This is interesting. Thanks for sharing. Yet another advantage of a national health care system (“Medicare for all”) would be better, more consistent, more centralized data. Health care (not just health data) in this country is in shambles. Exhibit A, a million Americans dead and one of the very highest death rates from covid in the world. It’s a global embarrassment that none of our leaders is talking about. A million dead and we’re just going about our business as if there’s not a huge problem here in this wealthy nation.
I stopped the video in the middle to send Dr. Rivers' statement "but if every jurisdiction is collecting different data or collecting it in different ways, it's really hard to aggregate that into a national picture that really gives us a sense of what's happening across the country," to my husband, who is a data scientist working for a formulation and contract manufacturing pharmaceutical company. Over and over again, he has heard various groups and projects say that they've got good data to work with, only to find, when the data is finally made available to him, that there are gaps and holes and no two sources collected the same data. Lots of lip service from the company about using data science and machine learning going forward, but no real investment in improving the data collection systems, programs, etc. Various groups that would need to participate in improving and using new data collection systems are not motivated to do so because it would be extra work for an outcome that they aren't invested in. My husband would like all data collection to be motivated by a project statement--the generators of the data need to be invested from the start, to understand how the improvements will benefit them in the long run. It is much more easily said than done, when one part of the company (or country) doesn't do the same thing as the other part of the company (or country). For my husband, better data means better modeling, control, etc. The money it saves the company or the ways it improves process outcomes can sometimes be abstract to the operators or researchers who collect the data. For the country, better data means better public health outcomes for the country, which can feel very nebulous and abstract to too many citizens. Better communication about why consistent data is so important is key to improving the situation. In a nutshell, that is what both Dr. Jetelina and Dr. Rivers are doing. Difficult problem. Thank you both for working on it!
Super fascinating! Full disclosure: I was a railroad data geek. Gathering, collecting and getting data from disparate sources to play nicely together was one of my work things.
Complete, timely and accurate data doesn't exist anywhere, and is likely impossible, but shouldn't be an excuse for not trying to improve. Glad to hear there are people working at it.
I wonder if the most bang for the buck might occur by working from the bottom up. That is, getting the actual data workers together to suggest standards rather than trying to mandate standards from the top down that would get stuck in the bureaucracy.
Not when some states have rules that restrict what they want revealed. Otherwise, yeah, it's a great idea. I just retired and left a state where the Governor had strong ideas of what he wanted reported, because he didn't want anyone getting the idea COVID might really be a problem. The epi's I interacted with at the State had never seen a governor so interested in what their routine messages to the public said, and certainly never cared about the TB data.
I hear you! Seems to be a problem in any large organization....particularly when there are public facing measures on display - especially when things are going badly. Gathering the data elements might be able to get done by decisions made at or near the bottom of the food chain. Displaying measures based on these and/or making it available to others would be more difficult, for sure, but at least the data is there.
Also, a general principle should be keeping the data at the most granular level possible. Summary level stuff terrible to deal with (as YLE indicated). No excuse not to in this day and age.
Thanks for all the great work you do in pointing out gaps in our nation's pandemic response.
With the upcoming wave of high school and college graduations, given the CDC's new "risk assessment" map based on hospitalizations (2-3 week lagging indicator), I am concerned that there is no easy way for elderly grandparents to assess their own risk. Many of them will be traveling by plane to attend indoor, unmasked graduation ceremonies, particularly in areas of high transmission such as the northeast.
Paxlovid and the other drugs for people recently infected have been kind of a game-changer for people I know who are elderly or otherwise at elevated risk. Paxlovid is *quite powerfully effective* as an antiviral. It really shortens and weakens the illness a lot. Some people can't take Paxlovid because it interacts with other meds they take, but there are alternatives that also work well. It's amazing how few people know about Paxlovid. Spread the word!
My understanding is that Pfizer ran its trials for Paxlovid only on the unvaccinated, only on those who had never had Covid, and before omicron. In the real world, when Paxlovid is given to the vaccinated, they don’t always clear the virus, and Covid symptoms return. Also, Paxlovid must be given in the first five days, and many people no longer test or cannot get results and doctor permission that quickly. It’s also in short supply and will be rationed (by vaccination status, by age, by comorbidities, by geography, maybe even by ethnicity) if we have another surge. It’s unclear if it will work against future variants. If you have kidney issues or take certain meds, you should avoid Paxlovid. So is it a game changer? I hope so, but it’s too early to tell.
