OHDSI Home | Forums | Wiki | Github

FDA/CBER COVID19 vaccine protocol posted and open for public comment; OHDSI comments drafted in MSTeams

(Patrick Ryan) #1


As most of you know, our colleagues at FDA/CBER, have the substantial responsibility of protecting US public health by monitoring the safety of biological products, which includes COVID-19 vaccines as they are authorized for use. As part of their strategy for safety surveillance, last Friday, they posted a protocol that outlines their proposed use of observational data, and it open for public comment.

The protocol can be found here: https://www.bestinitiative.org/wp-content/uploads/2020/12/C19-Vaccine-Safety-Protocol-2020.pdf

On behalf of the OHDSI community, @hripcsa and I have drafted comments and a response document, and have posted it in the OHDSI MSTeams environment under OHDSI/General/Files/FDA BEST COVID19 Vaccine Surveillance Protocol comments (link here for those with MSTeams account). We encourage others to review the FDA’s protocol and provide feedback. You can do so individually, or if you wish, you can review our comments and suggest edits/additions. Given the various activities in the OHDSI community that are highly related to this work, I suspect many will be interested, including @msuchard @schuemie @Christian_Reich @Daniel_Prieto @Rijnbeek @aostropolets @tduarte @David_Madigan @xintongli @Gowtham_Rao @Chris_Knoll @Azza_Shoaibi .

Given the holiday season is fast upon us, we’d like to submit our document by end of Wednesday, as an early Christmas gift to our FDA friends. So, if anyone wants to review the OHDSI response document, please suggest edits/comments by 10pmET on Tuesday.

Thank you all in advance for your collaboration and support of FDA and our other public health officials trying to keep us all safe!


Patrick and George

(Mark Danese) #2

Thanks Patrick. I don’t have access and can’t find a link to submit comments. So, if the OHDSI team hasn’t already found this, would you mind passing along this comment (assuming I am reading this correctly)?

This section on page 12 says that the clean window will be defined prior to the index date of vaccine administration. This is typical. However it doesn’t align with Figure 1 which shows a clean period being defined from the event, which doesn’t seem correct to me.

Text from protocol:
For each AESI observed within the risk window of interest, a clean window restriction defined prior to the COVID-19 vaccination will be implemented to more plausibly identify incident cases, as opposed to follow-up care to an initial diagnosis that occurred earlier.

Figure from protocol:

(Mengchun Gong) #3

Dear Patrick,

I sent the application for access to your draft and could you pass it?

This is great work and OHDSI should play a key role in this.

Best regards.


(Andrea Margulis) #4

I found what I was looking for and I am posting it here in case others are interested:
“The COVID-19 Vaccine Safety Surveillance: Active Monitoring Master Protocol will be posted for public comment from December 18, 2020 to December 28, 2020. The workgroup will consider all comments but will not reply to commenters. Any revisions made to the protocol subsequent to the public comment period will be noted in the change log document included in the amended protocol. Please send all comments to FDABEST@fda.hhs.gov.”
Source: https://www.bestinitiative.org/ (2020.12.23)

I missed the deadline for adding comments to OHDSI´s members´ comments. I could not see where I should be sending comments separately. Could you please share the link/email address/process for submitting comments to FDA?
Thanks in advance,

(Patrick Ryan) #5


Thank you all for your rapid and thoughtful feedback! I really appreciate it. It’s genuinely inspiring to see so many different people from various stakeholder groups, including collaborators from Europe, Middle East, and Asia-Pacific regions providing their insights to support the US FDA. It underscores that we’re all in this together to support and inform the global pandemic response, and all have an great opportunity to help our colleagues in public health around the world.

Attached is the final OHDSI community response document that @hripcsa submitted today.
For those who missed our deadline or have other thoughts to share, the FDA is still accepting comments until 28Dec2020. Additional details are here: https://www.bestinitiative.org/

I wish you all a Happy Holidays and I look forward to collaborating with you all in the New Year!



C19 vaccine safety protocol OHDSI comments 2020.pdf (1.1 MB)

(Patrick Ryan) #6


Our colleagues at FDA have posted a new protocol detailing their approach to calculating the background rates of adverse events of special interest (AESI) for the COVID vaccines, augmenting the initial protocol that was posted last month. The document (link here) is open for public comment until 8Jan2020.

