Pandemic Q&A--Winter Prognosis, Vaccine Waning, and Boosters

NewsWire: 11/22/21

Back in 2020 and early 2021, Demography published regular reports and videos on the global pandemic. Over the last year, we've moved on to other topics. But we still get plenty of Covid-related questions.

In this NW, we begin the first of what we hope will become a periodic Q&A report on the pandemic. Here we look at Covid expectations for this winter plus new studies on "vaccine waning" and the CDC's new booster recommendations.

•  What should we expect for Covid-19 this winter?
• Unfortunately, the leading indicators point to another surge this winter. We've gotten used to the master pattern by now. Change in direction of case count comes first, followed 2 weeks later by change in direction of hospitalization rate, followed 2-3 weeks after that by change in direction of deaths.
• Most recently, daily cases reversed direction on Oct 25 at 70.3K. Over roughly the same span, positive tests rates have been rising again. As expected, hospitalizations reversed on Nov 7 at 45.8. Deaths look like they've flattened  (at 1,008) and should be rising again before the end of Nov. (FYI, I use the NYT trackers because they are derived directly from the state and local data; I use John Hopkins for positive test rates.)
• What's driving the rise?
• Colder temperatures plus greater mobility and congregating indoors. Over the next month or two, temperatures are sure to get colder. As we have learned from studies on the original wildtype C19, the shift from hot Aug to cold Jan weather pushes Covid's R₀ down or up by about 10% from its mean. (R₀ or "R-naught," as you recall, is a basic measure of transmissibility--how many persons on average an infected person will subsequently infect.) The Delta variant of course has a much higher R₀ to begin with. Already, Michigan, Minnesota, and other colder-weather states are leading the national case resurgence, while the southern states are lagging behind. Greater mobility and mask-wearing will self-correct once the deaths start rising significantly. But this always happens with a 4-to-6 week lag.
• There are good reasons to hope that the coming surge won't be nearly as bad as last winter (peak: 3.5K deaths per day) or even this fall (peak: 2K deaths per day). Vaccination rates continue to climb, albeit very slowly, and well over 80% of the highest-risk 50+ adult population is now fully vaxed.
• On the other hand, we have not yet experienced a winter with the highly transmissible Delta variant. Delta will become more aggressive than ever among the unvaxed. It has already rendered much of America's "herd immunity" (against wildtype) moot. And, last but not least, the vax protection is losing its efficacy over time.
• So what is happening to vax efficacy?
• Vaccine efficacy refers to a vaccine's effectiveness (VE) relative to no vaccine at all. So if you vax 1,000 people and compare them to 1,000 similar unvaxed people, and then discover that 25 vaxed people got infected versus 100 unvaxed people, then the vaccine's VE against infection is deemed to be 75%. If only one vaxed person got infected, then the VE would be 99%. You can measure VE for infection, hospitalization, or death.
• There are three big things to understand about VE. It varies substantially according to (a) which vaccine you've used; (b) how long it has been since you've been vaxed; and (c) how old or sick you are.
• OK, let's start with the vaccine itself.
• The best two are the mRNA types, Moderna and Pfizer--with Moderna hands down the best. AstraZeneca is probably third. Novavax and J&J (one shot) are trailing at maybe fourth and fifth. China's low-tech (inactivated virus) vaccines like CoronaVac are considerably less effective.
