Post #1189: Part 2B of the new COVID odds, the recent findings from Israel.

Posted on July 27, 2021


Today is the start of my “all things vaccine” series of posts.  I want some reasonably solid answers to  a handful of straightforward questions pertaining to U.S. individuals who are fully vaccinated against COVID-19.

For now, I’ll settle for answers to one thing:  What’s up with the data being reported from Israel?  In a nutshell, based on a small sample of cases, the Israeli Ministry of Health suggests that the Pfizer vaccine’s ability to protect against infection by the Delta variant shows a nearly linear decline over time, and is almost nil six months after vaccination.  Let me emphasize that finding is based on a very small sample, and that protection against hospitalization and death remain high, it’s only protection against “any infection” or “symptomatic infection” that appears to fall off rapidly in their data.

In this post, my goal is to channel the spirit of Light and Pillemer.  This all about research results that disagree.  It is not sufficient merely to say that results disagree, and shrug your shoulders.  The point of a review of the literature is to identify why they disagree, and figure out a way to test that.

I’ll give you my conclusion up front:  I’m not sure yet whether that Israeli finding is real, or is an artifact of methods.  But it sure has the “flavor” of a real finding.  There’s some modest evidence that maybe some of their result is due to having vaccinated the highest-risk individuals first.  (So that their earliest cohorts differ systematically from their later cohorts.)  And, oddly, the fading of immunity is far more pronounced in their younger population, which is exactly the opposite of what would be expected based on the weakening of the immune system with age.  But in the main, from what has been published, this has the appearance of being a true finding, including a rough dose-response relationship (immunity declines monotonically with time from vaccination).

So why hasn’t the U.S. found that?  Why haven’t the British found that?  Well, that’s plausibly because Israel was ahead of both of us in getting their population vaccinated.  The most profound loss of immunity in the Israeli data was for those fully vaccinated in January 2021.  Those individuals account for almost 20% of the Israeli population, but maybe a percent or two of the U.S. or British populations.  That was followed by those vaccinated in February 2021, and so on.  Because we were so late to vaccinate, compared to Israel, those cohorts aren’t a very large fraction of the U.S. or British populations, compared to their impact in Israel.

I’ll say something further:  Given how important this is, how large and rapid a deterioration of protection that Israel appears to have identified, and how huge a sample size of Delta-infected individuals we now have in the U.S.A., it should be a piece of cake for any large state with an outbreak to confirm or reject the tentative findings of the Israeli Ministry of Health.  Just match up the rosters of the vaccinated and those who tested positive, and cross-tabulate by month of vaccination.  If the results are as the Israelis suggest, the results will jump off the page.

Bottom line:  If I can figure that out, so can every state public health agency in the country.  It should not be long before the current uncertainty over this is settled.

If I had to summarize, I’d say that everyone agrees that the vaccines provide good protection, out to at least six months, against hospitalization (and probably against death) from Delta.  But there’s some evidence (from Israel, but not yet from elsewhere) to show that protection against any infection fades rapidly over that time period.  Separately, it’s now clear that those who have a “breakthrough” infection (i.e., infected despite being vaccinated) are perfectly capable of spreading the disease, just like anybody else who is infected.  (No surprise there.)  Though likely at a lower rate than the un-vaccinated.  But in the context of Delta, where viral loads appear very high, there may still be significant disease spread via breakthrough infections.

One part of this that I have not yet examined is how unusual those Israeli findings are.  Whenever something has been presented as some new aspect of COVID-19, what I have generally found is that other viruses do the same thing, it’s just that nobody ever much noticed.  (E.g., about 16 percent of flu infections are totally asymptomatic.)   I’m familiar with vaccines that require boosters at five-year intervals.  I’ve never heard of one that more-or-less wears off within six months.  Is there precedent for this or not, that’s a question I still need to research.

A final complication is the for-profit drug industry.  Based on what I read, the vaccine manufacturers are pushing hard for booster shots.  I can’t say the approval process is corrupt, but neither does it appear perfectly pure.  So, science aside, the force of money is behind concluding that booster shots are needed, and money can be a corrosive and corrupting force.

