Pfizer in Practice

This week brought a medical article worth discussing: the New England Journal of Medicine published the results of a study of the efficacy of the Pfizer SARS-CoV-2 vaccine in the real world. The article provides compelling evidence that the vaccine works extremely well.

The data come from Israel, which has been doing a superior job of vaccinating its citizens. As of a week ago, two-thirds of the currently eligible population in Israel had gotten both of the recommended doses (individuals under age 16 and those who have had Covid-19 are not eligible). In Israel, health insurance is mandatory for all permanent residents; they must join one of four healthcare organizations called “funds.” The new study reports data from Israel’s largest health organization (Clalit Health Services) and includes information on a stunning 1.2 million people.

The study’s authors, led by Dr. Noa Dagan, used Clalit's integrated electronic medical record to capture health data for 596,618 people who received both doses of the Pfizer vaccine between December 20, 2020 and February 1, 2021. They then matched them, based on demographic and clinical characteristics, with another group of identical size who had not received any vaccine. Next, they looked at five different outcomes: documented Covid-19 infection; symptomatic Covid-19 infection, Covid-related hospitalization, Covid-related severe illness, and Covid-related deaths. Because the sample was so large, they were also able to collect extensive information about a number of interesting subgroups defined by age or specific co-existing health conditions such as cancer or diabetes.

The article includes an enormous amount of intriguing data. The most exciting results, from my perspective, address outcomes a week or more after receiving the second dose of the vaccine. At that point, the vaccine conferred 94% protection against symptomatic Covid-19 (95 percent confidence interval 87-98), 87 percent protection against Covid-related hospitalization (CI 55-100), and 92 percent protection against severe Covid (CI 70-100).  The efficacy in people over 70 was identical to those in younger individuals, and the rate in people with one chronic health condition such as diabetes was only slightly lower than in people without the condition. 

These numbers strongly resemble the results that Pfizer and BioNTech reported to the FDA in their application for emergency approval.  But, as the study authors point out, Pfizer drew its conclusions based on 44,000 people; the Israeli study is based on 1.2 million people. As a result, when Pfizer calculated the efficacy against severe Covid-19, they drew on a total of 10 cases (one of whom had been vaccinated and 9 of whom had not been); when the Israelis calculated the risk of severe Covid, their estimate was based on 229 cases, vastly increasing the credibility and certainty of the calculation. Moreover, Pfizer’s data was based on the somewhat artificial conditions of a clinical trial: for example, the subjects were all highly motivated to optimize their health and may have regularly worn masks and practiced social distancing; the Israeli study drew on real life experience, in which participants’ behavior reflected community norms.

The new study, like all studies, has its limitations. It excluded people living in nursing homes and medical personnel working on Covid units in the country’s hospitals, arguing that the rate of the disease in their particular communities, i.e. the nursing home or the Covid ward, was highly atypical. The study was performed during a period when the South African variant was very rare in Israel so we cannot draw conclusions about the efficacy of the vaccine against this strain. The information on the ability of the vaccine to prevent Covid-related deaths is limited because of the short follow-up period: there were nine Covid deaths in the fully vaccinated and 32 deaths in the unvaccinated group, but these numbers might change when more time elapses. The data on deaths may also be difficult to generalize as Israel has an unusually low case fatality ratio: in Israel, according to Johns Hopkins' "Our World in Data," is currently 0.7 percent whereas in the U.S. it is 1.8 percent.

Some of the study’s greatest strengths are also potential weaknesses: the “real world” nature of the investigation means it is an observational study rather than a randomized controlled trial, raising the possibility that the differences in outcomes between the vaccinated and the unvaccinated were related to some factor other than their vaccination status. Despite these limitations, the study provides very encouraging information.

The fact that the Israelis could carry out their study sends another message over and beyond the efficacy of the Pfizer BioNTech vaccine. The study could only be conducted because Israel did a good job acquiring and distributing vaccine. Early on, the country developed mass vaccination sites. Since everyone is enrolled in a health plan and the plans all have electronic records, there was no need to waste as much time on bureaucracy as we do in the U.S, where more time is spent filling out forms than on administering the shot. The study could only be conducted because of Israel’s electronic health records, which assured that information about who got what dose when, and the age, sex, and chronic medical conditions of each individual was digitally recorded. Finally, the entire rollout was centrally coordinated, assuring efficiency and consistency: from the outset of the pandemic, the Israeli Ministry of Health collected Covid-related data from all four health plans, negotiated to purchase vaccine from Pfizer, and organized distribution. The good news reported in the NEJM article is a result both of the biological properties of an mRNA vaccine that was designed in record time to deal with an international health crisis of enormous proportions, and of the characteristics of a health care delivery system that can actually deliver.