• What kind of numbers are we talking about? Well, within 2 or 3 months of vaccination, both Moderna and Pfizer had VEs in the low-to-mid 90s against infection and mid-to-high 90s against hospitalization and death. So that's really good. Compared to an unvaxed person, basically, it means that you have 1/20th the risk of infection and as little as 1/40th the risk of hospitalization.
• J&J had VEs in the 60s and 70s. And CoronaVac was in the 50s. So that's better than nothing, but not great. After resisting western mRNA tech in favor of their own vaxes, Chinese leaders are at last changing course and signing a joint venture with Pfizer to manufacture the mRNA vaccine for their own citizens. (If you think mandating an effective vaccine is a "draconian" response to Delta, try endless rolling lockdowns!)
• So what happens with time?
• Just about every type of immunity to any disease (after infection or vaccination) wanes over time. This happens because your body gradually "disarms" its immunity in order to prepare for possible new threats. The big first step in the disarmament is the lowering of serum concentration of IgG antibodies. It also happens because the pathogen itself becomes less recognizable due to subtle genetic drift or to shapeshifting into a new variant.
• After creating the new Covid vaxes, all the biologists on the development teams understood that some decline over time was likely. But no one knew the rate of decline. The data we had were fragmentary, small-scale, lacking controls, or not properly categorized by duration from vaccination. Finally, better data is becoming available. And the data are disquieting.
• Originally, early Israeli estimates or surveys elsewhere by Pfizer or Moderna estimated that infection VE declined by somewhere between 6 pp every month or every two months. New and better studies are showing that the decline is closer to 6 pp every month--and that the decline (measured this way) accelerates with each passing month.
• In October, a major USC study of members of Kaiser (the giant CA HMO; N=3.5M) shows that infection VE for the Pfizer vaccine dropped from 87% during the first month to 47% after five months. That's an average decline of 8 pp per month, with the decline accelerating parabolically. According to this study, the apparent good news is that hospitalization VE remained steady at around 90% after 5 months. But, after reading the fine print, we find that the authors define hospitalization VE as VE given infection. OK, fine, but that still implies that, as an ever-larger share of the vaxed are getting infected, an ever-larger share are entering hospitals!
• In November, another CA team conducted a similar study (N= 780K) on members of the Veterans Health Administration (VHA). It found a very similar six-month decline (from 87% to 45%) in Pfizer's infection VE. Unlike the USC study, however, it clearly stated that the rising rate of breakthrough infections "was not benign." The authors do not isolate the death or hospitalization VE. But they do confirm that mortality for vaxed vets who tested positive was definitely higher than for vaxed vets who didn't--which implies (what the USC could not say in so many words) that the higher breakthrough rate over time was accompanied by a commensurate increase in mortality rate.
• What about combining the choice of vaccine with the effect of time?
• The quick answer is this: The infection VE of the better vaccines declines more slowly; the VE of the worse vaccines declines more rapidly. A figure from the VHA study should make this clear.