Yes, I am aware of these limitations on Paxlovid, and understand that it was given to unvaccinated people. It is not actually in very short supply currently. Here is a site where you can see Paxlovid availability in every state, and how many courses of treatment each drugstore is holding:https://rrelyea.github.io/paxlovid/
A lot of the Paxlovid is sitting unused on pharmacy shelves, because patients don't know to ask for it and doctors sometimes don't know how to find a pharmacy that stocks it. Here is the Paxlovid situation in Oklahoma, for instance. https://rrelyea.github.io/paxlovid/?state=OK
There are currently 10316 courses of treatment in the state. Look down at the drugstores: Many of their graphs are *flat*: They got a shipment and have distributed little or none of it.
Site was built by a Microsoft engineer whose wife needed Evusheld and had a terrible time finding it. Purpose is to help other people find needed drugs.
2% of the subjects in the original study, all unvaxed, also had rebound. I haven't yet seen any evidence that vaxed people are more likely to suffer rebound than unvaxed. Maybe I missed it, though -- have you seen something?
I haven’t seen anything suggesting that vaxed are more or less likely to rebound than unvaxed, yet I hear the concern is that every time Paxlovid fails, the viruses that are selected out are smarter and more resistant, which risks breeding out a super-variant. The same concern holds for molnupiravir, which is more widely used in Europe. There’s also a question of whether different variants have different responses to Paxlovid, and whether Paxlovid will be less effective against B4 and B5. Lastly, there’s concern that Paxlovid isn’t an adequate solution to long covid. But this is the problem with our nation’s “eggs all in one basket” response so far: first betting everything on vaccines, next betting everything on antivirals.
Yes, I wonder about Paxlovid-resistant variants too. Still the failure rate of Paxlovid seems to be far lower than the failure rate of vaccination -- at least failure in the sense of vaxed people getting a mild or asymptomatic, but still contagious, case.I'm not sure which warrants worrying about more. At this point I'm at the limit of my interested-layman grasp of things.
"My understanding is that Pfizer ran its trials for Paxlovid only on the unvaccinated, only on those who had never had Covid, and before omicron." I'm pretty sure all that's true. It makes sense that they used unvaxed subjects. Their primary outcome measure was number of subjects hospitalized or dead. If they had used vaxed subjects they would have had very few hospitalizations or deaths per thousand, even for subjects on the placebo, because vaxing improves people's odds so much that the study would have had to have many thousands of subjects in both the Paxlovid and the placebo groups, so that there were enough deaths to give them valid data about frequency of deaths and hospitalizations .
As for the idea that being vaxed affects how effective Paxlovid is -- I don't seen anything about that except anecdote and speculation so far. I did have a talk with a scientist whose field is viruses, antiviral drugs, etc. Scientist said that Paxlovid works in a way that is completely independent of what the patient's immune system is up to. Paxlovid sort of snips the virus up into useless little pieces. Since Paxlovid's mode of action is completely independent of what the immune system is doing to attack the virus, it seems as though the effects of Paxlovid and vax-related immunity should be additive, rather than canceling each other out to some extent. (Though of course things don't always work out the way one feels like they will.)
The concept that Paxlovid is hampering the immune response is but one of several competing ideas as to the noted relapse/rebound infection seen in about 2% of those treated. As one of several, it'll have to be considered and tested. It is not yet "science".
Hey, I get that you are trying to be helpful, but it is kind of irritating to have you jump onto my comment and inform me that the idea that Paxlovid is hampering the immune response is "not yet 'science'." I just SAID that in the comment you are replying to. I said I hadn't seen anything but anecdote and speculation about the idea. "Haven't seen anything but anecdote and speculation" MEANS "not yet science."
No, have seen him on Twitter though. I try to avoid podcasts, but am impressed by this written summary. Can you tell me a place where I can find the archive of his summaries, and subscribe or something to get new ones as though come out?
