The Delta Variant
Following last week's newsletter, several readers asked me to discuss the Delta variant. I'm glad to. A lot of stats are being tossed around, and a preprint of a new study, released this week, seems at first glance to constitute bad news even for the vaccinated.
If you haven't gotten your vaccine, please stop reading this and go get one. Since June, unvaccinated people represent more than 97% of those hospitalized for COVID-19 and more than 99% of those who die from it (see here, for example).
The new study I mentioned addresses a question on everyone's minds: How dangerous is the Delta variant for vaccinated people? Let's call it the MC study, because all data came from the Mayo Clinic or hospitals in its health care system.
The purpose of the study was to compare COVID-related outcomes among Americans who had received the Pfizer vaccine, those who had received Moderna, and those who were unvaccinated. The sample was large (over 75,000 adults) and rigorously selected. The members of each group were matched on variables such as sex, race/ethnicity, and date of vaccination. (See my Appendix if you're unfamiliar with the term "matched.")
Data were collected for the MC study between January and July 2021 as the Delta variant spread and became responsible for over three-quarters of COVID-19 cases. The researchers found that vaccine effectiveness at preventing infection was high (86% for Moderna, 76% for Pfizer), and even higher at preventing COVID-related hospitalizations (95% for Moderna, 85% for Pfizer).
(As I noted last week, these percentages don't describe individual risk of infection, but rather how much protection the vaccine provides. 76% effectiveness, for example, indicates 76% fewer infections among the vaccinated group, as compared to the unvaccinated group, regardless of the actual numbers of infections.)
Now, you might be wondering: Why am I focusing on effectiveness stats, which are relative values? Why don't I tell you what percentage of people in each of the vaccine groups got infected? Isn't that the most relevant information for understanding how dangerous Delta is to vaccinated people?
Well, no, not by itself, although it's useful information. Overall, the researchers found that less than 1% of each of the vaccinated groups developed COVID-19 during the study. That's good news. But keep in mind that it represents one seven-month time period. The rate of breakthrough infections (i.e., infections among the vaccinated group) rose during that time period, and is presumably still rising, even though the overall incidence was still less than 1% by the end of the study. In other words, a small percentage of vaccinated participants who hadn't been infected by the end of the study will eventually become infected.
In short, breakthrough rates for short time periods may be underestimates. One hopes that the effectiveness stats would be somewhat more reliable, even though we know these will change too as new variants emerge. In fact, effectiveness stats are where the MC study seems to provide bad news. (Spoiler alert: It's not as bad as it sounds, in my opinion.)
As you saw a few paragraphs back, the overall effectiveness stats were lower in the MC study than in studies conducted prior to the emergence of the Delta variant, when effectiveness for both vaccines was around 90%. Even worse, just looking at July data, the MC researchers found noticeably lower effectiveness against infections – 76% for Moderna, 42% for Pfizer. That sounds worrisome, especially if you've had the Pfizer vaccine.
Here's the good news:
1. Effectiveness stats are, in essence, generalizations about thousands of people. Various people, who live in various places, and engage in various behaviors (not all of which are safe). You're one person, and to a great extent you choose your own behaviors. Your actual risk of COVID-19 depends most on those choices. Getting vaccinated was a great choice, for instance. Even if you happened to have the Pfizer vaccine, I wouldn't worry, so long as you're minimizing personal risk (more on that shortly). Breakthrough rates for those who received the Pfizer vaccine are surely highest among those with the least safe behavior.
Here's an analogy: You may know what percentage of people in your city were victims of violent crime over the past, say, 10 years, but that's not a very accurate indicator of your personal risk, in part because crime rates vary widely by neighborhood. If you live in a safe neighborhood, your actual risk is lower than the city-wide figure suggests. And, there's more...
Suppose you live in a relatively safe neighborhood. Call it neighborhood P. The violent crime stats for neighborhood P will be much more favorable than for the most dangerous part of town, neighborhood U. However, a lot of other people besides you live in neighborhood P. Those with the least safe behavior (e.g., late-night visits to neighborhood U to buy drugs) are the ones most likely to be a victim of a crime.
My point is that if you got the Pfizer vaccine, you're already in a relatively safe neighborhood, even if it's not the safest one in the city. And, you may actually be safer than other recipients of Pfizer if you do what you can to minimize risk.
2. The July data in the MC study showed that effectiveness against hospitalizations was high (81% for Moderna, 75% for Pfizer). This is consistent with other studies showing that the vaccines are particularly effective against preventing the most serious cases of COVID-19. Here again, individual behavior influences risk. The least safe behaviors (e.g., close contact with unmasked people in poorly ventilated spaces) will expose a person to the greatest viral loads and thus the possibility of more severe, hospital-necessitating infections.