• As we can see, the infection VEs of all three vaccines tracked by the VHA decline parabolically. For Moderna, the month 0 EV is 89%, declining to 58% by month 6. But for J&J, the VE starts at 86% but ends at a mere 13%.
• Needless to say, 13% is barely working at all. This means that, after six months, 100 unvaxed people get infected and 87 J&J vaxed people get infected. What's more, it looks pretty clear that by 7 months the J&J VE will be near-zero--as will the Pfizer VE by 9 or 10 months and the Moderna VE in about a year.
• But let me emphasize this: Even after the infection VE drops to zero, a vaccine will still confer very important advantages so long as the conditional VE of hospitalization or death given infections remain substantial. How much is "substantial." I don't think the conditional odds improvement is 90% (the figure cited in the USC study), but more likely in the range of 50-75%. Most importantly, unlike infection VE, this hospitalization or death VE does not seem to wane much over time.
• OK, what about the effect of age or health?
• Here, the rule is pretty simple: At older ages and among people with compromised immune systems (obesity, hypertension, kidney disease, immunosuppression therapy, etc. all classify here), the waning of infection VE is faster. Also, there is some evidence that the conditional hospitalization/death VE may not be as large--and (according to Israeli data) may itself wane a bit with time.
• For these groups, however, the absolute benefits of vaccination remain much higher. To illustrate, let's compare two situations. In situation A, you choose between a 0.01% and 0.05% risk of hospitalization (VE of 80%). In situation B, you choose between a 10% and 20% risk of hospitalization (VE of 50%). This example is extreme of course. But it makes the point.
• This analysis has a lot of moving parts. Can you simplify it?
• I hope so. Let's say that, one month after your Pfizer vax, your total VE against hospitalization is 95%. That is to say, you are 20X less likely to go to the hospital due to Covid than an unvaccinated person. Think of this protection as consisting of two factors: first, a 5X lower risk of getting infected and, second, a 4X lower risk of being hospitalized if infected. (These numbers are roughly consistent with a major UK study.)
• What the data seem to be showing is this: Your 5X lower risk of getting infected wanes over time, so that by 9 months or 18 months (depending on the vaccine) it approaches zero. But that still leaves you with a 4X advantage (an VE of 75%) against hospitalization.
• This conforms to our understanding of how the immune system works. Yes, a high concentration of serum antibodies provides a very effective defense against getting infected in the first place. But your immune system can't forever maintain a high antibody concentration against every threat it has ever encountered. Instead, while demobilizing the antibodies, it "memorizes" that particular threat by creating customized T cells and memory B cells which enable your body to mobilize quickly--in case you do get infected again--and defeat Covid much more easily than if you were infected the first time.
• Some vaccines have less anti-infection potency than others, and these tend to lose their potency faster. We don't yet know whether the vaccines differ much in their conditional VE against hospitalization. Also, older/less healthy people tend to lose their anti-infection protection faster (probably due to faster serum antibody decline). There is some evidence--not yet confirmed--that they also may possibly lose their conditional VE against hospitalization.
• Some of you may want to know about the VE against death. All of the studies I have seen conclude that there is no real difference between vaccine hospitalization VE and vaccine death VE. In other words, whatever your risk of death may be if admitted to the hospital with Covid (let's say it's 10%), that risk will be unaffected whether or not you have been vaccinated.
• This sounds like a very good argument for taking booster shots.
• If you want to maintain a high infection EV, yes, this is the way to go. The booster basically rejuvenates your immunity back to month zero. This is especially important for older and less healthy people--both because their absolute risk is higher and their rate of EV decline is faster. A few months ago, health officials thought boosters were only important for these groups--which is why the initial booster recommendation targeted them alone. But now we understand it is important for anyone who wants to maintain an optimal EV or who wants to minimize the risk of infection regardless of the sequelae.
• On Friday, largely on the basis of the new data, the CDC announced it now "recommends" a booster shot for anyone over age 18--not for just the old or sick--who received a Pfizer or Moderna vax at least 6 months ago. Specifically, it suggested that anyone over 50 (plus anyone at any age living in a long-term health care facility) "should" get the booster and that anyone else "may" get the booster. The CDC had already recommended boosters for anyone who got the J&J vaccine more than two months ago.
• Which type of booster should people get?
• The CDC makes no recommendation on which type of booster you may or should get. A recent NIH study, however, makes the choice a pretty simple one: Everyone should, if possible, get the Moderna booster. If that isn't possible, get the Pfizer booster. Basically, no matter what your initial vaccination, your best booster outcome is ranked in order of the VE of the original vax.
• What are the longer-term implications of vaccine waning and the new "booster regime"?
• Without any boosters, the long-term consequence of vaccine waning would have been--and still is, to some extent--a widening gap between the rate of infection transmission and the rate of serious illness among the vaccinated. This translates into a more dangerous world for the unvaccinated.