I felt much more calm about my grandma (who is boosted) when I read a recent nature paper showing vaccines were 99% effective against death. Her estimated risk went from 16% chance of dying if infected(based on another model) to 0.16%. If grandparents are not boosted and vaccinated I'd feel much more nervous. I do know there are other bad outcomes besides death, but somehow the paper just really made feel good because 99% is just so effective. Here is a link: https://www.nature.com/articles/s41591-022-01699-1
In a distributed public health surveillance system, such as we have in the US, actors at different points in the system have different strengths and opportunities, and different functions. An epidemiologist working at the local has all kinds of local context for the case reports, and the flows of syndromic surveillance and laboratory data, that are available in real time from their community. But they don't automatically have a view of similar data from neighboring communities or states, or for the US as a whole. An epidemiologist working at the national level may be able to see patterns across data from multiple jurisdictions that would not be noteworthy in the various jurisdictions themselves. So communication between workers at various levels is paramount in making the system work well. Also, the purposes of data collection and analysis are different at the local, state and national levels. Case reports for reportable diseases at the local level are most important for driving followup and intervention around individual cases, and for local cluster and outbreak detection. Discerning larger patterns over time and space is important, but not the only reason for data collection. At the national level, the detection of patterns over time and space, and their investigation, is the primary reason for data collection. With modern information systems, a national worker can have access to raw data from the local level just about as fast as the local worker, but won't have the advantage of the cleaning and editing that happen at the local level.
I had an idea about a way to get decent data about case counts: Why not have a system where a random group of individuals in each state, a group that's representative of the state's demographics, are selected to test themselves say once or twice a week. They could be paid in cash or lottery tickets or something. Or, we could make it like jury duty -- no pay, just your turn to do it this month. Have people send in photos of their rapid test results and use time stamps on them to make sure they're genuine.
Really!I didn't know that. Thought maybe my idea was impractical in some way I hadn't thought of. Why the hell isn't the US doing it, then? Can you send me to some info about how the UK does it?
When I worked on data analysis for fisheries, we often had to address problems that arose from different groups and agencies having different standards and priorities for the data they collected.
It was not unusual to encounter situations where a group would collect data on “their favorite species” and lump the rest into “other”. Of course, each group would have a different favorite species (depending on what they were studying), which made the “other” category awfully nebulous, and well, … oh well.
Thank you. I’m envious that you had the chance to meet with Dr. Rivers, and very appreciative for the excellent interview and questions.
I’m a data geek, as are we all… the painful lack of data has made communicating the concerns and risks much too difficult and has allowed disinformation to prosper because we lacked the information we needed to counter it.
And I agree with @Dan Jepson about seeking input from the PH community.
I suspect Caitlin can leverage a bit of the work on crowdsourced data and statistical analys to make sense of home RAgtest results, but her point about reporting bias were spot on.
Data geek here, but I prefer the term "Quant" as I've been called one. So data collection standardization is to me paramount. The worst of all worlds is to run an info aggregation system reporting to external end users, where the daily, weekly, and quarterly input is coming in subject to periodic and arbitrary changes in protocols from the reporting entities. Here in Oregon, the OHA has made numerous changes in reporting since early 2020. Usually the changes have been improvements as more reporting data elements were added to the picture, but others were a bit baffling to me at least. There are too many semi-independent reporting baronies and fiefdoms, both public and private. Here in Oregon we were flying blind for quite a while since there was woefully little testing and elr data was lagging but now we are up to speed through the very hard work of the folks there. Widely available private testing though has muddied the picture again alas.
Working with sparse and disparate data has sorta been the hallmark of one of my careers. Figuring out the differences and how to safely and appropriately normalize them is critical. Not easy, but critical.
Thank you! Looking at the current pandemic, I feel that a key to the work of the public health cohort has been data disaggregation with an eye to getting a better grasp of how the pandemic was affecting the entire population broken down into various groups. Ideally, the more dimensions the better, but there is a point of diminishing returns of course in analysis... In any case, this pandemic is going to have a long tail and there will be full employment for the quants studying it- likely for decades.