3. The worst news from the MC study represents just one month of data. And, there are other studies. For example, a study published last month in the renowned New England Journal of Medicine (NEJM) compared the effectiveness of the Pfizer and AstraZeneca vaccines among nearly 20,000 people in England. The approach to sampling and other aspects of the NEJM study's methodology were rigorous. Among fully vaccinated participants, effectiveness against Delta infection was 88% for Pfizer and 67% for AstraZeneca. For those who've had the Pfizer vaccine, this is better news than what the MC study offers. (You might even call it great news, since earlier studies showed that Pfizer had almost the same level of effectiveness for the Alpha strain – roughly 90%.)
Why the discrepancy between the two studies for the Pfizer vaccine? 42% is strikingly less than 88%. Which stat do you trust? (News media are already beginning to toss around that 42% figure, so this won't be the last time you hear about it.)
Usually, researchers propose reasons for discrepancies between their results and older ones. The MC authors failed to do this, noting instead that their findings are consistent with the NEJM study in showing lower vaccine effectiveness for the Delta variant. Perhaps the MC authors were being cautious, but I believe they had an obligation to address – or at least acknowledge – the 42% vs. 88% discrepancy.
Discrepant findings are sometimes traceable to methodological differences between studies, but I'm not sure how that applies here. I would only note that the NEJM study seems to be stronger, methodologically speaking. For example, every participant in the study with COVID-19 was identified as having either the Alpha or the Delta variant. (That 88% figure I mentioned was strictly for Delta infections. Pfizer's effectiveness against the older, Alpha strain was 93.7%.) In the MC study, Alpha and Delta were not distinguished at the level of individual participant. Instead, conclusions about Delta were grounded in data on estimated state-level rates of this strain. This of course weakens what can be concluded specifically about the Delta variant.
Another concern I have about the MC study is that no rationale is given for the mathematics behind case-rate calculations. The technical details are explained, but not justified. For these and other reasons, I believe the NEJM study represents better methodology. I may be missing something, of course, since I lack biomedical expertise. And, as the authors of both studies noted, we don’t know to what extent their findings were influenced by variables that weren’t measured or controlled for. All the same, I’m slightly biased in favor of the NEJM findings, at least until more data are available.
So, what's the bottom line? How dangerous is Delta for the vaccinated?
The best answer to that question will not only emerge over time, but also evolve as new variants or subvariants emerge, and as boosters are rolled out. In the meantime, I think that some degree of positivity is warranted.
1. Among those vaccinated in the U.S., the incidence of breakthrough infections is, for the moment, still less than 1%, regardless of which vaccine you've received. Although this figure is likely to increase over time, it does represent a time period extending back as much as two months earlier than March, when the Delta variant was first identified in the U.S.
Since I keep saying "less than 1%", you may be wondering: How much less? Quite a lot, actually. State-level data published on July 30 show breakthrough rates ranging from .01% (Connecticut) to .029% (Alaska). Let's go with the worst-case scenario. That is, let's assume you live in Alaska, and let's round that .029% up to .03%. That's 3 in 10,000. In other words, as of July 30, about 3 in 10,000 vaccinated residents in Alaska had gotten COVID-19. The actual number is probably higher, owing to undetected mild cases, and it will most probably increase over time. For the moment, from the perspective of individual risk, it might still be considered a small number.
(To put that number in context, about 3 in 10,000 Americans dies every two years from an overdose of opioids – and, as you know, your actual risk of dying from opiods depends on your behavior. The chances are zero if you don't use them, but greater than 3 in 10,000 if you do.)
As I keep stressing, vaccine effectiveness is the key stat, but for the moment, at least, your personal risk seems low if you're vaccinated and engaging in safe behavior. Which brings me to the other reason for some degree of positivity.
2. Although the advice from leading experts and organizations keeps changing, the basic prevention strategies haven't changed much: Get vaccinated, socially distance to the extent possible, avoid unventilated public spaces and/or dense crowds, wear the right kinds of masks, etc. In short, we know what to do.
I still haven't directly answered the question that prompted this newsletter ("How dangerous is Delta to the vaccinated?") From a statistical perspective I would conclude that Delta doesn't seem very dangerous yet if you're avoiding risky behavior. However, it's impossible to put an exact number to that statement, in part because it's impossible to know how much you reduce a currently small risk by the specific behavioral choices you make in social settings.
Numbers aside, nobody, including experts, can tell you how you should feel about the stats we throw at these kinds of questions. Some people find stats like 3 in 10,000 scary, some don't. Now that you've seen some data and you know how to minimize risk, I hope you choose caution over fear.
Thanks for reading!
Appendix: Matching
"Matched" means that the groups were identical in composition to each other with respect to the matching variables. If there were, for example, 94 Hispanic women from Wisconsin ranging in age from 18 to 25 in the Pfizer group, then each of the other two groups contained 94 women with those characteristics.)