or therapeutics to the people who need them?

Dan Brown: I think we're still learning how big these kind of gaps are. I think this kind of more upstream investment is quite unusual for GiveWell. We're starting to consider [00:34:00] more of these. My instinct is that they do exist for exactly the kind of reasons that you mentioned, we've seen examples of this in the past.

I could also imagine it's kind of maybe a bit less, like, sexy or exciting or something. The kind of really hard work to take a product that works, and you've kind of gone past the initial innovation part, and then it's like actually how do you scale this? And, you know, I think often that can be quite a good fit for an organization like GiveWell.

Elie Hassenfeld: We love doing the boring stuff where you just need to slog through over a long period of time. Sort of our, maybe that's our bread and butter.

Dan Brown: Exactly, yeah. So yeah, I think in this case, we feel reasonably confident from speaking to other people in the area that this wasn't going to be funded on the timelines that we funded it, so we kind of brought for—I think this kind of work would have been funded at some point, but probably further down the line. And so, certainly how we're thinking about and modeling the cost-effectiveness of this grant itself is, like, we're speeding up the process. But I do think medical oxygen in general, I think there was a lot of investment during COVID for obvious reasons. But I think there are potentially gaps there, and that's something we often see with horizontal interventions that are cutting across lots of different disease areas. It sometimes fits less well with funder strategies and things like [00:35:00] that.

Elie Hassenfeld: And then, just tell me what else we're thinking about, you're thinking about in medical oxygen as a whole. So we're talking about this particular device. I know you're excited about this area as a whole. Is this the only thing we're doing? Are there other areas of focus to address this problem?

Dan Brown: We're very, very early stage. There's lots of things that we could do. We are currently investigating some other programs that are looking at systems-wide interventions for pediatric wards, but I think focused predominantly around oxygen concentrators, those portable devices that I mentioned before, and another piece of equipment called a pulse oximeter, which is used to sort of diagnose whether a child is hypoxemic.

But I think that oxygen is quite exciting to me because it is just, like, a very complicated chain from production all the way down to a patient receiving oxygen. So you've got these multiple different methods of producing oxygen, and some of them make sense for some facilities, others make sense for other facilities. You've got a couple of new things coming on the market. There's a whole distribution piece. So once you've, you know, got the oxygen produced somewhere, there's still a lot that goes into transporting it, having the infrastructure and facilities to be able to get it to [00:36:00] the bedside. You then have a diagnosis piece working out whether children are hypoxemic. There's often shortages of these pulse oximeters, the equipment that's used to diagnose oxygen levels in the blood. And there's also a part of this that's just about the triage process itself and getting, you know, this routine part of when a child comes into a hospital, their oxygen levels are checked routinely and you're not missing hypoxemic kids.

And then there's another whole question on what some people refer to as the last yard of oxygen supply. So you may have, you know, a reliable oxygen supply, kids are getting diagnosed, but you've got the oxygen kind of at the bedside in the wall outlet or in an oxygen cylinder, but you may lack the face masks or the nasal cannulas or the like tubing to take it that last yard right to a patient's face.

And so there's lots of potential bottlenecks all the way across the system. And I think, you know, depending on the country and the level of facility and so on, there's going to be a potentially different story as to what the bottlenecks are. So, yeah, we're still learning a lot about this space, but I think there's a lot of potential to find highly cost-effective interventions [00:37:00] across that chain.

Elie Hassenfeld: Yeah. Well, it's really amazing. Really glad that we're trying to do this to address this serious problem. Let's talk about one more grant, support for this randomized controlled trial of work that's being done by Noora Health. So again, maybe start us off by just explaining what they do, what problem they're solving, and then what we're trying to learn from this trial.

Dan Brown: Yeah. So the program that we're supporting is called the Care Companion Program for Maternal and Neonatal Health. It's in India. And this program basically trains family caregivers, so in this instance, that's sort of mothers themselves, but also their mothers or mothers-in-law, maybe their partners, to provide care during the prenatal and postnatal period.

So there's like a range of different practices that caregivers are trained to support. That includes things like skin-to-skin contact with newborns, exclusive breastfeeding, handwashing, dry cord care, so keeping the umbilical cord clean and dry and letting it fall off naturally to make sure it doesn't get infected, but also how to sort of recognize warning signs of illness and know when to bring a newborn back in to [00:38:00] seek care.

So yeah, this is a, we think, a really exciting program because we think it could reduce neonatal mortality, and neonatal mortality rates have been fairly stable over the past three decades and are kind of increasingly make up a large share of under-five deaths.

But as you mentioned, as part of this funding, we're also supporting a randomized controlled trial of the program. And I think we're excited about that because the existing evidence for this program I think is limited to one pre/post study, so actually sort of looking at neonatal mortality before and after Noora Health's intervention, but without a control to compare to. And we know that neonatal mortality rates have been coming down in India generally anyway, so, you know, it's just difficult to know how much to attribute to the program versus changes that would've otherwise happened.

So I think this study will give us more rigorous, more credible evidence on the impact of the program, and we're very excited about the theory of change. We think it makes a lot of sense to us, the organization are already operating at a large scale. And we think that sort of better understanding the effect of the program is probably [00:39:00] one of the big barriers to us funding it at a larger scale.

