Vaccine Rollout as Wheeled-Luggage Problem

by AllAmericanBreakfast5 min read13th May 202124 comments

28

Covid-19World Modeling
Frontpage

According to FiercePharma, Pfizer, AstraZeneca, Moderna, Johnson & Johnson, and Novavax plan to produce a combined ~9 billion doses this year. They are collaborating with other companies to scale up production, and have facilities all over the world. They're licensing their product to nonprofit producers, and some have committed to providing the vaccines at cost.

The fact that these companies project to produce enough vaccine to vaccinate the entire world this year is great. Why didn't they have it ready to go earlier?

Note that this is a similar question to "Why did it take so long to invent wheeled luggage?" Why has it taken so long to ramp up production, when the first mRNA vaccines were created in January 2020?

Jason Crawford proposes epistemic standards for this sort of question:

On the other hand, if you want to argue that something could have been invented much earlier than it was, you have to do better than glancing at its high-level concepts or components. You need to rigorously examine every part, material, and manufacturing process, and rule them all out as gating technologies. Any detail, even a minor one, can become crucial—especially when we remember that inventions need not only to work but to be practical, which includes performance, reliability and cost.

We want to identify the fundamental bottleneck for ramping up vaccine supply. I've seen lots of attempts. They usually boil down to the idea that with better planning and less regulation, we could have spent a relative pittance for more production capacity early on, and might have been able to produce the billions of needed doses many months sooner.

That explanation does not meet the Crawford Standard. Instead, we need to rigorously examine every aspect of vaccine manufacturing and rule them all out as bottlenecks.

First, we need to consider planning and regulation failures prior to 2020. What failures of planning and regulation prevented us from setting up sufficient emergency vaccine production capacity before COVID-19 emerged? Could we have had a global network of pandemic vaccine factories on standby back in 2015?

Secondly, we need to consider the practical and material challenges producers must have faced once they'd developed their vaccines. If they'd had the money and governmental support, could they have ramped up production faster than they did? Or would they have faced problems sourcing labor, materials, transport, and storage capacity?

Are these 9 billion planned doses being produced in factories that already existed at the start of 2020? Or have these companies built or renovated additional facilities? How much delay was caused by planning failures and red tape, and how much was a practical problem?  Have these companies had to retrain workers? How hard was that?

By the Crawford Standard, while we can reasonably complain that planning failures and regulation probably slowed the ramping up of production, we need to do a lot more work to assign them the lion's share of the blame.

FastCompany has an article on how Lonza, a vaccine manufacturer, sped up the vaccine production spin-up process from 2+ years to about 8 months. They developed this concept about a month prior to the COVID-19 pandemic. The article is itself a look at some really innovative thinking by a manufacturing company that's actually trying. Good job, Lonza.

Their approach involved building adaptable buildings in advance, which can then be pre-approved and repurposed when they have a specific vaccine production need. My impression is that it still takes a long time to build the adaptable buildings. Once built, though, it only takes 8 months to set up manufacturing within them. Even starting in late January, though, they wouldn't have been ready to go until September 2020.

This is as fast as a company capitalizing on a whole new approach to vaccine factory construction could go. It seems reasonable to expect that repurposing existing infrastructure not specifically designed for adaptability would take even longer.

I have no idea how complicated the work of manufacturing vaccine is. But given that we've seen a major batch failure in the J&J vaccine after a quality control failure, it's not exactly flipping hamburgers. So on top of the time and labor it takes to build a factory,  with all the complex equipment that also needs to be manufactured (perhaps requiring additional factories to build that!), there's also the challenge of finding or training a high-skill workforce on a global scale to operate a bevy of new vaccine factories. You can't buy this stuff at Target.

We can cynically assert that in fact, governments are acting to actively harm vaccine production as some sort of signalling game. What about the more obvious-seeming explanation that they would want to take credit for a strong pandemic response? That vaccine manufacturers want to make money and help those bureaucrats get some credit for success?

It seems like a more reasonable explanation to expect that regulatory and other governance failures are responsible for only a part of the rollout. Practical issues must be a big part of the explanation as well.

If true, this helps explain why human challenge trials might not have helped as much as you'd expect. If even Lonza wasn't going to be ready to produce vaccine at scale until September, then an HCT wouldn't speed things up, at least not by much.