I suspect we, and Dr. Rivers' team, will end up having to look at some of the rather significant work that's gone into crowd-sourced data. If we got sufficient fragmented and disparate, yet related, datasets we could get useful statistics out of that...
Great informative discussion! After 17+ years on the public health disease intervention front lines several critical points/needs were identified/mentioned. Please seek input from local, district, regional, and statewide public health staff on what is needed and known/anticipated barriers as a critical step in developing any strategy prior to implementation. I acknowledge the messiness of this approach or procedure process.
This is really helpful information. It has been much harder than I would have ever expected to get data-based answers to the basic questions people are interested in-what is my risk? What is my communities risk? Is the vaccine still working?
Wonderful work! Thank you for being such a good influence on the influencers! How exciting to get this validation. This is yet another sign that for all its imperfections and mistakes, the right person is leading CDC. I recently heard a one hour interview with Dr Walensky and realized how we have all been hearing only soundbites from her for 2 years. It was good to see her sharp intelligence, humility and determination to keep learning and improving shine through. Sharing it here with your readers https://milkeninstitute.org/video/rochelle-walensky-cdc-director
Dr. Jetelina, could you address this preprint study (see below) that purports to show that the mRNA vaccines did not reduce all-cause mortality but that the adenovirus-vector vaccines did? I also include the perspective and analysis of the preprint study by Martin Kulldorff, an epidemiologist. Thanks.
Thanks, it does come across as less than objective, but I can't pin down why. Also, I am not facile enough with stats to adequately analyze the original paper (The Lancet preprint). I hope Dr. Jetelina considers it worth her (and our) time to look at it and offer her critique.
This is interesting. Thanks for sharing. Yet another advantage of a national health care system (“Medicare for all”) would be better, more consistent, more centralized data. Health care (not just health data) in this country is in shambles. Exhibit A, a million Americans dead and one of the very highest death rates from covid in the world. It’s a global embarrassment that none of our leaders is talking about. A million dead and we’re just going about our business as if there’s not a huge problem here in this wealthy nation.
I stopped the video in the middle to send Dr. Rivers' statement "but if every jurisdiction is collecting different data or collecting it in different ways, it's really hard to aggregate that into a national picture that really gives us a sense of what's happening across the country," to my husband, who is a data scientist working for a formulation and contract manufacturing pharmaceutical company. Over and over again, he has heard various groups and projects say that they've got good data to work with, only to find, when the data is finally made available to him, that there are gaps and holes and no two sources collected the same data. Lots of lip service from the company about using data science and machine learning going forward, but no real investment in improving the data collection systems, programs, etc. Various groups that would need to participate in improving and using new data collection systems are not motivated to do so because it would be extra work for an outcome that they aren't invested in. My husband would like all data collection to be motivated by a project statement--the generators of the data need to be invested from the start, to understand how the improvements will benefit them in the long run. It is much more easily said than done, when one part of the company (or country) doesn't do the same thing as the other part of the company (or country). For my husband, better data means better modeling, control, etc. The money it saves the company or the ways it improves process outcomes can sometimes be abstract to the operators or researchers who collect the data. For the country, better data means better public health outcomes for the country, which can feel very nebulous and abstract to too many citizens. Better communication about why consistent data is so important is key to improving the situation. In a nutshell, that is what both Dr. Jetelina and Dr. Rivers are doing. Difficult problem. Thank you both for working on it!
Super fascinating! Full disclosure: I was a railroad data geek. Gathering, collecting and getting data from disparate sources to play nicely together was one of my work things.
Complete, timely and accurate data doesn't exist anywhere, and is likely impossible, but shouldn't be an excuse for not trying to improve. Glad to hear there are people working at it.
I wonder if the most bang for the buck might occur by working from the bottom up. That is, getting the actual data workers together to suggest standards rather than trying to mandate standards from the top down that would get stuck in the bureaucracy.
Not when some states have rules that restrict what they want revealed. Otherwise, yeah, it's a great idea. I just retired and left a state where the Governor had strong ideas of what he wanted reported, because he didn't want anyone getting the idea COVID might really be a problem. The epi's I interacted with at the State had never seen a governor so interested in what their routine messages to the public said, and certainly never cared about the TB data.