Elie Hassenfeld: Can you say more about the outcome indicators here that we're measuring? Like, what's the main outcome of interest that you're focused on to see how successful the program is?

Dan Brown: The main outcome for the randomized trial is neonatal mortality. We're actually funding two studies. There's a smaller scale study that, like a kind of pilot study, that will happen first, and in that pilot study, the outcome measures are more around morbidity. So, indicators like jaundice and fever, and that kind of thing. The measures are going to be self-reported. But yeah, they're kind of direct morbidity and mortality measures. Because we think in this case we actually can get a sample size that would allow us to understand effects on that without the costs becoming, like, very, very large.

Elie Hassenfeld: One thing that I think is so interesting just about this whole conversation is that we're only scratching the surface of what, you know, your team is focused on. But even across just these three grants, we're looking at very different kinds of programs that we're supporting. So it just feels [00:40:00] like it spans this spectrum of, you know, product innovation and scale up to, you know, linking into an existing government health system to just doing this almost, like, pseudo-traditional GiveWell work of evaluating a great organization's program to see how cost-effective it is and whether we should scale it. And I don't know, I just think that breadth is so cool.

Dan Brown: Yeah, I totally agree. I love being in the New Areas team, and it's not just the level of the intervention or the type of intervention but also the cause area, right? Every time we start a new grant, it may be, like, a totally different disease, and we have to understand how that works, and work out what the existing status quo for care is, and so on. And yeah, it's just a, I think, a really cool team to work on because you get to learn about a ton of new stuff and have that real breadth.

Elie Hassenfeld: Yeah, I mean, now I'm sad that we only scheduled an hour for this conversation. Maybe we should, you know, have, like, some—I'm like, "I only have about 17 hours worth of questions." But yeah, when you think about the team and the future and what you're focused on, would just love to hear an area that you think could be a major area of focus for GiveWell in the future that's, you know, not currently one of our biggest focus areas but, you know, might be coming out of the work [00:41:00] that your team is doing.

Dan Brown: It's genuinely hard to narrow down because I think so many of these areas are, you know, really do look very, very promising in the sense we've identified some cost-effective interventions. We think there's a lot of room to absorb more funding in future. I think probably the main focus of our time recently has been on family planning, medical oxygen, some work on TB, HIV, and health system strengthening. And we're also considering a scaling up of door-to-door distribution of oral rehydration solution to tackle diarrhea.

So I think there's a kind of core set of things that we focus a lot of our time on where I do think there's a reasonable chance that from the learning we get from these initial grants, we would want to, you know, really double down in our investment of time and funding. And which of those we take off and we end up spending a lot more time on really depends on what we find from these initial sets of grants.

So yeah, I think honestly, the thing I'm most excited about is the breadth itself, and the fact that we can cover so many areas despite having such a small team, and really just learn about a ton of different things, and I think that just maximizes the chance that we do find something really, really promising for GiveWell that turns into a whole new sub-team and a [00:42:00] whole new portfolio of lots of large-scale grants.

And I think the only other thing I'd say with that in mind is that we do really want to grow our team. You mentioned at the start we're hiring for senior researchers for our team. I'm also very excited to try to set up a specific kind of cause prioritization sub-team within new areas. Our process for, like, choosing what we work on is, you know, has to be, like, pretty quick and instinctive and subjective. And I think having a bit more time on that would be really helpful, and also may help us bump up some of the really emerging stuff that we haven't spent much time on at all yet, things like anti-hypertension drugs and applications of AI to global health, and all sorts of things where we're just starting to learn, but I think probably much earlier in our journey. So would definitely put a plug in if people are listening to the podcast and are excited about this to look for those jobs on the GiveWell website.

Elie Hassenfeld: The small team is a bug, not a feature. And we'd like it to be less small as soon as possible because there's just so much to do. Well, it's really amazing what you're doing, Dan, and, you know, to the whole team, so just thanks for walking me and anyone listening through, like, just the tip of the [00:43:00] iceberg of what you're doing. It's really great.

Dan Brown: No, my pleasure. And it really is me sitting here and taking credit for all the work that everyone I mentioned earlier does. They're so genuinely incredibly impressive team, the amount that they're able to get through. So, yeah, all credit to them.
--

Elie Hassenfeld: Hey everyone, it's Elie again. Hope you enjoyed this conversation. I think this work is just so interesting, so exciting to see the way in which we've been able to broaden the scope of what we do, both in terms of different cause areas, but also different ways of supporting programs that help people that are, you know, much broader in scope than what GiveWell is traditionally known for, while also maintaining that core of rigor, evidence, cost-effectiveness, and learning as really the foundational principles that we rely on in everything we do.

You know, we only talked about a small slice of what this team is doing, and this team is, you know, really needs to grow significantly. So we're hiring to grow this team so that we can cover more ground, see more [00:44:00] areas, and hopefully find even more ways to help people cost-effectively in low-income countries.

Thank you as always for your interest and support of our work. We really appreciate it.