By the Crawford Standard, the burden of proof is not on me to demonstrate that we couldn't have gone much faster in producing vaccine.

To claim we could have gone much faster, if it weren't for X, is to claim that there was no other factor Y that was a more fundamental bottleneck than X. The burden of proof is on people making arguments of this kind. They need to investigate the details of the manufacturing process make a careful case that this could have been accomplished faster. Alternatively, they need to supply a better epistemic standard than Jason Crawford's.

This would undoubtedly produce useful insights, and I think it's a more productive way to steer the conversation than complaints about political blame-mongering. That blame-mongering might just be a reaction by politicians who are fundamentally not able to make things go faster, yet who are expected to do so and are unfairly blamed for not having a magic wand. I would be much more interested to learn about what goes into building a new vaccine factory.

28

24 comments, sorted by Highlighting new comments since Today at 7:39 AM
New Comment

Biotech billionaire Winfried Stoecker invented and used his vaccine based on technology that's easy to scale up in April 2020 for himself because he wanted a vaccine. He waited. Then he vaccinated his family. He waited again and vaccinated a bunch of friends. Then in September he suggest to the RKI in Germany to just vaccinate everybody in Germany by the end of the year. Then the RKI first ignored him and then sued him.

So the timeline for Germany to vaccinate without government interference would be something like end of 2020. However if the government wouldn't have been in the way of progress Stoecker would likely have been interested earlier into vaccinating people besides himself, family and friends. If that would have happened end of April, German being well vaccinated would be in end of August or September. If you would have wanted to do human challenge trials first maybe + 1/2 months. 

The vaccine might have be a bit less effective then the one's we have now but also with less side effects then the one's we use. 

This leaves the question of why those companies are producing vaccine's with processes that aren't easily scalled up. This is a similar question as to why Scott Siskind couldn't get any melatonin in his hospital. It's the pharma ethos to use technology that can be protected with patents to produce products that can't simply be copied by their competition. 

This means using complicated new technology instead of tried and tested and easy to produce technology. 

But given that we've seen a major batch failure in the J&J vaccine after a quality control failure, it's not exactly flipping hamburgers. 

The main difference is that hamburgers would never be approved by the FDA if the FDA could decide about whether to approve them. Producing hamburgers to the quality standards that are involved isn't easy either. Just because the batches didn't meet the high standards that Johnson&Johnson had for them doesn't mean that they wouldn't be useful for vaccinating people.

I wouldn't be surprised if the standards are significantly higher then for the average drug that goes over the counter at our pharmacies and that is produced by an Indian or Chinese generics company which doesn't care much about quality control. 

The burden of proof is still (by the Crawford Standard) on you to demonstrate that Steocker's vaccine was "easy to scale up." I agree with you in my OP that regulation and ineffective governance are partly to blame for production delays. What's in question is the proportion of the blame.

Yes, even small delays are highly consequential in an exponential growth scenario. But determining the proportion of delay due to governance vs. inevitable pratical issues is the question at hand. The point of this post is that to determine the delay due to governance, you need to thoroughly assess the delay due to practicalities, and also investigate that there are other issues beyond these two categories. Simply naming yet more examples of governance failures doesn't really help address this.

The burden of proof is still (by the Crawford Standard) on you to demonstrate that Steocker's vaccine was "easy to scale up."

The argument here is that a biotech billionaire is likely able to know what can be achieved in a 3-month timeframe with his technology. 

But lets look at his description:

Man nehme dreimal 15 Mikrogramm rekombinante RBD der S1-Untereinheit (Arg319-Phe541) für eine Person. Als Adiuvans habe ich Alhydrogel von InvivoGen verwendet: Ordentlich durchschütteln und davon 200 Mikroliter mit der Tuberkulinspritze aufziehen. In eine größere Spritze 10 Milliliter Kochsalz aufziehen und die 200 Mikroliter dazugeben, mischen. Davon 500 Mikroliter pro Schuss, mit denen man seine Portion Antigen vermischt. Alles hübsch steril.

