What the Numbers Say: Endemicity - What's Next?
On March 10, Michael Li (University of Alberta) moderated a panel of experts on the mathematical approach to COVID-19 endemicity. Hear what the numbers are saying about the upcoming removal of masking mandates, future variants and the role mathematics can play in future public health emergencies.
Q: How would you define endemicity and how do we know when we’re there?
Steven Hoffman: On the one hand, it's a technical term in epidemiology that just means it's a disease regularly found in a certain population or a certain area. There isn't a clear threshold, like a number to ascribe whether something is a pandemic versus an endemic. What’s really interesting about it, however, is that it's also a very social term in that we, as a society and societies around the world, are basically deciding whether we are in an endemic phase or about to become an endemic phase.
What's [also] really important, right up front, is to debunk a myth that I see a lot in the news media about endemic being a good thing. In many respects, when a disease becomes endemic it means that it's here to stay. Too often in the news media, it becomes associated with all the freedoms or the restrictions or the protections that have been put in place over the past couple of years, and that it suddenly means we get back to normal.
The reality is that if we are in an endemic phase where this is here to stay and we have to live with it, it's more reflective of raising the white flag of surrender. [It suggests] that we tried to do something humans have never done, which is to beat back a global pandemic using new technologies and new approaches to protect millions of lives. We are, however, at a place where there's so much virus circulating that we are on a path towards endemicity. And that's not a good thing, even though certainly in the news media and with our lives, it does mean that we just have to learn to live with that.
Amy Hurford: I think about this from a mathematical point of view, and I think of endemicity as when you have a relatively flat number of cases. The underlying mechanism here is often that immunity at the individual level will change through various factors like births and vaccinations, but at population level it is relatively constant.
Q: Is there any sense of what our future will look like? Will immunity wane and new variants emerge? And will we require frequent booster shots?
Jane Heffernan: An infectious disease can be ever-changing, so we can expect to see new variants arise. For this, the probability of seeing a new variant arise is going to depend on the number of people infected. The more people infected, the higher the probability that a new variant will arrive. And certainly, as pathogens evolve over time, they naturally select for fitness characteristics that allow them to be more transmissible or have more asymptomatic infection so that they are more transmissible because they want to be replicated. It's hard to project exactly what the variants will look like, but I will say that with the vaccines and the new vaccine technologies that can be revamped in a pretty fast timeline, we might be able to get new vaccines that will be able to quickly generate immunity in individuals that are vaccinated against these new specific variants.
Q: Would endemicity promote new variants? And could Omicron hasten the transition from a pandemic to an endemic state?
AH: Yeah. I think endemicity would select for new variants and I expect new variants to emerge. That's what we've seen happen over and over again. And the only way that's not going to happen is if we manage to eradicate COVID and I don't think that's going to happen.
I had to take this idea of constant immunity – which comes from the mathematical models, but is an epidemiological definition that overlooks evolution – and modify it. Maybe one of the take-homes from the pandemic is that we need to be thinking more about evolutionary epidemiology and not just thinking about pathogens that are static. We also need to be thinking about how our actions, like using antimicrobials or therapeutic agents, are going to select for certain characteristics and how our behaviour might select for different things as well.
Could omicron hasten the transition from pandemic to endemic? I think under my definition, yes, because I'm looking for a levelling-off of immunity, and omicron is infecting a lot of people very quickly so there's a lot of natural immunity building.
Q: How do modellers distinguish whether COVID is the direct cause of death or indirect cause of death?
JH: I'm going to talk about my favourite thing and it's called sensitivity analysis. It would be nice to know exactly what deaths were due to COVID and which are with COVID, though COVID can also exacerbate other conditions. So, when we're taking this type of data into our modelling, it's important for us to see how sensitive our results are to what the input is and what fraction of these deaths we expect are either due to COVID or with COVID. It’s the same as with the number of cases when PCR testing was more available to everybody, we had to consider when individuals were choosing to go to get PCR tested versus if they had to go get PCR testing. So, we do a big sensitivity analysis over all of these numbers and see how robust our results are. That also helps us get those minimum and maximums and confidence intervals that we can then use to inform decision makers.