FDA has again requested our community’s support to provide feedback. @hripcsa @msuchard and I will again coordinate a collective response. We will again use our OHDSI MSTeams environment to capture feedback. If you are interested in contributing, please share your feedback by 8pmET on Thursday, 7Jan2020, and that’ll give us Friday to clean up and send along to our FDA friends.

To get the ball rolling, I have opened a shared Word document under OHDSI/General/Files/FDA BEST COVID19 Vaccine Surveillance Protocol comments (direct link here for those with an OHDSI MSTeams account). I’ve put in some high-level bullets of some of my immediate observations, It’s currently raw, unedited and in no particular order, but I’m hoping it beats than just starting a blank document and telling you all ‘have at it’ :slight_smile: .

Thanks again for your rapid response and critical input to support FDA’s public health mandate.



(Patrick Ryan) #7


Thank you again for your rapid response to support our friends at the FDA. Based on the collective input of the community, @hripcsa and I prepared the attached document, which was submitted as our response today. We look forward to collaborating with you all as we continue to work with our partners at FDA, EMA and other public health agencies around the world to generate reliable evidence about COVID-19 epidemiology, prevention and clinical management from our international observational data network that can be used to promote better health decisions and better care.


Patrick and George

OHDSI comments to FDA COVID19 AESI background rate protocol.pdf (119.8 KB)

(Christian Reich) #8


Nice comment, and looks like the lack of controls was already addressed.

Question: I am struggling with the differential diagnosis between Encephalomyelitis and Transverse Myelitis proposed by the FDA. One problem is that in our cohort definitions, the Encephalomyelitis also includes the Transverse one, at least at the mapped ICD-10-CM level. (It also brings in a lot infectious conditions in, like Bacterial meningitis, unspecified or Influenza due to other identified influenza virus with encephalopathy.) Of course, we could tune the Conceptsets to clean that up and make them non-overlapping. But is that clinically possible? The Brighton Collaboration Case Definition is the same for both, splitting the problem into encephalitis (CNS-based), myelitis (peripheral) and disseminated encephalomyelitis (both). Does the distinction make sense?

(Matt Spotnitz) #9

@Christian_Reich I agree that it will be challenging to make phenotypes that differentiate these diseases. Encephalomyelitis and transverse myelitis can have similar clinical presentations and overlapping condition codes.

(Matt Spotnitz) #10

Generally, neurological diseases can have overlapping clinical presentations. Therefore, phenotyping them may be improved by use of laboratory and imaging results, which may be difficult to approximate from other variables. Would our research efforts benefit from measurement concept harmonization or extraction of data from imaging reports?

(Christian Reich) #11

Thanks, @mattspotnitz, that’s what I thought. ICD-10-CM G04 “Encephalitis, myelitis and encephalomyelitis” and its descendants (now part of Encephalomyelitis) makes it hard to have confidence that we can pull them apart.

However, I need to correct myself with what I said above about the infectious myelites in the definition, as well as the Transverse Definition being also part of the Encephalomyelitis one: I didn’t pay attention to the lacking checkbox in the “Descendants” column in the Conceptset definition. But still, there are other viral and parasitic myelites we may want to remove.

Definitely. Won’t make a difference for claims based definitions, but we should make use of what’s in the EHR if we can.

(Vojtech Huser) #12

The cohorts posted on Atlas are very interesting to learn from. You search ‘fda aesi’
Thank you for creating them !

Example here: https://atlas.ohdsi.org/#/cohortdefinition/335

OHDSI phenotype methodology should perhaps have a new feature to automatically add clean window as IC1 based on user’s definition of Initial Event.

Another nice feature is: Condition must be during dates in inpatient visit. E.g., in this definition https://atlas.ohdsi.org/#/cohortdefinition/344/definition

Finally, making “primary condition” international is a challenge. This definition limits the diagnosis by type https://atlas.ohdsi.org/#/cohortdefinition/348/definition I wish conceptsets would be supported for attributes. Issue filed here https://github.com/OHDSI/Atlas/issues/2424