Mit einem einzigen 2000-Liter-Reaktor kann man 45 g Antigen pro Tag produzieren, das würde für 1 Million Personen reichen. Mittels eines Hochdichte-Kultursystems schafft man die fünffache Menge. Innerhalb eines halben Jahres könnte man Impfstoff für 80% der Bevölkerung Deutschlands in einem mittelgroßen Laborraum produzieren.

Kochsalz is salt or NaCl which is easily available in very large quantities. 

The antigen production seems to require a 2000-liter reactor for producing enough for 1 million people per day and he suggests that more advanced technology can do 5 times as much in a middle sized labatory room, so that doesn't seem to be a constraint.

That leaves the Alhydrogel of InvivoGen. Current cost seems to be 248,00 € for 250ml which is a bit more then what you need for 1000 dosis. That leaves the question of how fast that can be scaled up. It's based on Aluminium hydroxide which needs Aluminium is easily available and turning it into Aluminium hydroxide is not very expensive. 

In addition to just using Aluminium hydroxide InvivoGen does something they call ultrasonication with it. 

I'm not sure how we would go about determining how fast it is to scale up Aluminium hydroxide to 2 million for Germany's 80 million inhabitants or 200 million worth of it for the world population. 

Vaccine production is, as far as I can tell, not exponentially increasing over time. I don't have data on the world as a whole but in the UK at least vaccinations-per-day isn't even increasing, it's holding steady!

This is not what you would expect to see if vaccine production was a priority, if it was being increased as fast and as much as possible, cost be damned. If that's what was happening we should see exponential growth in vaccine production.

The pattern of production we see (steady or slowly increasing) is what we would expect to see if cost was a major constraint for vaccine production, and in particular, if vaccine producers have high fixed costs (the factories, equipment, etc) that they are trying to amortize over many months of production. Better to run one factory for ten months than ten factories for one month, since the latter plan costs almost 10x as much.

I've heard that vaccine producers can't charge high prices and instead have to sell at low prices negotiated by governments that ban them from selling to anyone else. If that's true, then that also fits nicely with this theory of what's going on.

So, is this not what's going on? I'm curious to hear counterarguments.

You are right that some vaccine patent-holding companies are selling at cost. I don't know if that means that vaccine producers are manufacturing them at cost. Companies like Pfizer don't necessarily do all their vaccine production in-house. They contract it out.

Operation Warp Speed planned 300 million doses by January 2021, but fell short. According to Airfinity data, 413 million doses have been produced worldwide as of the beginning of March. If these companies' plans bear out, we'll have 9 billion doses by year's end. That looks like exponential growth (or probably sigmoidal) to me.

The key thing to remember here is that vaccine production is not the same as vaccination.

By the Crawford Standard, it's not enough to find a piece of supporting evidence (i.e. vaccination rates) for a claim that the fundamental bottleneck for faster production (i.e. cost) has been identified. You have a hard job to do. You need to conclusively demonstrate that there are no other factors that might be major bottlenecks. Unless you've looked in detail at the economic and logistical challenges of building more factories and manufacturing more vaccines, you can't rule that out as a major bottleneck.

I didn't mean to suggest they were manufacturing at cost; quite the opposite! I was saying they weren't going as fast as they possibly could, e.g. as fast as they would go if they were being paid $10K per vaccine -50% for each month of delay.

Thanks for the data point about doses manufactured so far; that does indeed look like they are ramping up production, though idk if it's exponential, I'd want to see a graph. This is good evidence against my theory.

https://www.nytimes.com/interactive/2021/world/covid-vaccinations-tracker.html is a graph of vaccine doses administered worldwide. It looks like in the beginning there was exponential growth and now there's linear growth. 

Right, but again, I’m talking in the OP about production, not administration.

If production were exponential but administration were exponential-then-linear then there should be massive stockpiles of unused vaccines by now. Are there?

The Crawford Standard simply states that it's difficult to assert something could have been invented earlier but wasn't, it doesn't tell you why it couldn't have been invented.

The why could be lack of enabling technology, time taken to ramp up production, or regulation.

Given we know that regulation was the limiting factor here, there's no reason to assume other things were also a limiting factor - i.e. we know that vaccines were produced as soon as they were approved. What's the chance that all the other limitations (e.g. ramp up time) happened to take exactly (or close to) the same amount of time?

