Where have all the vaccines gone?

The year 2020 was unlike any other in our memory, a year when many countries around the world have been paralyzed by the COVID-19 pandemic caused by the SARS-CoV-2 virus. The year 2021 was supposed to be different – the turning of a new leaf, the beginning of the end of the pandemic. Yet should we really be surprised when the beginning of 2021 turns out to be qualitatively similar to the end of 2020?

As I write this in late January from the United States, much of the recent media attention has focused on the distribution and administration of two newly approved vaccines against COVID-19. The development of the vaccines seems like a miracle but really is the culmination of decades of previous work combined with a lot of funding and time over the last year. But the distribution of the vaccines is another matter altogether, with nearly everyone in agreement that the vaccine supply chain has failed us [1].

But why has the distribution and administration of the vaccine failed so miserably? And with current reporting that more than 40 million doses have been distributed and less than 20 million doses administered into people’s arms…where exactly is all that vaccine? Why is it so slowly making its way into the arms of those who need it? Most importantly, what will need to be done to fix it?  

The Vaccine System

Before delving into the problems and solutions, let us begin with the status quo. The public health system in the U.S. is remarkably decentralized, involving a spider web of local or regional health departments, healthcare providers and federal agencies. During the H1N1 pandemic in 2009-2010, the Centers for Disease Control and Prevention (CDC) was the ringmaster, leading the coordination of health departments in “jurisdictions” representing states, territories and a small number of cities.

The influenza vaccine for H1N1 was produced and ready for distribution by October 2009. Throughout fall 2009, the CDC received shipments of the vaccine at distribution centers (DCs) managed by McKesson, and the CDC allocated a limited supply to each of the health jurisdictions. The health jurisdictions would determine exactly which providers would receive how many doses from that allocation and submit the orders through the federal vaccine system VTrckS.

Vaccine was shipped out from the DCs, in most cases to arrive directly at the locations of providers (e.g., physician offices, clinics, hospitals, employers, universities, pharmacies, health departments, etc.). Providers were responsible for administering vaccine to individuals, reporting administration of doses to the state, and the state in turn reported to the CDC. Some states used their immunization registries (originally built for children’s records) to record and report each vaccine administration, while others did not. The CDC also created a pharmacy initiative to allocate doses directly to commercial pharmacies.

There were initial challenges in the H1N1 vaccine distribution (delayed vaccine, limited quantities, difficult to find) but ultimately the system distributed more than 100 million doses of the vaccine in less than four months, vaccinating more than 90 million people, beginning with priority populations and continuing until everyone had the option of a vaccine. That pandemic had some different characteristics (less infectious, less fatal, vaccine demand declining over time, one vaccine dose required and no ultra-low cold chain). Yet there is much that can be understood about the current vaccine distribution by understanding the decentralized system.  

Why is the COVID-19 Vaccine Different?

So where is the 50% of the COVID-19 vaccine that has been distributed but not yet administered? Well, it’s complicated, likely at a number of distinct locations for several different reasons. There is not publicly available information on where it is, but I can conjecture what is happening.

First, recognize that the reporting of supply is of doses “distributed,” which really means shipped. The reporting of administered doses is allowed up to 72 hours afterward. Thus, a portion of the gap is because the two values are not a direct comparison. Let’s assume two days to ship and two days to record.

What is the lead time to administer doses that have just arrived at a provider? I expect that most of the initial providers are spreading the appointments out over one week, keeping in mind that not much happens on the weekend. Thus, I think another three days could be added for the lead time to administer. We now have one week from reported shipment until expected recorded administration [2].

Operation Warp Speed recently announced that the inventory associated with second doses had been shipped immediately rather than being held for later. Using the raw data from the CDC [3], I estimate the average amount shipped every seven days was 10.6 million over the last three weeks [4]. Accounting for this small amount of lead time for transportation, administration and recording means that the percentage administered would be closer to 75% of doses distributed. That looks better, doesn’t it? 

Supply Chain Factors

Other factors in the supply chain can help explain the gap further. Any supply chain can have demand uncertainty, and buffer inventory or leftover inventory may result. Many locations seem to have high demand in the hospitals in cities, while some reports indicate that vaccine hesitancy may be higher [5] among residents or staff in rural nursing homes. Addressing vaccine hesitancy can take time and education. We haven’t even touched the perishability of the vaccine. There are reports of leftover vaccine [6] at the end of the day that may be wasted, e.g., because it was brought to room temperature for use and there are no more people in the priority population to receive it. This is not likely to be a huge component of the gap (especially given the recent attention to it), but it may contribute a little.