Q: What is the global perspective on the need for public health measures, especially given that Omicron is still very active and very serious in some parts of the world?
SH: Taking the global perspective is exactly what we have to do. If we think that by order of council in a cabinet [somewhere] that we can decide when the pandemic ends, we're going to be disappointed when we find new variants emerge, then come back to us through travel and trade in other places.
The reality, then, is that the best approach to getting through this pandemic would be to make sure that we address this pandemic everywhere. That has *not* happened in the world. We have had a very inequitable response. In Canada, we've had the benefit of rapid access to vaccines, diagnostics, and now increasingly, other medical countermeasures like antivirals. Most places in the world haven't had that kind of access. So we could get to a place where we’re temporarily endemic in Canada, only for this virus to continue to fester and evolve elsewhere in the world and come back to Canada causing new waves, because there isn't the immunity to these new variants in the same way that we're starting to see now. The global perspective is not as optimistic as our political leaders have decided around March 21. And that's a problem that affects all of us.
Q: And on that very note, we know that the Ontario government decided to lift most of the public health measures on March 21. There are concerns that removing the masking mandate in schools, for example, is premature and could result in avoidable school closures. Do we have enough data to predict what may happen here?
JH: There's lots of discussion about masking right now. Are we going to keep our masks? Are we going to relax the mask mandates? Or we will relax the mask mandates, and what are people going to choose after March 21st? There are still a lot of things to think about, especially in schools. There’s some research that came out recently to show that children are less transmissible compared to adults for COVID-19. There are also things that say children have a higher level of contacts compared to adults. Maybe those two things will weigh out and make things similar in terms of contacts and transmissibility.
We also can consider other aspects, like people going back to work, more sports and after-school programs, as well as just individuals going out to the movies or a hockey game. There are lots of things happening after March 21st that will enable more transmission of the current variant and maybe even a new variant.
In terms of looking at data, we try to incorporate decision-making – like how individuals consider trade-offs between infection and wearing masks, for example. And we can incorporate this type of decision-making into our mathematical models because we've also seen how people have been behaving over the entire pandemic, and then project what the maximum level of taking off masks might look like and what the minimum level might look like. Then we see if those [results] are acceptable for healthcare. For example, are we going to expect more infections that are going to cause severe infection that will require healthcare and ICU and ward beds? There is data we can use from the past and from current Omicron cases and infections and hospitalizations, so we should be able to get some idea of the minimum and the maximum of what we expect to happen after March 21st.
Q: Every province seems to have its own responses to the pandemic. Should we be looking at a one-size-fits-all policy response or a more nuanced approach that takes small jurisdictions versus large cities into account?
AH: Smaller jurisdictions have, for some aspects of the pandemic, had different best responses from the rest of the country. If you're in a region that sees a lot of importation or there's a lot of high-density housing or public transit use, in some ways that's going to mirror what it's like for a variant to be more transmissible. There are also different healthcare capacities to consider. So, part of the reason why Newfoundland and Labrador took to managing the pandemic with strict border restrictions and an elimination strategy was partly due to geographic and social factors: Newfoundland being an island and Labrador also having a land border with Quebec. But there also isn’t a lot of travel relative to somewhere like an Ottawa, which is between Quebec and Ontario. Seeing a lot of inter-provincial travel by road, that would be really hard to understand.
Newfoundland and Labrador [also] have a lot of rotational workers – people who are working out of province – and that's another local characteristic that has affected some of the best strategies. I don't think there's a one-size-fits-all approach. However, with these variants that are now so transmissible, some of the things that we might have been able to take advantage of in Newfoundland and Labrador, like low transmission for geographical reasons, are less accessible to us now because the way the transmission has evolved with successive variants.
