The speed at which COVID-19 (C-19) vaccines have been researched and developed is a testament not only to human determination, but also to advances in science, technology, and data and analytics.  Several viable vaccines will soon be manufactured at scale in many places around the world simultaneously. And the United Kingdom has already authorized a vaccine for use, with other countries expected to follow shortly.

Now we have to get those vaccines from the pharma companies’ manufacturing facilities to billions of people around the world in the shortest possible time. This will be one of the greatest logistical challenges we’ve ever faced. And we can’t do it well without advanced analytics.

The challenge ahead

Distributing a vaccine is complex, even in a country like the US with experienced distributors, transportation providers and an infrastructure for determining how many vials should be sent, where and when.  

For C-19, this complexity is magnified by the fact that there are several promising vaccines, most of them requiring two doses. Each will come in multi-dose vials (due to economic and raw-material constraints). Each requires a different time between doses. And, of course, a recipient’s second dose needs to come from the same manufacturer as the first.

Additionally, among the two leading vaccines, one requires an ultra-cold chain (about -100°F), has a shorter shelf-life, and needs more delicate handling than the other.  

If that wasn’t enough, there’s a relatively limited supply of vaccines in these early months. So a series of complex decisions will need to be made to minimize waste and increase availability for prioritized populations.

In total, the US could require about 600 million doses, assuming near-perfect planning and efficient execution. And that’s a big assumption, given that responsibility for the final allocation, planning and distribution will be split among state health departments, each of which has its own way of operating.

<<< Start >>>

It’s the most complex supply-chain challenge since the mobilization of Allied troops to begin the liberation of Europe in 1944—and maybe even more daunting.

<<< End >>>

Analytics: The best medicine

Here’s why the distribution challenge is so formidable. Imagine a state health department places a C-19 vaccine order to the federal government for 500,000 doses. A few days later, the state receives confirmation that it will be allocated 215,000 for that period, as supply is limited. The state currently has a combination of 646 dispensing sites (e.g., hospitals and pharmacies). The population of the Tier 1/high-priority population varies between cities and towns across the state, and the dispensing locations' capacity can only operate for so many hours and vaccinate so many people per day. Some of the vaccines have a relatively short shelf-life and minimum order quantities of 1,000 units—too much for some sites, not enough for others. 

So, where and when do you place the vaccines to ensure not a single injection goes to waste?

Accenture has developed a cloud-based mixed-integer linear programming (MILP) optimization solution, which leverages commercially available software. This solution's objective is to maximize the vaccination rate and immunize the highest number of people in the shortest possible time (subject to all the above mentioned physical and policy-related constraints). The same model can simulate demand and supply changes, leading to decisions on how many additional vaccination sites need to be opened in different locations. 

The solution helps users quickly answer questions like:

  1. How many doses are needed?
  2. Where do we need to have the vaccine (i.e., at which dispensing site)?
  3. When do we need to have the vaccine in the dispensing locations?
  4. What is our population per tier per geographic boundary?
  5. Where do we have an oversupply of doses?
  6. Where do we have the potential for vaccine waste?
  7. Which locations are likely bottlenecks (and potentially need a temporary dispensing location)?

How would it work in practice?

In the following four figures, we show how the solution would work in a diverse and sprawling area like California’s East Bay.  Figure 1 is the starting point. It provides a general overview of population demand by tiers, dispensing site locations, vaccine supply by manufacturer, demand-supply balance, and more.

<<< Start >>>

Figure 1 provides a general overview of population demand by tiers, dispensing site locations, vaccine supply by manufacturer, demand-supply balance, and more

Caption: Figure 1: Illustrative vaccine planning and forecasting allocation overview

<<< End >>>

Figure 2 shows the optimized, recommended allocation plan based on maximizing prioritized population vaccination in the minimum amount of time. This allocation answers when, where, what and how many doses are required—a major step toward effective, efficient vaccination.

<<< Start >>>

Figure 2 shows the optimized, recommended allocation plan

Caption: Figure 2: Illustrative optimized recommended allocation forecast

<<< End >>>

Timing is everything, so Figure 3 also shows the vaccinated population forecasted by tier and period, and the cumulative vaccinated population. In other words, the MILP solution provides an expected measure of success.

<<< Start >>>

Figure 3 shows the expected performance of the program over time

Caption: Figure 3: Illustrative vaccination forecast by population tier

<<< End >>>

Figure 4 shows the potential vaccine waste and dispensing-site capacity utilization to help ensure that as few doses as possible go unused. This element of the solution considers minimum-order shipment quantities at dispensing locations for the different vaccine manufacturers (which could vary from 100 to almost 1,000 doses per shipment).

This view highlights potential bottlenecks and helps decision makers determine where and when to open or expand dispensing-site capacity, and potentially transfer vaccines for sites that don't have the demand to consume the minimum order quantity supplied by manufacturers during the period.

<<< Start >>>

Figure 4 shows the potential vaccine waste and dispensing-site capacity utilization

Caption: Figure 4: Illustrative potential vaccine waste and disposal

<<< End >>>

Distribution: The next Herculean effort

Planning capabilities will be as important to the success of vaccination programs as the vaccines themselves.

Flawless execution is only possible if a reliable planning capability is embraced as a prerequisite to the forward deployment of inventory, as well as the mobilization of distribution infrastructure to send vaccines to thousands of dispensing locations large and small, urban and rural.

In other words, bioscience must be followed by data science. Making the best use of advanced data and analytics capabilities is as essential as the syringes and needles that will deliver the vaccines. Leveraging indispensable models such as Accenture’s will ensure that the Herculean task of vaccinating the world doesn’t become Sisyphean.

<<< Start >>>

<<< End >>>

For more on supply chain topics overall, please visit our supply chain blog.

Special thanks to Jordan T. Loehr, Rubén González-Rodríguez, and Tomás Sánchez-Lombardi

Pierre Mawet

Managing Director – Supply Chain and Operations, Planning and Fulfilment, North America

Ely Colón

Senior Principal – Supply Chain and Operati​ons, Data Science

Subscription Center
Subscribe to Business Functions Blog Subscribe to Business Functions Blog