Should We Be Counting COVID Cases?

How to think about rising numbers

For those of you who follow this for discussions of schooling, our team has a new NBER working paper out this morning on the impact of pandemic schooling mode on test scores. You can see it here.


It’s a confusing time in the pandemic. On one hand: vaccine access is high, boosters are widely available, kids 5-11 can be vaccinated, new therapeutics are coming soon. On the other hand: cases are up, especially in colder areas where things are moving inside. It seems likely that we’ll see further case increases around the holidays. 

This has raised the question: How much should we care about cases at this point? Does it matter how many cases there are, when vaccinations are high? Should we even be counting cases at all?

I see this question from two angles: individual and public health. I’m going to focus largely on the second today, but I’ll start with the first, which I think is less complex.

Should you change your behavior based on number of cases? 

If you’re fully vaccinated, broadly speaking, I’d say no. As individuals, the things in our power are to (a) protect ourselves and (b) take steps to limit the impacts of our behavior on others. You can protect yourself by getting vaccinated, and by getting a booster if you want another layer (which you probably should). If you have kids 5-11, you can protect them by getting them vaccinated. If you have kids 0-5, you can protect them by getting yourself vaccinated. Wearing a high-quality mask when you’re traveling or in crowded indoor settings can add a layer of protection.

Vaccines also protect others, and you can take another step in that direction by adding a couple of layers to your safety procedures when you gather — don’t be around other people when you’re sick, and use rapid tests. NB: A much longer form of this same argument is here, and a discussion of holidays in particular is here.

These are precautions that I believe many of us will continue to do for the foreseeable future, more or less regardless of case rates. They recognize the ongoing presence of COVID-19, and actions that we can take to keep ourselves safe. But once you do these, you’re very well-protected from serious illness yourself, and you’re doing a solid part to limit your impact on others. You may or may not choose to go to the gym (I would), but that decision probably shouldn’t be based on the particular case rates this morning versus yesterday. 

Should public-health officials be counting cases?

The above is an argument that for fully vaccinated individuals, case rates shouldn’t dictate behavior. A possible corollary is that we shouldn’t track cases at all and should focus instead on hospitalizations and deaths. This discussion is often framed as “Should we count cases?”, although I think that is a bit misleading. Counting cases is just about getting data; more data is generally a good thing, since you can always ignore it. 

In practice, what I think is meant is closer to “Should we base policy on cases?” or, perhaps even better, “Should lowering case rates be the goal?” 

This goal-setting question is where I want to focus.  

When it comes to this next phase of the pandemic, there are at least two possible goals. One policy goal is lower case rates. A second is lower rates of serious illness, hospitalizations, and deaths. These are closely linked, but they are not the same. When case rates go up, serious outcomes are expected to increase. But how much they increase depends on the conversion rate. In a sense, we can think of focusing policies on two levers: (1) reducing cases and (2) reducing the rate at which cases become serious illnesses or deaths.

This distinction matters because the policy levers we have to address them differ. Our approaches are really in three categories: policies that do not matter at all and we shouldn’t do; policies that impact case rates; and policies that impact conversion rate from case to hospitalization or death. (I’m going to refer to these all as policies even though some of them, like vaccination, are more individual behaviors; I’m thinking of the policy as “encouraging this behavior.”)

Policies that do not matter: There are a number of policies we have tried that appear to have little or no impact on COVID case rates. These include school closures, extra cleaning procedures, and restrictions on outdoor activities (including any outdoor masking other than in very crowded settings). We shouldn’t be doing these!  

Policies that may affect case rates: There are many approaches to lowering case rates, but I’d list five: (1) vaccinations, (2) various lockdowns or restrictions on gatherings, (3) mask mandates, (4) ventilation improvements, and (5) better testing access and encouraging testing. 

Policies that affect conversion of case rates to serious outcomes: Just two: vaccinations and better therapeutics (like the new pills from Pfizer and Merck).  

If our primary goal is to decrease case rates, we need to focus on the policies that accomplish that. If our primary goal is to limit serious outcomes, we have the full set of policies to choose from.  

A reasonable question you may be asking: Even if the goal is only to limit serious outcomes, why not just do all the policies? If we lower case rates and lower the conversion rate, it’s a double whammy!

The issue with this is that our resources — financial, but also attention and focus resources — are finite. Every policy cannot be the most important policy. When we try to get people to do everything, we may fail to help them see which things they should prioritize. 