Of course, that does not explain why North Dakota (ND) and West Virginia (WV) are performing better than other states. One factor for WV is that the state partnered with community pharmacies [7] for their long-term care program. This initiative is going faster than the federal pharmacy partnership because WV did a lot of upfront planning before vaccine shipped.

Several interesting elements appeared when I reviewed the ND vaccine plan [8]. The state: 1) has end-to-end visibility of inventory until it is administered, 2) estimated the vaccine demand to be served for each provider using seasonal influenza vaccination doses administered, 3) surveyed to understand vaccine hesitancy by location, and 4) automated the vaccine administration recording. ND is able to match supply and demand much more accurately than many states. 

Other Challenges

Many more challenges I have not addressed relate to expansion of eligibility, potential inequities in access to the vaccine, accuracy of forecasted supply and upcoming shipments, performance measures (speed) not matching the goals of the program (reducing mortality, morbidity, inequities and keeping society functioning) and more. The system will be even more complex in the upcoming weeks as vaccine is shipped to thousands of locations, accounting for the second dose shipments becomes more complicated, and additional vaccines may be approved. 

What is the Fix?

What needs to be done to improve the distribution and administration? For jurisdictions that do not have inventory visibility, data integration solutions can be set up to provide that visibility. Jurisdictions may want to consider having a “master waiting list” for people to be notified when there are new shipments in order of their priority (similar to student registrations at large universities). Increasing the supply would help, especially as the number of shipping locations increases. Addressing vaccine hesitancy will continue to be important, and addressing the equity in vaccine distribution and administration should also return to the forefront. Mass vaccination clinics can quickly vaccinate populations, although if the goal truly is to reduce mortality and morbidity there should be processes to manage the demand.

The funding allocated in December 2020 to jurisdictions may help in some of these elements, but it will take time. There are systemic problems, and I don’t see a lot of low-hanging fruit to quickly fix them while focusing on efficiency (speed), effectiveness (outcomes) and equity (valued in public health).

In the meantime, we can all help slow down the pandemic by buying ourselves a little more time and demonstrating a little less panic. Consider getting a screening test for COVID-19 even if you are asymptomatic, wear your mask or face covering anytime you are around someone not in your immediate household, and don’t breathe each other’s air. 

Notes & References

1. By “us,” I mean those currently living in the U.S. You could also include those living in some other countries (e.g., France comes to mind as a location currently struggling with vaccine distribution) but most of my remarks will be driven by the U.S. systems.
2. For the long-term care program [9] coordinated with commercial pharmacies, this lead time is even longer. The CDC is currently reporting doses administered but not doses distributed [10]. A portion of the doses distributed comes from the jurisdiction allocations. If commercial pharmacies are slow, then states may appear slow at administration even though they no longer have control of those doses.
3. https://data.cdc.gov/Vaccinations/COVID-19-Vaccine-Distribution-Allocations-by-Juris/saz5-9hgg/data
4. As a rough estimate, I am using all first dose shipments listed for distribution from 1/1/2021 to 1/25/2021 with second dose shipments 21 days later for Pfizer and 28 days later for Moderna associated with the first doses given the week of 1/10/2021 and 1/4/2021, respectively. The latter two sets of doses were originally planned for distribution the week of 1/31/2021 and 2/1/2021.
5. https://www.wsj.com/articles/nursing-homes-grapple-with-staff-hesitant-to-get-the-covid-19-vaccine-11608477474
6. https://www.propublica.org/article/covid-vaccine-wastage
7. https://www.npr.org/sections/health-shots/2021/01/07/954409347/why-west-virginias-winning-the-race-to-get-covid-19-vaccine-into-arms
8. https://www.health.nd.gov/sites/www/files/documents/COVID%20Vaccine%20Page/Covid-19_Mass_Vaccination_Plan.pdf
9. https://www.npr.org/2021/01/05/953558314/covid-19-vaccine-may-not-get-to-long-term-care-facilities-quickly-enough
10. https://covid.cdc.gov/covid-data-tracker/#vaccinations-ltc