• To illustrate the point, consider an (idealized) scenario in which two-thirds of the population are all vaccinated on the same day. In the first few months--not considering the impact of new variants--you would see a huge decline in population R₀ since very few of this large vaccinated group would any longer be getting infected and passing on the infection. That would confer large benefits to the nonvaxed by slowing or even stopping transmission in many regions. But as more months pass, infections and transmissions would rise again. Even though much of the serious illness protection would remain for the vaxed, the transmission rate (R₀) would climb back to where it was before--or maybe even higher, since a larger share of "breakthrough" infecteds would be asymptomatic and continue to socialize.
• The "recently vaxed"/low transmission scenario may have described what was happening in Jun and Jul of last summer, when cases, hospitalizations, and deaths sank to their pandemic low-point in the U.S. To be sure, the Delta variant played a major role in pushing up cases and deaths in Aug and Sep. But the lengthening average duration since original vax dates also played a role.
• It was in early Aug that the CDC, for the first time, was blindsided by the cluster of superspreader events in Barnstable County, Massachusetts. The CDC analysts were shocked to learn that, of the 469 cases identified, 74% of them occurred among fully vaxed persons--and, of the 5 who were hospitalized, 4 of them were fully vaxed (though none died). Ever since, we have all become quite used to reports of vaccinated transmissions. In just the past couple of weeks, we have seen significant outbreaks (higher case and hospitalization rates) in Vermont, New Hampshire, and New Mexico. These three states have full vax-to-population ratios of 72%, 64%, and 63%, respectively--all higher than the U.S. average of 59%.
• To date, 69.1% of Americans age 12+ are vaxed. Of these, 17.1% have received boosters. For Americans age 65+, the figures are 86.2% and 39.1%, respectively. (CDC figures.) With each passing month, the vaccine waning gradually increases transmissibility among all the vaxed, while boosters decisively reduce transmissibility among those who get boosters. The net impact on transmissibility (R₀) is ambiguous, since it depends on how many more vaxed Americans continue to get boosters and how often they do so.
• My take, then, is that we need a steady rise in booster uptake--ultimately reaching the same population share as the initial vaccine uptake--to keep R₀ from rising. (Further growth in the initial vax rate of course would help even more.) Are we going to get that? We don't yet know.
• So is the "pandemic of the unvaccinated" fated to continue indefinitely?
• In a word, yes. In a 2/3 vaxed world without boosters, the overall infection and transmission rate would probably return to somewhere near or even above its pre-vax rate, with a disproportionate toll on the unvaxed and a higher death rate for all groups. In a 2/3 vaxed world with boosters, the outcome changes. The overall infection and transmission rate drops significantly, which lowers the death rate for all groups. But those serious illnesses and deaths are now shifted even more disproportionately than before onto the unvaxed. Thus, the booster regime is an even purer "pandemic of the unvaccinated."
• From early summer on, as vax rates grew fastest among the oldest age groups, health officials in almost every state noticed a rapid shift in the mix of hospitalized patients: Increasingly, they were younger, mostly Xers in their 30s, 40s, and 50s. Why? Because, increasingly, these were the unvaxed Americans.
• Are there vaxed Americans who get breakthrough infections and die? Of course. In its recent analysis, the WSJ estimates that there have been to date about 13M breakthrough infections and 20K breakthrough deaths. This amounts to roughly 15% of total reported infections and 12.5% of total reported deaths in the second half of 2021 (when nearly all of them must have occurred). These percentages are strikingly small considering the large share of Americans now vaxed--and the even-larger shares of older and sicker Americans who are now vaxed.
• Without boosters, this share would have risen along with the total death toll for both groups. With boosters, this share will decline along with the total death toll for both groups.
• In its estimated age breakdown of the breakthrough population (by hospitalization, ICU admittance, and death), the WSJ implies that the breakthrough toll is especially severe for the elderly. 

• Well, of course it is. But the researchers do not establish that this breakdown is any more skewed against the elderly than an age breakdown of unvaxed infecteds. When I compare these multiples (for example: under 18 to 65+), they look no different to me than the prevalent multiples observed before the vaccines were widely available. In other words, I see no change in the multiples--and therefore no clear evidence, yet, that the conditional VE of hospitalization or death given infection has changed substantially for the old versus the young. IMO, the recent growth in the total death rate among vaxed Americans is driven primarily by their growing infection rate, not by any large age-related decline in the ability of the vaxed to survive an infection.
• We will continue to track the data to see whether such an age-related decline does appear over time among the vaxed who don't get boosters. Until then, we should assume that the main advantage of getting a booster shot is to drastically reduce the probability of getting infected in the first place.

Stay tuned for future Pandemic Q&As. Among upcoming questions in our queue: Can breakthrough infections cause Long Covid? Is natural immunity superior to vaccine immunity? What does "endemic Covid" look like in the long run? And more.

To view and search all NewsWires, reports, videos, and podcasts, visit Demography World. For help making full use of our archives, see this short tutorial.