One possible explanation is that, with emergency approval, the FDA maximized the time they took to study the drug, without actually shortening significantly the time it took to deploy it. They timed their emergency authorization to coincide with the readiness of manufacturers to have adequate vaccine produced to begin distribution.

Sure anything's possible - I haven't seen any evidence that they did that, nor do I think they ever claimed they did that, nor does their behaviour match that strongly. E.g. they still haven't authorized AstraZenica for God knows what reason, but it's currently being produced at a very decent rate.

If vaccine production in the USA is no longer the bottleneck (which it isn't, which is why we're shipping it overseas now), then it makes some sense for regulators not to approve it. At least if their model of vaccine hesitancy is "the more ostentatiously we ban and slow down drug approval, the safer the drugs we do approve will seem." AZ vaccine gets to be the "proof" that the FDA's emergency approval really means something (i.e. they don't just approve whatever).

Plus, we don't have any trouble shipping it overseas to meet demand in areas where vaccine availability truly is a bottleneck, because we're shipping them a vaccine deemed to be "second-class."

But that's pure speculation that I made up off the top of my head.

My core argument is this:

  1. Ignoring the possibility that we could have scaled up vaccine manufacturing capacity (not research capacity) years before COVID struck....
  2. Some people think they know what the primary bottleneck was, or is, to having more "vaccines in vials" (as opposed to shots in arms). Governance is one possibility. Practical challenges in scaling up manufacturing is another. There may be yet other hypothetical bottlenecks.
  3. By the Crawford principle, the burden of proof is on those who claim to know what the "vaccines in vials" production bottleneck. They must prove that it was not some alternative bottleneck, with an exhaustive examination of the details of each possible bottleneck.
  4. Until somebody does that, we should regard each hypothesis of a vaccine production bottleneck with open-minded skepticism, and assume that there were some significant devils in the details that would have made it practically difficult, on a civilizational level, to produce vaccine earlier than we did.
  5. The most useful way we can find an answer to this question is by looking at the most neglected possible bottlenecks.
  6. My anecdotal observations are that there's much more attention given to the governance-bottleneck hypothesis than to the manufacturing-bottleneck hypothesis. I assert that addressing this imbalance would be more useful than yet more investigations and harangues on the governance-bottleneck hypothesis.

I disagree with point 3.

Given that we can show governance was definitely a critical bottleneck (production proceeded as soon as governance allowed), why is the burden of proof on us to show that no other bottlenecks happened to also be critical. My prior would be that it is unlikely that 2 bottlenecks happened to be about the same time. Not crazy unlikely, but in the 20/30% range.

In other words, bad governance was definitely an issue and must be fixed. Maybe there's some other issues that also have to be fixed, in which case, sure, bring me some evidence of them and we'll work on solving them.

Given that we can show governance was definitely a critical bottleneck (production proceeded as soon as governance allowed)

One possibility is that the FDA, already concerned about vaccine hesitancy, put off approval until they knew the manufacturers had scaled up adequately.

I think under your hypothesis, the timing was something like this:

Producers scale up and have lots of vaccine ready to go months prior to December (when Pfizer was approved). The FDA refuses to approve the vaccines for months, despite the producers being ready to go with a big stockpile and production capacity, until December. Once they give emergency authorization, the vaccine distribution begins immediately on the stockpile they'd built up.

An alternative hypothesis is this:

Producers can make enough vaccine for early trials, but don't manage to scale up to produce millions of doses until December. The FDA takes the opportunity to gather data on safety and efficacy until that time. However, they don't want to look like they're issuing emergency approval just because the vaccine's ready to go. They want it to look like it's because they have reached satisfaction with the safety and efficacy data. So that's the line they put out to the public. Yet in reality, they give approval right around the time when producers, distributors, and vaccination sites are ready to go.

It's really not obvious to me that one or the other of these is correct. In the former, regulatory decision-making is the bottleneck. In the latter, regulators are striving to make regulation have the appearance of a bottleneck, when in fact the true bottleneck is manufacturing capacity.

why is the burden of proof on us to show that no other bottlenecks happened to also be critical.

