How Minnesota Does Breakthrough Events

Dave Dixon is dogged in his pursuit of the best data.  He asked the DOH to explain the breakthrough identification process.  We know there are issues in fully identifying those who are vaccinated and then matching them to cases, hospitalizations and deaths.  You can see from the answer below to questions 1 and 2 that the state consistently errs on the side of not identifying a case as vaxed.  It is apparent to me that the state is consistently misclassifying a large number of events as unvaxed when they are in fact occurring in vaxed persons.  This is to some extent intentional, as it supports the states messaging on getting vaxed.  But it is an atrocious betraying of public duty.

Here is Dave’s request:

“I am curious about the process MDH uses to identify breakthrough Covid cases. Can you briefly explain the process and data that is used? Specific questions: 1. Are cases reviewed electronically, such as in a data base process, or manually? 2. What information types are used to match cases to vaccination records; name, birth date, social security number, or some other identifying information? 3. What is the process to ensure that breakthrough cases are not missed? Is there a manual review or audit process? 3. How common is it for small errors in the data, such as name spelling, etc. to affect the matching process? Does MDH maintain any statistics on the incidence of matching errors? 4. What % of cases are interviewed to determine if they are actually vaccinated, but not in MDH’s data bases (such as vaccinated out of state, or VA, etc.)? Please provide any information that you are able to. If these types of questions are beyond the scope of the IDEPC Comment process please let me know and I can file data practice act requests instead.”

Here is the explanation DOH sent to Dave:

“Questions 1 and 2: All reported COVID-19 cases are reviewed manually. Case information is matched to vaccination information that is reported to the Minnesota Immunization Information Connection (MIIC) via a bulk matching process using patient first and last name and date of birth. If these demographics are not sufficient to identify an exact match (e.g. multiple people with the same first name, last name, and DOB are identified) then immunization information will not be pulled from MIIC. We also collect self-report vaccination information from patients and health care providers. In order for this information to be considered sufficient to determine if a case meets the vaccine breakthrough case definition, the self-reported information must contain the vaccine manufacturer and administration dates. Case and vaccination information are then reviewed to ensure they meet the vaccine breakthrough definition. This includes: 1) Checking that an individual has received an FDA authorized or approved COVID-19 vaccine; 2) Checking that an individual has received the appropriate number of doses to be classified as fully vaccinated or fully vaccinated and boosted, and 3) Comparing vaccination dates to specimen collection dates to ensure that the case occurred at least 14-days after the completion of the vaccination primary series or administration of a booster dose.

Question 3. Data undergo a series of both manual and automated checks every week to identify if errors have occurred. However, no process could completely capture every single vaccine breakthrough case in the state. We are currently improving the matching processor to take into account additional factors – this helps us identify more misspellings, reordering of names/last names/hyphenated names, and other common errors in data sets this large. Not all cases are interviewed and even with case interviews patient recall of exact vaccination dates may not be perfect, patients can choose not to report vaccination information, or may not readily have access to their official vaccination records. Not all providers report vaccination data to MIIC, including federal providers such as the VA, or out-of-state vaccination sites, however some do get reported to us and we take all the data we can, knowing that there are some gaps. In some instances, we do not have enough patient information to match a case to the vaccination record in MIIC. However, we estimate that we are able to match at least 90% of our cases to their vaccination information, therefore the impact on the overall rate is likely small.

Question 4. As we have transitioned to a more sustainable model for disease follow up, we are interviewing very few cases on a regular basis. We try to follow up more intensively where epidemiologically necessary/beneficial, but most cases do not get interviewed. In general, when we’ve had larger waves of case data and we’ve looked at how much vaccine data has come through via our matching process vs. self-report, the self-reported answers that were NOT reported through other means were a tiny percentage of the overall data. We try to be transparent in putting out all the data we receive or are able to match and we will continue to post new data to the website as it becomes available. However, vaccine breakthrough data is just a piece of what’s happening in terms of disease transmission and vaccination. Efficacy data from controlled studies is often updated and posted on the CDC website and through other sources – vaccine breakthrough data should not be mistaken or misinterpreted as true vaccine efficacy.”