One of the most intensively researched areas today is the creation of an effective universal vaccine against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). An effective SARS-CoV-2 vaccine is the best single tool to help the world revert to a more normal situation by extinguishing viral spread without the need for lockdowns and travel restrictions. However, even when a vaccine becomes available, there is likely to be a resource bottleneck, keeping in view the massive need for this intervention. As the researchers say wryly, “With the hope of producing a vaccine in the near future comes the difficult task of deciding who to vaccinate first as vaccine shortages are inevitable.”
This new study by the researchers at Fred Hutchinson Cancer Research Center and the University of Washington, Tacoma and published on the preprint server medRxiv* in August 2020 describes one viewpoint concerning the optimized allocation of the vaccine, showing that under different efficacy scenarios, either the old or the young may be the best candidates for initial vaccination.
The researchers used a mathematical modeling tool along with optimization algorithms combining vaccine effectiveness (VE) with a number of available doses. This led to the construction of an age-stratified model with four parameters: deaths, symptomatic infections, and maximum non-ICU and ICU hospitalizations. These operate under a wide range of assumptions.
For one, they considered children to be at lower risk than middle-aged adults (20-65 years old), with adults above 65 years were relatively more susceptible. Secondly, they assumed that natural and vaccination immunity lasted a year or more. They considered a 20% infection/natural immunity rate at the beginning of the scenario, and the absence of any social distancing. All healthcare workers on the front line are considered to have been immunized.
They defined five vaccination groups based on the currently known association of disease severity and mortality rate with age:
- children (aged 0–19)
- adults 20 – 49 years old
- adults 50 – 64 years old
- adults 65 – 74 years old
- those 75 and older
Combining this with assumed VEs of 10% to 100%, and population coverage from 10% to 100%, they arrived at some definite conclusions.
First, with a VE of 50%, and high population coverage, the pandemic can be significantly controlled.
Low VE, Optimal Allocation to High-Risk Groups
The optimal allocation strategy to bring down the death rate remains the same with VE between 10% and 50%. That is, when the VE is low, older people should be first vaccinated, as they are at high risk for death. If 35% or more of the population is vaccinated, with a VE of 50%, more than half the deaths can be prevented.
High VE, Changing Allocation Patterns
When the VE goes to 60% or above, a threshold phenomenon is apparent; optimal allocation still prioritizes the high-risk groups (those above 75 years), but when coverage increases, vaccine allocations switches to high-transmission groups (those between 20 and 50 years, and children).
Once 60% of the population is infected, herd immunity will be achieved. This means that with a VE of 80% to 100%, 40% of the population must be vaccinated since 20% is already immune. If the VE is higher, younger people should be vaccinated first as they are the primary vehicles of transmission, to maximize population coverage. The rationale is that this reduces the number of spread events, symptomatic infections, and non-ICU hospitalizations. This causes the curve of the epidemic to flatten, and the number of deaths drops steeply.
With increasing vaccine availability, the optimal allocation switches back to high-risk groups again.
With a VE of 60%, if 70% of the population is optimally vaccinated, the pandemic will be completely controlled, that is, zero spread will occur. On the other hand, if the VE is 70%, only 50% vaccination coverage is required.
In order to maintain the number of non-ICU hospitalizations due to COVID-19 below 10% of hospital beds, and prevent ICU overflow, vaccines with a VE of 50% or more are required. If the VE is 60%, the same goals can be attained with 54% optimal population coverage.
Optimal Allocation Differs with Objectives
An important point here is that optimal allocation is more economical than pro-rata (proportionate) allocation with respect to achieving the desired goals. With the latter, at a VE of 60% as above, 67% of the population would need vaccination to attain these objectives.
The optimal allocation strategy is most important when faced with vaccine shortages. When VE is 100%, and vaccine stocks cover only a fifth of the population, optimal vs. pro-rata allocation can avert 32% more deaths. With VE 60% and population coverage of 60%, it is possible to prevent 32% more symptomatic infections using this technique of allocation.
With increasing VE, both pro-rata and optimal allocation converge with increasing population coverage. With 60% or 70% coverage, the high-transmission groups are prioritized to accomplish all of the four objectives.
In other words, the researchers say, “Optimal vaccine allocation differs for different objective functions.” Moreover, the optimizer does not aim to use up all the vaccine. For instance, with a VE of 90%, the optimizer visualized 75% vaccination coverage, which makes sense since, at this point, complete containment will occur. No gains are expected with coverage beyond this point.
One limitation of this study is the expected difference in pre-existing immunity in different parts of the world, due to the changing dynamics of the pandemic. To cover this scenario, they repeated their analysis with 10%, 30%, and 40% of the population being immune at the outset of the study. They found the same pattern, with the only difference being in the threshold at which the allocation group switches.
With a 10% pre-existing immunity level, the switch is observed at 80% vaccination coverage, but at a 40% immunity level, it occurs at 40% coverage. At lower levels of pre-existing immunity, the optimal allocation is for high-risk groups. With higher immunity levels at the start, a more even distribution is observed.
Relationship of Optimal Allocation Methods to Parameters
The researchers looked at how the optimal allocation strategies operated concerning key parameters of viral spread and the course of the natural infection. That is, they compared how it would work if all age groups were equally or differentially susceptible to the infection. They found no significant change in allocation strategies.
As the VE increases to 60% and more vaccines are available (70% or more), more is allocated to children. To reduce symptomatic infections, more vaccines are allocated to children than with the earlier model. When the objective was to bring down hospitalizations, high-transmission groups were favored in this model.
The outcomes were found to be consistent when optimized for four of the least specific parameters. And when the number of infections was much higher at the peak, that is, at 10,000 current infections, as in the exponential growth phase, the outcome was highly similar.
The researchers conclude: “We have identified optimal allocation strategies, and once more information about vaccine characteristics is known, validating our allocation strategies with more complex models is welcome.” It is noteworthy that this model operates on conservative estimates, which means that in real life, even fewer vaccines might be sufficient to extinguish the pandemic. Social and ethical factors must be included in any optimization model to take account of financial, ethnic, and systemic risk factors that could hinder equitable vaccine distribution to high-risk groups.
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.