Manufacturing a novel vaccine at mass scale, and setting up the cold-storage and clinic sites to distribute and inject it, was a huge logistical project. It has lots of technical details. The FDA knows how to evaluate safety and efficacy; it's what they do. By contrast, the manufacturing was a novel challenge. It probably put strains on the global supply chain and the world's vaccine manufacturing expertise. Overall, it's a more complex and novel challenge than testing and regulation, with far more ways for delays to crop up.

So when there's a reasonably plausible explanation for why governance and manufacturing bottlenecks appear to have coincided (with the root cause actually being manufacturing), and we also know that the manufacturing is the novel/complex part, and we can more obviously see how politicians and manufacturers would both stand to benefit from an efficient rather than an inefficient process, it seems like you should at least take that possibility very seriously?

we can more obviously see how politicians and manufacturers would both stand to benefit from an efficient rather than an inefficient process

In a world where it would have been possible to vaccinate every person on earth for the cost of $1000 because we have very efficient technology for vaccination nobody would have paid billions for vaccines and those companies wouldn't have made a lot of profit. 

Moderna and Pfizer used the most complex technology for producing vaccines and are also paid more per vaccine then other companies (and those contracts were made before we knew anything about effectiveness). 

Using simply solutions is just not the way profits are made in biopharma. 

An alternative hypothesis is this:

Producers can make enough vaccine for early trials, but don't manage to scale up to produce millions of doses until December. The FDA takes the opportunity to gather data on safety and efficacy until that time. However, they don't want to look like they're issuing emergency approval just because the vaccine's ready to go. They want it to look like it's because they have reached satisfaction with the safety and efficacy data. So that's the line they put out to the public. Yet in reality, they give approval right around the time when producers, distributors, and vaccination sites are ready to go.

If that was correct we'd expect that FDA approval process tends to vary by how quickly manufacturing capacity can be ramped up. The key criticism I have of the FDA is not allowing challenge trials. If you are correct and that was intentional to give time for capacity to ramp up, we would expect to see drugs which can be ramped up more quickly use challenge trials - but that's not something we've seen. If anything this was a ridiculously fast approval, for a process that usually takes years.

Furthermore I find it highly unlikely that producers would ramp up production just as quickly before results of trials are in as afterwards. So by not allowing challenge trials, less was invested in ramping up, so the ramp up itself (if it was a critical bottleneck) would have taken longer.

and we can more obviously see how politicians and manufacturers would both stand to benefit from an efficient rather than an inefficient process

The overwhelming evidence I've ever seen is that politicians and government orgs are highly inneficient. My prior on them being efficient here is extremely low.

If that was correct we'd expect that FDA approval process tends to vary by how quickly manufacturing capacity can be ramped up.

I think the right reference class here is emergency use authorizations, not FDA approval. Under my model, yes, their emergency use authorization timeline would vary by the speed of the manufacturing ramp-up. I might expect a global pandemic context to be a special context even among EUAs, but let's not add epicycles at this point.

Another factor to consider is that the interplay of manufacturing and regulation may vary depending on whether the treatment is among the first, or whether it is a latecomer. For example, Novavax only got preliminary data on its vaccine in Jan 2021. It's only started scaling up after that. When they got the prelim data, the CEO said it would take 5-6 months to scale up to 150 million doses.

So why is it that they can get to that scale in 5-6 months, if it took 8 months+ to scale up vaccine production for Pfizer or Lonza, as I claim? One possibility is that their vaccine is easier to manufacture than others. Another is that they're able to use global production capacity that's been scaled up over the last year and a half.

A third, of course, is that it really doesn't take more than 5-6 months to scale up to 150 million+ quality-controlled doses if you're adequately incentivized. But then, we'd expect Pfizer to be cranking out that much right now. The number is closer to 20 million per month, even though they're a much bigger company than Novavax.

Furthermore I find it highly unlikely that producers would ramp up production just as quickly before results of trials are in as afterwards.

Novavax's decision to start scaling up only after getting their prelim results is some evidence in favor of your argument. Likewise, the fact that, back in July 2020, Operation Warp Speed offered them $1.6 billion to scale up, but only after their data was in showing that the vaccine works.

It makes perfect sense that companies are risk-averse, and don't want to invest lots of money in scaling up without good data showing it's not going to be a wasted investment. An HCT would give them that confidence faster.

The overwhelming evidence I've ever seen is that politicians and government orgs are highly inneficient. My prior on them being efficient here is extremely low.