Here’s my take on it:  Exceptional claims require exceptional proof.   And now that the Israelis have found this, that finding should be replicated in short order.  Either Israel is the bellwether on this, because Israel vaccinated first.  Or those new Israeli results are an artifact of methods or (far less plausibly) simply small numbers of cases.  But it really should not take long for others to duplicate the basic result of an apparent and rapid decline in immunity against the Delta variant within during the first six months of immunization.

Does this mean you should get a booster shot?

Does this mean you should get a booster shot?  I’m not sure, and it’s not clear that you can get one in the U.S. at this point anyway.  That sort of thing requires FDA clearance.  Worse, it appears that the vaccine manufacturers are pushing for it, which means that there will be pressure to approve regardless of scientific merit.

There’s a cogent discussion of both sides at the University of Minnesota Center for Infectious Disease research and policy website.  It appears clear from that discussion that more-or-less the only evidence suggesting the need for a booster, at this point, is this recently released study of a small number of recent cases from Israel.  Hence, the Israeli results appear to play a pivotal role in the evolution of the booster shot issue in the near future.  Quoting CIDRAP:

Our official U.S. position on booster shots can be found at this URL.

Americans who have been fully vaccinated do not need a booster shot at this time. FDA, CDC, and NIH are engaged in a science-based, rigorous process to consider whether or when a booster might be necessary. This process takes into account laboratory data, clinical trial data, and cohort data – which can include data from specific pharmaceutical companies, but does not rely on those data exclusively. We continue to review any new data as it becomes available and will keep the public informed. We are prepared for booster doses if and when the science demonstrates that they are needed.

It gets back to that core question:  Is the effect the Israelis are seeing real, and neither we nor Great Britain is seeing it because we started vaccinating later?  Or is the Israeli effect some sort of artifact of methods, or a mere statistical fluke — plausibly because they vaccinated the most at-risk first?  That’s what FDA and CDC need to decide.  And I don’t think it ought to take them long to figure it out.

A quick summary of Israeli and British evidence so far.

If I pick up British studies of vaccine effectiveness, they seem to match what I expect to see.  And they roughly match the clinical trial results, plus or minus the usual differences between clinical trials and observational data.  And they match what I would call the consensus view in America, as well as matching what I would call the normal behavior of vaccines.

All of that research — both clinical trials and observational data — says the Pfizer and Moderna vaccines work pretty well, even against this new Delta variant.  And that immunity lasts for quite a long time.

In a nutshell, clinical trial data and observational data out of Great Britain suggest that the Pfizer vaccine is about 80% to 90% effective in preventing  symptomatic illness from the Delta variant.  And more effective than that at preventing severe illness, hospitalization, and death.

Here’s a table from their most recent vaccine surveillance report, available at this URL.  The results are a bit worse than the clinical trials estimate, but these are observational data — the people who did and didn’t get vaccinated might have been systematically different from one another.

Source:  British Ministry of Health, COVID vaccine surveillance report — Week 29, available at URL cited above.  Numbers are percents.

You have to go to their background methods document (at this URL) to get more information on methods, but even with that, there is no explicit mention of adjustments for age or comorbidities of the vaccinated population.  There’s just a passing mention that these effects did not vary much by age.

By contrast, there’s the information just released by the Israeli Ministry of Health.  Let me emphasize that this is  very new and based on few cases.  It boils down to “immunity to Delta mostly fades after six months” and as a corollary, overall immunity conferred against Delta is rather modest, once the vaccinated population ages into that six-months-plus state.  You can find the source document at this link.

As a result, the most recent press release from the Israeli Ministry of Health says that, at present, the Pfizer vaccine is only 41% effective in preventing symptomatic illness from the Delta variant. (Although, to be fair, that’s based on a tiny number of cases, and the 95% confidence interval is 9% – 61%).  But the vaccine remains quite effective at preventing severe illness and hospitalization.

Source:  Israeli Ministry of Health, URL cited above.

