Early evidence from Israel suggests that the Pfizer vaccines are working

A group of researchers in Israel have put together a preliminary report tracking cases and hospitalizations since their nationwide vaccination program started in late December. While preliminary, this gives us a first glimpse of the potential population-wide effects of vaccination.

As emphasized in a previous blog post, the rapid vaccination of the entire Israeli population along with its outstanding electronic medical records make their population a treasure trove of information for understand the effect of SARS-CoV-2 vaccination on the population, and this is one of the first thorough reports of such.

Background

Israel launched its SARS-CoV-2 vaccination campaign on12/20/2020, starting at 50k per day, up to 150k per day by 12/24, and >225k per day by 1/21/21. They first prioritized the >60 year old population, with >75% receiving their first shot by 1/7/21, and by 2/2/21, almost 90% have received a first dose and >75% have received both doses.

From available data, assuming that 16% of the Israeli population is >60, I estimate that ~25% of those <60 have received first vaccine dose and ~10% have received both vaccine doses by 2/2/21.

Here I will highlight two primary analyses in their study:

1. Tracking cases and hospitalizations over time in the 0-59 and 60+ age groups to see if there is evidence of reductions after vaccination.

2. Comparison of cities vaccinated "early" (>85% of >60 population vaccinated by 1/10/21) or "late" (<70% of >60 population vaccinated by 1/10/21)

Results and Commentary

Here is their Figure 2 presenting results of cases and hospitalizations over time in the two age groups:

Looking at cases and hospitalizations, we see a drop-off starting early-mid January. Now, given Israel had a surge of cases in December and instituted a 3rd lockdown on 1/8/21, one would expect a natural drop off in January, vaccination or not. So the fact that these decreased is not sufficient evidence to infer any effect of the vaccines.

However, If you split out by younger (0-59) and older (60+) populations, you see very different curves. It is clear that the cases and hospitalizations, while coming down slightly for the younger group that has not received much vaccine, is much less than the older population that has been extensively vaccinated, which has decreased sharply since mid-January. From Table 1 of their paper we see a reduction of 15.7% in cases from 14 days back in the younger population, and a reduction of 40.1% in the older populations. For hospitalizations, the older group decreased 32.3% in 14 days while the 0-59 group remained roughly the same (+0.3%).

One would naturally expect more severe cases in the older 60+ group, and indeed this was the case in early January, but since mid-January notice how the lines cross, as the number of moderate and severe cases in the 60+ population has sharply decreased while for the younger population it has not. From Table 1, we see a reduction of 31.6% in moderate and severe cases from 14 days back in the older group, but actually an increase of 16.5% for the younger group.

This type of differential pattern whereby the case and hospitalization counts decreased more sharply in the 60+ population was not seen in the previous lockdowns, and could be evidence of some vaccine effect.

This hypothesis is further supported by the analysis of "early" vs. "late" vaccination cities, from Figure 3 in their paper.

Note that the differential pattern with the sharper decrease in the older 60+ group is much more strongly evident in the early vaccinated cities than the later ones, with the timing of the decrease shifted later and decreasing slope less steep for the late vaccinated cities.

Conclusions and commentary:

While these results are clearly preliminary, they provide promising signs that SARS-CoV-2 vaccination, in this case with the Pfizer vaccine, might have a substantial effect on COVID-19 cases, hospitalizations, and moderate/severe hospitalizations when rolled out at the population level, especially among the older population.

The timing of vaccination is partially confounded with the 3rd lockdown put in place on 1/8/21, so it is not entirely clear how much of the decrease is due to vaccination, due to the lockdowns, or due to a natural reduction from the December surge.

However, the fact that the vaccination was not rolled out uniformly in the entire population, but was done much more rapidly in the older 60+ population and more rapidly in some cities than others, enables some decoupling of these confounders and assessments of whether the older group and cities with earlier vaccinations indeed experience a sooner or sharper decrease, which is what would be expected if the vaccines were having an effect in reducing infection and sickness.

The differential reductions highlighted in this paper and seen in the plots above provide support the hypothesis that the vaccines are effective in reducing infection and sickness.

These effects might seem relatively small, but it is remarkable to see such a noticeable effect given how short a time has passed since vaccination. Based on the phase 3 trial results, the protective effect of the vaccines takes a couple weeks to kick in, and we know that cases typically do not manifest for 5-10 days after infection, and there is typically another week or more delay before hospitalization. So the reductions of cases and hospitalizations in mid-late January likely reflect a reduction in infections back in the first week or two of January, early in the vaccinations. The next 3-5 weeks of these results should tell us a lot, since if the vaccination is having as strong effect as expected, we should see the case and hospitalization counts decline even more strongly as the higher vaccination rates and second dose effects kick in.

These data also have great value in assessing other questions, including single vs. two dose efficacy and validating vaccine safety, and I hope research groups with full access to these data will rapidly conduct and disseminate analyses like this one to provide information about the effects of population-wide vaccination to the rest of the world. Update 4/19/21: A paper based on an updated version of the analysis in this report was just published in Nature Medicine, and provides strong evidence for societal effect of vaccination based on careful analysis of observational data.