The important comparison here is the inefficiency of the politicians vs. the difficulty of the manufacturing process. It doesn't matter how slow the politicians go if the manufacturing scale-up process is even slower than that.

Anyway, I find that overall, there's reason to think and some evidence in favor of regulation being an important source of slowdown. I find your argument that HCTs could have reduced uncertainty around trial outcomes, thus accelerating scale-up efforts, particularly compelling.

At the same time, the discrepancy between Pfizer and Novavax's current and projected scale-up timelines, even months post-EUA for Pfizer, makes me think that manufacturing must also be an important source of slowdown. That may be a tighter constraint than improving regulation. We can't kick science and manufacturing to make it go faster, but sometimes we can kick the government to make it go.

From a political perspective, then, it makes some sense to focus on the failures of government. From an epistemic standpoint, though, I think we really do need to spend more time understanding the manufacturing problems.

Finally found the key quote, from an article in Nature.

"Could other companies pitch in to manufacture more? Making mRNA vaccines has a simplicity about it, but scaling up is tricky, says Zoltán Kis, a chemical engineer at the Future Vaccine Manufacturing Hub at Imperial College London (see ‛Messenger RNA: the science of speed’). Because it is a new process, there’s a shortage of trained personnel. “It’s very hard to find these people who are trained and also good at it,” he says.

But the key bottleneck in mRNA-vaccine manufacture is a worldwide shortage of essential components, especially nucleotides, enzymes and lipids. This is because relatively few companies make these products, and not in sufficient numbers for global supply. Moreover, these companies are proving slow to license their manufacturing so that others could do this."

Once again, this updates me even more strongly in favor of manufacturing, not governance and regulation, being the key bottleneck.

"First, we need to consider planning and regulation failures prior to 2020. What failures of planning and regulation prevented us from setting up sufficient emergency vaccine production capacity before COVID-19 emerged? Could we have had a global network of pandemic vaccine factories on standby back in 2015?"

I am of the opinion that such failures are more than sufficient, given that there have been successful human trials for mRNA vaccines (for cancer immunotherapy, mostly, AFAICT as a layman) dating back to the early 2000s, and even more so to the early 2010s. The fact that designing the Moderna vaccine took two days and could be scaled up in a matter of months once it became clear that the vaccine would work and get used implies that all the critical technical and production problems already had known solutions. It would require a very large coincidence for there to have been very significant problems that just happened to get solved or become solvable within a few months of the first pandemic of this scale in generations.

But no, I don't think there is a reasonable regulatory and planning framework that would have vaccine factories "on standby" in case of a pandemic. That would be very inefficient. Rather, I think a better regulatory and planning framework would have long since approved and scaled up at least a few and possibly many mRNA vaccines/other therapeutics, such that the factories that were already in use and operating profitably could be quickly repurposed or duplicated. This version of Earth would already have established protocols for testing and approving mRNA therapeutics, reducing the hand-wringing CYA antics of our regulatory agencies in the face of what outsiders saw as clear evidence of safety and efficacy. It would also have already built public trust and awareness of mRNA therapeutics, reducing hesitancy in much of the population. 

Personally I really don't care whether the world-as-it-was-in-January-2020 could have developed a vaccine more than a couple of months faster given better emergency regulatory changes or government funding or whatever. The correct solution to a pandemic is to have a better designed our regulatory infrastructure, and the innovation culture built around it, going back decades. We could have recognized that the FDA correctly declined to approve thalidomide in the 1960s instead of ignoring that and adding new hurdles for subsequent drugs. We could have learned from the AIDS crisis that it really doesn't matter what group first suffers from a disease. We could have not thrown away the already-developed well-regarded pandemic playbook the US developed within the past decade.

This doesn't seem right to me - if it takes 8 months minimum to turn over a factory to produce a new vaccine, how come there's a reasonable (if not high) rate of production within a few weeks of each new vaccine coming on board? Had factories already started turning over for each individual vaccine just in case 8 months ago?

The 8 months is mostly turning a building that wasn't a factory that could produce mRNA vaccines into a factory that can produce mRNA vaccines. The decision to build those factories was indeed made before the vaccine was approved.

[+][comment deleted]7mo 3