Vaccination is lagging? (Can’t) Pool me once shame on you, (can’t) pool me twice: shame on me.

The New York Times had a few interesting articles last weekend, trying to answer (at least partially) an interesting conundrum: Why is the case that states have administered only a fraction of the vaccines already distributed to them (about half across all states) while at the same time we hear about mass cancellations of appointment for vaccines due to shortages.

I am writing this post in an attempt to make sense of this. We initially focused on the approval process, testing the different drugs, the manufacturing capabilities, and then the distribution (and the need for a cold supply chain). But one thing is clear: we used the time until the vaccines got developed, tested, and approved to think about distribution, but evidently, very little thought was given to the last step, which is how we are going to actually deliver and administer these in the most efficient possible way.

The complexity of this process is that it includes three key components, each of which is constrained, and solving issues in one is going to expose the problems in the other ones. The three systems are:

1. Manufacturing of vaccines

2. Distribution of the vaccines from the manufacturers to the point of delivery

3. Vaccination: the process of actually giving the shots

The first fact that we should all agree and understand is the following:

"Dr. George Rutherford, an epidemiologist at the University of California, San Francisco, said the most obvious problem with vaccine administration in the San Francisco area was clear: "There's not enough doses, period," he said. "That's it. Everything would work fine if you had enough doses."

So, while we see issues with the administration of the vaccine, the root cause is the scarce capacity of vaccines. To understand why this is the case imagine a world where this capacity was not insufficient; say we had 600 million dosages in the US (more than enough to vaccinate everyone above 16), and so did every other country. Through employers, pharmacies, and primary care physicians, I can see many simple ways in which you can vaccinate most people interested in being vaccinated pretty quickly.  

If you have sufficient capacity of vaccines, the question of prioritization is less of an issue as long as everyone can be vaccinated within a few weeks. Again, this is under the clearly unreasonable assumption of sufficient capacity, but if this is the case, a matter of a few days will not mean life and death.

But here is the first issue: vaccine capacity is limited. This means that some priority should be given to some groups over others. I am not going to get to the question of how to prioritize. We can agree that if the goal is to maximize the reduction in mortality rate due to Covid, age should be the primary criteria, followed by pre-existing conditions, followed by the risk people are exposed to due to their profession. This seems to be somewhat the underlining priority determined through the CDC "1, 1A, 1B, 1C, 2" scheme. One can debate whether people in the medical profession should have received it before older people and whether teachers should have received it before or after. Still, I don't think any of this explains the discrepancy we are trying to solve above.   

To the need to prioritize, one has to add one more topic that exacerbates these delays, and that is equity considerations. I am not here to argue that these should not be accounted for. 

But it should be clear that if the goal were to ensure that the number of people that get vaccinated is maximized, the best way would have been a "First come, first served" system (even within a particular prioritization bucket). The issue is that this disregards the fact that certain groups of people have access to technology and information in a way that is highly correlated with income, age, and race. In a recent survey, most people agreed that speed should not be the only considerations and that equality and equity do matter.  

So how can we explain the impact of these on the over "efficiency" of the system? The main idea is simple: The more centralized the decision making and the more the system pool its resources together, the more efficient the system is. Efficient in reducing waste. Efficient in handling uncertainty (both in supply and demand, i.e., handling no-shows), and thus more efficient in utilizing its resources to maximize their impact. With every decision to make things less pooled and add more constraints, we are reducing the system's efficiency. While the considerations I mentioned until now are ones we tend to agree on: age-based priority and equity considerations, the point I am trying to make is that this lack of central decision making and resource pooling where it's possible is really making things worse.

If you look at the countries that are the most successful in administering the vaccine: Israel is at the top with having more than 50% of the population already vaccinated before the last day of January (the US is at 8.9%, the UK at 13.1%, but France at 2.1%). Israel managed the entire project at the country level and decentralized the actual administration to two HMO-like systems (Israel is a single-payer health care system). Furthermore, even as the priority was followed in making the appointments, people that went at the end of the day to the vaccination centers could get shots if there were any remaining vaccines. While this is a violation of the equity considerations above, it does mean that very few shots had to be discarded due to expiration. In other words, all resources are pooled.

Among the US states: West Virginia is first, South Dakota is third, and New Mexico is fourth. I have not followed West Virginia closely, but both South Dakota and New Mexico manage the entire effort at the state level. A person has to register with the state and is notified where to go. I don't know how it works in every state, but in the state where I reside, Pennsylvania, the allocation is made by the state, but the registration is done at the county level. You can register at the county you work and the county you live in, and you can see how this results in redundancies and mismatches, exactly when you want to be the most efficient. Again, every time you decentralize the resources, you hurt the system's ability to deal with uncertainty, no-shows, spoilage of shots, and ultimately the system's efficiency.

Now, it is clear that centralized decision making is not a panacea. The World Health Organization's initial reaction to Covid and instructions on the fact that masks are not helpful showed that there are very few centralized authorities you should trust. In the vaccination effort, the CDC paid Deloitte 44 Million dollars to develop a system to manage the role out.  

"In May, it gave the task to consulting company Deloitte, a huge federal contractor, with a $16 million no-bid contract to manage "covid-19 vaccine distribution and administration tracking." In December, Deloitte snagged another $28 million for the project, again with no competition. The contract specifies that the award could go as high as $32 million, leaving taxpayers with a bill between $44 and $48 million."

The system is so cumbersome and buggy that even the few states that tried to use it backed away.

I want to go back to my initial point: the main issue is the lack of sufficient capacity of vaccines. If we can solve this issue by expediting the approval of effective drugs such as the Johnson and Johnson one and approving the AstraZeneca one (approved in the EU, but not the US), things will get better. We are ultimately constrained by that. But the constraint is amplified by the lack of resource pooling: while the federal government in the US has demonstrated very little capability to deal with the pandemic (which means that it's not too late to build before the next crisis), states should aim to pool and centralized their resources as much as possible, and potentially even collaborate across states, in an effort to reduce spoilage and maximize the rate by which the vaccines are administered.