Conflicting CDC Disclosures on Source of Vaccine Efficacy "Denominator Data"
Executive Summary:
The immunization registry used to determine the vaccination status of COVID vaccine study subjects appears to be different from the immunization registry used to determine the number of vaccinated people in the population in CDC studies, at least according to some disclosures. If that is the case, inequality between the two registries can produce the appearance that COVID vaccination has a protective effect when, in fact, it has none.
Two Types of Immunization Registries
Each individual’s COVID vaccinations are recorded in two parallel immunization registries, in the United States; a regional registry and the CDC registry. Similar information is recorded in the two immunization registries except that personal identifiers are removed from the records in the CDC registry. Typically, the registry region is the state where the vaccination was administered. In California, for example, records of COVID vaccinations are collected by the California Immunization Registry.
The regional and federal registries are independent of one another. Immunization reporting to the CDC registry appears to be a requirement for all providers of COVID vaccination. COVID immunization reporting to the regional/state registries may be limited by regulations. Patient privacy may play a role in limiting sharing of immunization information with regional/state immunization registries.. Also, reporting to regional/state registries for COVID immunization has not been uniformly required. In California, for example, it appears that reporting of COVID vaccination was not mandatory until recently.
COVID-NET
CDC clearly discloses that the COVID hospitalization-by-vaccination-status is derived from the regional/state immunization registries. However, CDC has provided conflicting disclosures about whether, additionally, the hospitalization-by-vaccination-status is based on the CDC immunization registry.
COVID-NET is a CDC program that estimates rates of hospitalization according to COVID vaccination status. “Hospitalization by vaccination status” are published on the CDC website. The website also provides estimates of “rate ratios” that compare the rate of hospitalization amongst the unvaccinated with the rate amongst the vaccinated. COVID-NET findings have also been published in journal formal. In one publication, calculation of the rate ratios was defined:
“… incidence (cases per 100,000 person-weeks) was calculated by dividing the total number of unvaccinated hospitalized persons by the sum of unvaccinated persons in the underlying population each week; the same method was used for incidence calculations in vaccinated persons with and without a booster dose. Incidence rate ratios and 95%CIs were calculated.”
The definition can be expressed as follows:
“State immunization information system data were linked to cases …”
However, the origin of the two denominator quantities, “Population of vaccinated” and “Population of unvaccinated” is described in various ways:
A CDC website suggests that "numerator" and "denominator" data are exclusively derived from regional/state immunization registries. A footnote to the website for “Hospitalization by vaccination status” makes the following statement: “State-based immunization information systems (IIS) data provide the number of residents within defined geographic areas who have received one or more doses of COVID-19 vaccine every week (denominator data). These data are used to define the number of people in a geographic area who received their primary vaccination series, a booster or additional dose status or are unvaccinated, based on the number of doses received, the vaccine product and the date(s) of vaccination administration.” The term “State-based immunization information systems” may refer to the region/state immunization registries but that is not entirely clear. The period of the described study is January, 2021 to the present.
A journal article suggests that "denominator" data is derived from the CDC immunization registry. Havers et al (A) reported on COVID-NET stating: “… vaccine coverage for the underlying catchment area were determined by IIS data, as previously described …”. The reference is to Moline et al. The methodology is described there as follows: “Population level vaccination coverage was determined using deidentified person-level COVID-19 vaccination data reported to CDC by jurisdictions, pharmacies, and federal entities through the IISs, …. Vaccine Administration Management System, … or direct data submission.” This does not seem to be referring to the regional/state immunization registry system. The period of the described study is January 1, 2021, to April 30, 2022.
A preprint states that "denominator" data is derived from the CDC immunization registry. Havers et al (B) reported on COVID-NET. A figure caption stated: “COVID-NET catchment population by vaccination status a and week, used as denominator in population-based-rate calculations …” A footnote states: “Data from CDC immunization information systems county-level data for the Coronavirus Disease 2019 …” In this description of the methodology used in COVID-NET, the “denominator data” clearly originates from the CDC immunization information system. The period of the described study is January 1 – July 24, 2021.
A government publication suggests that COVID-NET "denominator" data may be derived from state immunization information systems. Havers et al (C) states: "Vaccination status ... was determined for individual hospitalized patients and for the catchment population using state immunization information systems data." However, no estimates of the rate of hospitalization by vaccination status is provided in this publication. The only outcome related to COVID vaccination that is provided in this publication is: "Nearly one half of adults hospitalized during the BA.2 period had received a primary vaccination series and ≥1 booster or additional dose." The period of the described study is June 20, 2021–May 31, 2022.
Conclusion
Differences between the regional/state immunization registries and the CDC immunization registry could impact the apparent effect of COVID vaccination on the rate of hospitalization. For example, if the apparent size of the vaccinated population is reduced in the regional/state immunization registry relative to the CDC registry, then the apparent rate of hospitalizations that are attributed to vaccination would appear to be disproportionately small relative to the general population. In that case, a vaccine that provided no protection against COVID-19 might appear effective. Furthermore, there is speculation, of major deficiencies in the regional/state immunization registries.
Given this potential flaw in CDC studies of COVID vaccines, the CDC finding of a protective effect of COVID vaccination is suspect. Furthermore, proof of the protective effect of vaccination against COVID-19 has been the bedrock of the US response to this disease. It is startling that the origin of the supporting data is uncertain.