And, they get that result because they estimate that the protective effect of the vaccine fades rapidly over time.  Here’s the key graph.  The blue bar shows effectiveness of vaccine for those vaccinated on January 2021.

Source:  Israeli Ministry of Health, URL cited above.

So, what’s the true story?  Is the Israeli experience really different, or is this just an artifact of methods, or possibly even one of mis-reporting of the facts?

There are a few obvious differences between the British and Israeli experience.  The British used a mix of three different vaccines, while the Israelis used Pfizer only.  The British (I think) went in strict age order, while the Israelis prioritized the elderly, the high risk, and health care workers first.

But in this context, the main difference is that the Israelis were vaccinated earlier.  Their contingent of aged, fully-vaccinated individuals is a much higher proportion of the population.

At the end of January 2021, about 20% of the Israeli population was fully vaccinated.  At that same point, thanks in part to a British decision to prioritize the first shot over full vaccination, less than 1 percent of the British population was fully vaccinated.

And what about the U.S.A.?   Our experience is more like the British. We have relatively few vaccinated individuals who fall into those January and February cohorts.

That said, we have a huge population.  It seems like a simple cross-reference of vaccination rosters and test rosters could easily identify this cohort-age effect in the U.S.A.  Just take any large state with a lot of the Delta variant — California, say — match up the two lists, and check the breakthrough rate by month of vaccination.

Given our huge sample size compared to Israel, and the severity of the current outbreak, if the vaccine protection deteriorates to the degree shown in the Israeli findings, any large U.S. state ought to be able to identify that in a heartbeat. For sure, if I had access to the underlying data files, I’d have done it by now.

A few other thoughts.

These Israeli findings are from the same group that documented excellent success of the Pfizer vaccine against the Alpha variant (published in The Lancet.)  That suggests the likelihood of mere methodological error is small.  It also suggests that this change in due to the Delta variant (that is, they appear to have found no such effect for the Alpha variant).

In my limited experience, I’ve found that generational effects like this can arise in observational data for spurious reasons.  E.g., perhaps the people who opted in earliest are the ones most at risk.  Hence, during the next outbreak, they’ll have the worst results.

In this case, that’s clearly true.  Israel decided to vaccinate the elderly and those at high risk first.  (Along with health care workers).  As did most other nations, including the U.S.

In other words, there may be other risk factors that are correlated with the various monthly cohorts.  For example, if Delta was a particular problem for health care workers, the earliest cohorts if the Israeli vaccinated would show the highest rate of breakthrough infections.  That wouldn’t necessarily be due to fading of immunity. That could be due to the order in which occupations were given priority in vaccination.

I can only guess that the text of the document mentions this possibility.  That seems to be the gist of Fortune’s reporting of these results.  As I can’t read the text, I am not able to verify that.

In that regard, I note one oddity of the results:  This presumed “fading of immunity” effect for Delta is much more pronounced for those under age 60.  I would have thought, given that immune systems weaken with age, that it should have been the other way around.  If the vaccine-induced immunity is going to give out, you’d think it would give out first and most in the oldest.  But that was not true.

Source:  Israeli Ministry of Health, URL cited above.

From my experience, I can tell you something that I would like to see, if possible, from the Israeli Ministry of Health.  A lot of researchers find something where they expect to see it, and stop looking.  Few of them go on to make sure that they don’t find it where they don’t expect to see it.

Israel should repeat the analysis using the time period when the Alpha variant was dominant, if that is possible.  Pool the observations using a “proportional hazards” model if sample size is small.

If this apparent fading of immunity is really just a generational effect (an artifact of vaccinating the most at-risk first), that should have appeared for the Alpha variant as well.  But if it does not, that strongly increases the inference that this has something to do with the Delta variant, and is not an artifact of the ordering of vaccinations over time from most to least at-risk.

Finally, this isn’t my last word on this.  This is just getting the facts down on paper, and trying to figure out what causes the difference between the Israeli an and British findings.  I need to let all this percolate a bit before addressing it again.