I hear you! Seems to be a problem in any large organization....particularly when there are public facing measures on display - especially when things are going badly. Gathering the data elements might be able to get done by decisions made at or near the bottom of the food chain. Displaying measures based on these and/or making it available to others would be more difficult, for sure, but at least the data is there.
Also, a general principle should be keeping the data at the most granular level possible. Summary level stuff terrible to deal with (as YLE indicated). No excuse not to in this day and age.
Thanks for all the great work you do in pointing out gaps in our nation's pandemic response.
With the upcoming wave of high school and college graduations, given the CDC's new "risk assessment" map based on hospitalizations (2-3 week lagging indicator), I am concerned that there is no easy way for elderly grandparents to assess their own risk. Many of them will be traveling by plane to attend indoor, unmasked graduation ceremonies, particularly in areas of high transmission such as the northeast.
Paxlovid and the other drugs for people recently infected have been kind of a game-changer for people I know who are elderly or otherwise at elevated risk. Paxlovid is *quite powerfully effective* as an antiviral. It really shortens and weakens the illness a lot. Some people can't take Paxlovid because it interacts with other meds they take, but there are alternatives that also work well. It's amazing how few people know about Paxlovid. Spread the word!
My understanding is that Pfizer ran its trials for Paxlovid only on the unvaccinated, only on those who had never had Covid, and before omicron. In the real world, when Paxlovid is given to the vaccinated, they don’t always clear the virus, and Covid symptoms return. Also, Paxlovid must be given in the first five days, and many people no longer test or cannot get results and doctor permission that quickly. It’s also in short supply and will be rationed (by vaccination status, by age, by comorbidities, by geography, maybe even by ethnicity) if we have another surge. It’s unclear if it will work against future variants. If you have kidney issues or take certain meds, you should avoid Paxlovid. So is it a game changer? I hope so, but it’s too early to tell.
Yes, I am aware of these limitations on Paxlovid, and understand that it was given to unvaccinated people. It is not actually in very short supply currently. Here is a site where you can see Paxlovid availability in every state, and how many courses of treatment each drugstore is holding:https://rrelyea.github.io/paxlovid/
A lot of the Paxlovid is sitting unused on pharmacy shelves, because patients don't know to ask for it and doctors sometimes don't know how to find a pharmacy that stocks it. Here is the Paxlovid situation in Oklahoma, for instance. https://rrelyea.github.io/paxlovid/?state=OK
There are currently 10316 courses of treatment in the state. Look down at the drugstores: Many of their graphs are *flat*: They got a shipment and have distributed little or none of it.
By the way, the data for the site I sent you to is drawn directly from the government site giving data about drug availability and distribution: https://healthdata.gov/Health/COVID-19-Public-Therapeutic-Locator/rxn6-qnx8
Site was built by a Microsoft engineer whose wife needed Evusheld and had a terrible time finding it. Purpose is to help other people find needed drugs.
Awesome data and links, thank you!
What did you have in mind about that -- what would the benefit be? Not saying there isn't one, just don't know what you have in mind.
2% of the subjects in the original study, all unvaxed, also had rebound. I haven't yet seen any evidence that vaxed people are more likely to suffer rebound than unvaxed. Maybe I missed it, though -- have you seen something?
I haven’t seen anything suggesting that vaxed are more or less likely to rebound than unvaxed, yet I hear the concern is that every time Paxlovid fails, the viruses that are selected out are smarter and more resistant, which risks breeding out a super-variant. The same concern holds for molnupiravir, which is more widely used in Europe. There’s also a question of whether different variants have different responses to Paxlovid, and whether Paxlovid will be less effective against B4 and B5. Lastly, there’s concern that Paxlovid isn’t an adequate solution to long covid. But this is the problem with our nation’s “eggs all in one basket” response so far: first betting everything on vaccines, next betting everything on antivirals.
Yes, I wonder about Paxlovid-resistant variants too. Still the failure rate of Paxlovid seems to be far lower than the failure rate of vaccination -- at least failure in the sense of vaxed people getting a mild or asymptomatic, but still contagious, case.I'm not sure which warrants worrying about more. At this point I'm at the limit of my interested-layman grasp of things.