Put simply: if we tell people to stay home when possible, wear a mask all the time, stay distant from others, get vaccinated, and test if they gather, they may find it impractical to do all of these things. Of this list, most public-health officials would agree that getting vaccinated is the most important. But if we fail to prioritize, it’s a lot to expect everyone to see that. At the individual level, we also run into confusing interactions between policies. Mask mandates that include vaccinated people may make sense, but there is no question that they have affected trust in vaccines.

The same issues arise in the focus of government efforts. Do we prioritize vaccine distribution or testing? Boosters for which age group? Choices have to be made. 

If we recognize that we cannot do everything, the question of where we want to focus our efforts comes down to identifying our goals and, within those, thinking about the most effective approaches. Looking at the list above, it is difficult to miss that by far our most effective policies are those that affect the conversion from cases to serious illness. Fully vaccinated people are still, even without a booster, in the range of 90% less likely to be hospitalized or die. The new Pfizer therapeutics are showing efficacies of 89%. Those are both huge numbers. 

In contrast, many of the policies that target cases have much lower efficacy. Lockdowns likely mattered some (see e.g. this paper on cross-country lockdowns), but the evidence on exactly which restrictions are important is not clear. And within the U.S., there isn’t a clear cross-sectional relationship between COVID levels overall and lockdown choices. 

Masks impact spread, but the effects are moderate. The most compelling evidence comes from Bangladesh, where a randomized trial estimated perhaps a 30% reduction in symptomatic illness from surgical masks with perfect compliance (cloth mask effectiveness was not significant). With the actual compliance in the trial, the reduction was only about 10%. 

Appropriately used, testing has the potential to be more effective — frequent rapid testing could remove perhaps 90% of infectious cases from circulation — although it requires ongoing buy-in from individuals. Ventilation improvements are also likely to matter a lot, although they require investments. Notably, vaccinations — especially including a booster, but even without — are among the most effective ways to lower transmission. Even with the possibility of breakthrough infections, vaccinated individuals are less likely to become infected, meaning they will be less likely to spread the virus. 

Pulling this together

  • If our goal is to reduce serious outcomes, we should prioritize three things: vaccinations (lowers conversion from case to serious outcomes; lowers case rates); approving therapeutics and making them widely available (lowers conversion); and making rapid testing widely available and easy, and providing guidance about use (lowers case rates). 

    Together, these have the potential to somewhat impact case rates and to hugely impact conversion rates, and as a result dramatically limit serious outcomes. Prioritizing these things doesn’t mean ignoring the others. It doesn’t mean saying masks do not matter at all. But it’s about where we put our resources and what we emphasize in public messaging.

  • If our goal is to seriously target lower case rates, we need to move on effectively all of our case-reduction policy levers. We will need more continued and enforced mask mandates, more vaccines (and more boosters), probably more lockdowns or movement restrictions, and also better testing access. Even with more restrictions, we may be unable to shift case rates very much, although it seems likely it could have some impact.

These are statements about what is possible. Where does it leave us on how we might think about crafting policy?

Most importantly, it should make clear that the first step in making policy is to say what our goal is. Our current policy discussion is muddled because we haven’t said what we are aiming for. For public-health officials, municipalities, states, and even the federal government, the first step is to state what you are trying to do.

If the goal is to lower serious illness and death, we should pivot our focus away from marginally effective policies that target case rates and focus all in on the smaller set of highly effective policies that can achieve the goal. This doesn’t mean we shouldn’t count case rates (data is good!), but it would mean that we shouldn’t make policy based on these rates. The focus should be on hospitalization, on tracking breakthrough hospitalizations in particular, on targeting therapeutics to areas with less vaccination, etc. The booster shot focus in this case should be on older adults or those who are immune compromised.

On the other hand, if the goal is to lower case rates — to go toward COVID-zero, even if we recognize that arriving there is hard — we should acknowledge that this means leaning very strongly into the set of policies that may impact case rates and (by extension) leaning away from some of the policies that impact conversion. For example: if our goal is lowering case rates, expanded rapid testing may be as or more important than vaccines. 

In the end, goal clarity can lead to policy clarity. I have a view here, in case it is not blindingly obvious, that we should focus on serious illness and death and lean into vaccines, therapeutics, and testing. But I know that for others, the goal of lowering case rates is more central — on the view, perhaps, that we do not know the long-term consequences of COVID so we need to be as careful as possible to avoid infection. It is OK to differ in our goals and to debate the right ones. This debate cannot happen, though, if we do not say what the goals are.  

I asked at the top: Should we count cases?  

And I answer here: It depends on whether that’s the information we want to know.


Did You Miss It?