SH: Part of the variation across provinces and the way they responded to this pandemic was based on natural geographic factors. Some of it was just a response to case counts and other metrics. But other parts of the differing responses were based on different social values and different political preferences. And in a democracy, one reason why you design federal systems like in Canada is to enable more localized responses to different issues. Whether a public health crisis like a pandemic is one that should be [considered] at the provincial level, that's a question that was decided in 1867. But my point in saying that is governments at different points in this pandemic knew there had to be layers of protection, but they had choices… and having those choices plays into [living in] a democratic society. And yes, let's get ideology or partisan politics out, but we don't want to get politics out because that's taking democracy out, and that means citizens aren't able to steer the kind of response that they expect to [get from] their government. It also means citizens can't hold their governments to account. We do have that power, and it is democracy to get to do that every time we go to the polls.
I think the starting point is recognizing that scientific evidence represents only one input of many legitimate inputs into decision-making. Some of those other inputs are, for example, what's legal or what's ethical. Others might be related to questions of what we can afford. Others might be related to what we have promised in international obligations to other countries and our or human rights implications.
If you're a political leader or if you're a civil servant working in government, there's a lot of noise. It becomes really, really difficult to figure out what to do, so I do have some sympathy. That being said, as researchers we have options. First, we have to do rigorous work. We have an obligation to do everything we can to try to get those research findings into policy or at least the policy process to be considered as part of response. We need to think in networks. Individual policy decisions are not made by a single person; they're often touched by 100 people before a decision is made. And we need to also know that our research can help hold political leaders and governments accountable for their decisions there afterwards.
Q: Given all that we have learned over the past two years, can you all share your insights on the role mathematical modelling should play in public health going forward? We know the interaction between scientists and the government can sometimes be strained. What would be an ideal way for scientists and the government to interact?
JH: We've experienced many different interactions between government and scientists and industry and scientists over the pandemic. Those that are the most fruitful are ones that incorporate questions from government to scientists, so we can try to tackle something that's going to be meaningful. Also, those that provide data where we can use that data to inform our model so that we make sure that the results that we get from our modelling are applicable.
We can also try to incorporate political decision making in looking at mandates and incorporate that into a sensitivity analysis that we can really project into the future. But one of the earlier questions used the word “predict”, and I answered with the word “project”. We can project. We can't necessarily predict exactly what's going to happen, because the knowledge that individuals get and that governments get from scientists means what the model projects will probably not be correct because it's changed the behaviour ahead of time. In terms of moving into the future, there needs to be some more discussion as to how modelling can be incorporated into decision-making in a way such that it's not believed that the models have to be perfect or right, but they can help inform.
SH: We do not have enough mathematical modellers of infectious disease and other public health matters in our country. I think there's widespread recognition of that now. If anyone in the audience is interested in a career, a very vibrant career, mathematical modelling in public health is going to be an amazing career opportunity. I'd say though, at the same time, we need to think about where mathematical modelling can be helpful and we have to be really careful that the work doesn't get misappropriated or misused by those who are then tasked with making decisions on its basis.
AH: At a basic level, there can be a misconception that the more factors you put in the model, the better the model is. As a result, the public can think that if we make a really simple model, we're just not very good at our job or we're not very aware of what's going on. But if you get models that are suggesting different things, then you want to dig back and know why, and it helps to not have put everything in the kitchen sink in your model to go through that process.
SH: I love having the final word. I'll just say the previous comments really highlight the importance of the choices that get made when models are being developed. Which then highlights the importance of ensuring that the mathematical modelling community is very diverse, with people coming from all backgrounds, all genders, socio-economic statuses, ethnicity, and communities around the world. We make choices and sometimes we all have unconscious biases that can accidentally result in models working for some populations and not others.
The best modellers are making those critical choices very specific to the question being asked and answered. In that respect, making sure future training programs are as inclusive and as encouraging of diversity as possible is going to be critical to make sure that mathematical modelling can continue to have the impact on public health as it did during the COVID-19 pandemic.