"My understanding is that Pfizer ran its trials for Paxlovid only on the unvaccinated, only on those who had never had Covid, and before omicron." I'm pretty sure all that's true. It makes sense that they used unvaxed subjects. Their primary outcome measure was number of subjects hospitalized or dead. If they had used vaxed subjects they would have had very few hospitalizations or deaths per thousand, even for subjects on the placebo, because vaxing improves people's odds so much that the study would have had to have many thousands of subjects in both the Paxlovid and the placebo groups, so that there were enough deaths to give them valid data about frequency of deaths and hospitalizations .
As for the idea that being vaxed affects how effective Paxlovid is -- I don't seen anything about that except anecdote and speculation so far. I did have a talk with a scientist whose field is viruses, antiviral drugs, etc. Scientist said that Paxlovid works in a way that is completely independent of what the patient's immune system is up to. Paxlovid sort of snips the virus up into useless little pieces. Since Paxlovid's mode of action is completely independent of what the immune system is doing to attack the virus, it seems as though the effects of Paxlovid and vax-related immunity should be additive, rather than canceling each other out to some extent. (Though of course things don't always work out the way one feels like they will.)
The concept that Paxlovid is hampering the immune response is but one of several competing ideas as to the noted relapse/rebound infection seen in about 2% of those treated. As one of several, it'll have to be considered and tested. It is not yet "science".
Hey, I get that you are trying to be helpful, but it is kind of irritating to have you jump onto my comment and inform me that the idea that Paxlovid is hampering the immune response is "not yet 'science'." I just SAID that in the comment you are replying to. I said I hadn't seen anything but anecdote and speculation about the idea. "Haven't seen anything but anecdote and speculation" MEANS "not yet science."
No, have seen him on Twitter though. I try to avoid podcasts, but am impressed by this written summary. Can you tell me a place where I can find the archive of his summaries, and subscribe or something to get new ones as though come out?
I felt much more calm about my grandma (who is boosted) when I read a recent nature paper showing vaccines were 99% effective against death. Her estimated risk went from 16% chance of dying if infected(based on another model) to 0.16%. If grandparents are not boosted and vaccinated I'd feel much more nervous. I do know there are other bad outcomes besides death, but somehow the paper just really made feel good because 99% is just so effective. Here is a link: https://www.nature.com/articles/s41591-022-01699-1
In a distributed public health surveillance system, such as we have in the US, actors at different points in the system have different strengths and opportunities, and different functions. An epidemiologist working at the local has all kinds of local context for the case reports, and the flows of syndromic surveillance and laboratory data, that are available in real time from their community. But they don't automatically have a view of similar data from neighboring communities or states, or for the US as a whole. An epidemiologist working at the national level may be able to see patterns across data from multiple jurisdictions that would not be noteworthy in the various jurisdictions themselves. So communication between workers at various levels is paramount in making the system work well. Also, the purposes of data collection and analysis are different at the local, state and national levels. Case reports for reportable diseases at the local level are most important for driving followup and intervention around individual cases, and for local cluster and outbreak detection. Discerning larger patterns over time and space is important, but not the only reason for data collection. At the national level, the detection of patterns over time and space, and their investigation, is the primary reason for data collection. With modern information systems, a national worker can have access to raw data from the local level just about as fast as the local worker, but won't have the advantage of the cleaning and editing that happen at the local level.
CMS is also the governing body over all plans subject to the Affordable Care Act, which includes millions of people with private insurance. The hospital data they require for the PHE is for all patients at the hospital -- not just Medicare and Medicaid patients. https://www.hhs.gov/sites/default/files/covid-19-faqs-hospitals-hospital-laboratory-acute-care-facility-data-reporting.pdf
I had an idea about a way to get decent data about case counts: Why not have a system where a random group of individuals in each state, a group that's representative of the state's demographics, are selected to test themselves say once or twice a week. They could be paid in cash or lottery tickets or something. Or, we could make it like jury duty -- no pay, just your turn to do it this month. Have people send in photos of their rapid test results and use time stamps on them to make sure they're genuine.
This is done in the UK, very successfully.
Really!I didn't know that. Thought maybe my idea was impractical in some way I hadn't thought of. Why the hell isn't the US doing it, then? Can you send me to some info about how the UK does it?
When I worked on data analysis for fisheries, we often had to address problems that arose from different groups and agencies having different standards and priorities for the data they collected.
It was not unusual to encounter situations where a group would collect data on “their favorite species” and lump the rest into “other”. Of course, each group would have a different favorite species (depending on what they were studying), which made the “other” category awfully nebulous, and well, … oh well.
Thank you. I’m envious that you had the chance to meet with Dr. Rivers, and very appreciative for the excellent interview and questions.
I’m a data geek, as are we all… the painful lack of data has made communicating the concerns and risks much too difficult and has allowed disinformation to prosper because we lacked the information we needed to counter it.
And I agree with @Dan Jepson about seeking input from the PH community.
I suspect Caitlin can leverage a bit of the work on crowdsourced data and statistical analys to make sense of home RAgtest results, but her point about reporting bias were spot on.
Thanks!
Data geek here, but I prefer the term "Quant" as I've been called one. So data collection standardization is to me paramount. The worst of all worlds is to run an info aggregation system reporting to external end users, where the daily, weekly, and quarterly input is coming in subject to periodic and arbitrary changes in protocols from the reporting entities. Here in Oregon, the OHA has made numerous changes in reporting since early 2020. Usually the changes have been improvements as more reporting data elements were added to the picture, but others were a bit baffling to me at least. There are too many semi-independent reporting baronies and fiefdoms, both public and private. Here in Oregon we were flying blind for quite a while since there was woefully little testing and elr data was lagging but now we are up to speed through the very hard work of the folks there. Widely available private testing though has muddied the picture again alas.
Working with sparse and disparate data has sorta been the hallmark of one of my careers. Figuring out the differences and how to safely and appropriately normalize them is critical. Not easy, but critical.
Thank you! Looking at the current pandemic, I feel that a key to the work of the public health cohort has been data disaggregation with an eye to getting a better grasp of how the pandemic was affecting the entire population broken down into various groups. Ideally, the more dimensions the better, but there is a point of diminishing returns of course in analysis... In any case, this pandemic is going to have a long tail and there will be full employment for the quants studying it- likely for decades.
I suspect we, and Dr. Rivers' team, will end up having to look at some of the rather significant work that's gone into crowd-sourced data. If we got sufficient fragmented and disparate, yet related, datasets we could get useful statistics out of that...
What can I say! long neglect is hard to fix quickly
Great informative discussion! After 17+ years on the public health disease intervention front lines several critical points/needs were identified/mentioned. Please seek input from local, district, regional, and statewide public health staff on what is needed and known/anticipated barriers as a critical step in developing any strategy prior to implementation. I acknowledge the messiness of this approach or procedure process.
YUP!
This is really helpful information. It has been much harder than I would have ever expected to get data-based answers to the basic questions people are interested in-what is my risk? What is my communities risk? Is the vaccine still working?
Wonderful work! Thank you for being such a good influence on the influencers! How exciting to get this validation. This is yet another sign that for all its imperfections and mistakes, the right person is leading CDC. I recently heard a one hour interview with Dr Walensky and realized how we have all been hearing only soundbites from her for 2 years. It was good to see her sharp intelligence, humility and determination to keep learning and improving shine through. Sharing it here with your readers https://milkeninstitute.org/video/rochelle-walensky-cdc-director
Dr. Jetelina, could you address this preprint study (see below) that purports to show that the mRNA vaccines did not reduce all-cause mortality but that the adenovirus-vector vaccines did? I also include the perspective and analysis of the preprint study by Martin Kulldorff, an epidemiologist. Thanks.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4072489
https://brownstone.org/articles/have-people-been-given-the-wrong-vaccine/
Thanks, it does come across as less than objective, but I can't pin down why. Also, I am not facile enough with stats to adequately analyze the original paper (The Lancet preprint). I hope Dr. Jetelina considers it worth her (and our) time to look at it and offer her critique.
Excellent. Thanks again.