[M T]

Thanks, g1313.  That's what I expected. 

Goose, the assertion "putting together a priority list less than a week before they needed to start administering vaccine" isn't consistent with the info in the article that g1313 linked to (which, to be fair, was posted after your posting).
--------

Even worse than I expected:  Not the kerfuffle amongst who got it first at Stanford, but Stanford's share of the vaccine to SCC.

The dashboard indicates Stanford got about 5,000 (+/- 500 as I'm estimating from a graph) Pfizer doses in the first week (about 1/3 of all of SCC's share of 17,550) and are expecting another 5000 Pfizer doses and 13000 Moderna doses.  (SCC hasn't said how many Pfizer doses it expects, but that again is 1/3 of expected 39,300 Moderna doses.)

I hope it is exceptional, but at least one of the Stanford resident went to a different hospital to get a vaccination, bypassing the allocation to Stanford and taking a dose away from someone else's allocation.

What about LTCF residents?  They were supposed to be in Phase 1a.   SCC doesn't list any LTCFs whose name begins with "Stanford".  Their allocation timeline doesn't mention any LTCF doses.

Out of roughly 60K doses to SCC for the first 2 weeks, how many are going to the people dying the most?
California guidelines for Phase 1a:

During Phase 1a of allocation, COVID-19 vaccine should be offered to the following persons in California:
. Persons at risk of exposure to SARS-CoV-2 through their work in any role in direct health care or long-term care settings.
. . o This population includes persons at direct risk of exposure in their non-clinical roles, such as, but not limited to, environmental services, patient transport, or interpretation.
. Residents of skilled nursing facilities, assisted living facilities, and similar long-term care settings for older or medically vulnerable individuals.

It appears that the CDPH included those residents, but then probably left room for misinterpretation in the wording of "Recommendation B" in (my emphasis added)

Persons exposed though[sic] work in health care or long-term care settings, by:
1. Type of facility or role
2. Location of facility
3. Attributes of individuals

Could SCC take this to mean that residents of LTCF were the last to get anything out of Phase 1a, because they are not exposed through work?
When I look at the Stanford allocations and compare them to the county allocations, it seems that through week 2, not one dose will go to a LTCF resident while all doses ONLY go to workers at the 11 hospitals in SCC.

It seems "th[r]ough work" was likely not intended to exclude LTCF residents. In Recommendation B1 (corresponding to "1. Type of facility or role"),
Tier 1 includes various hospitals and "Skilled nursing facilities, assisted living facilities, and similar settings for older or medically vulnerable individuals
Also, in concordance with ACIP, residents in these settings". (Note that the types of facilities are not ordered under the various tiers, so hospitals are not ordered ahead of SNF or paramedics.)

So, IMO, these guidelines indicate SCC should have been delivering vaccines to LTCFs and their residents concurrently with delivery to the hospitals, not subsequently.

-------

Flu shot records are kept for health care personnel at acute care hospitals in California. In 2018-2019, for the Stanford Healthcare (Santa Clara County only) + Lucille Packard Children's Hospital, there were 16824+6243 = 23,067 health care personnel ("Total number of Health Care Personnel (HCP) vaccinated. HCP include all paid (employee) and unpaid (licensed independent practitioners, other contract personnel and adult students/trainees and volunteers) persons working in health care settings and have the potential to transmit influenza to and from other HCP and patients.")
Stanford expects to have 23,000 doses of vaccine through this week.
The total number of HCP for the 11 hospitals in SCC was 63,824. The ratio of 23,067/63,824 matches closely to the ratio of total doses in the first week & Stanford's prediction of week 2.

It would seem SCC has not allocated vaccine to anything other than to health care personnel at acute care hospitals, even though the others (LTCF, paramedics, dialysis centers) are in the same tier.

---------
For Tuoumne county, the hospital HCP is 3.9% of the county population. For SCC, it is 3.3% of the county population. For San Francisco, it is 5.5% of the county population (the most in California). For some rural counties with a hospital, it is about 0.8%. If vaccine distribution is based solely on population, the smaller this number, the more quickly vaccine gets out to the general public).

[M T]

Thanks, g1313.  That's what I expected. 

Goose, the assertion "putting together a priority list less than a week before they needed to start administering vaccine" isn't consistent with the info in the article that g1313 linked to (which, to be fair, was posted after your posting).
--------

Even worse than I expected:  Not the kerfuffle amongst who got it first at Stanford, but Stanford's share of the vaccine to SCC.

The dashboard indicates Stanford got about 5,000 (+/- 500 as I'm estimating from a graph) Pfizer doses in the first week (about 1/3 of all of SCC's share of 17,550) and are expecting another 5000 Pfizer doses and 13000 Moderna doses.  (SCC hasn't said how many Pfizer doses it expects, but that again is 1/3 of expected 39,300 Moderna doses.)

I hope it is exceptional, but at least one of the Stanford resident went to a different hospital to get a vaccination, bypassing the allocation to Stanford and taking a dose away from someone else's allocation.

What about LTCF residents?  They were supposed to be in Phase 1a.   SCC doesn't list any LTCFs whose name begins with "Stanford".  Their allocation timeline doesn't mention any LTCF doses.

Out of roughly 60K doses to SCC for the first 2 weeks, how many are going to the people dying the most?
California guidelines for Phase 1a:

During Phase 1a of allocation, COVID-19 vaccine should be offered to the following persons in California:
. Persons at risk of exposure to SARS-CoV-2 through their work in any role in direct health care or long-term care settings.
. . o This population includes persons at direct risk of exposure in their non-clinical roles, such as, but not limited to, environmental services, patient transport, or interpretation.
. Residents of skilled nursing facilities, assisted living facilities, and similar long-term care settings for older or medically vulnerable individuals.

It appears that the CDPH included those residents, but then probably left room for misinterpretation in the wording of "Recommendation B" in (my emphasis added)

Persons exposed though[sic] work in health care or long-term care settings, by:
1. Type of facility or role
2. Location of facility
3. Attributes of individuals

Could SCC take this to mean that residents of LTCF were the last to get anything out of Phase 1a, because they are not exposed through work?
When I look at the Stanford allocations and compare them to the county allocations, it seems that through week 2, not one dose will go to a LTCF resident while all doses ONLY go to workers at the 11 hospitals in SCC.

It seems "th[r]ough work" was likely not intended to exclude LTCF residents. In Recommendation B1 (corresponding to "1. Type of facility or role"),
Tier 1 includes various hospitals and "Skilled nursing facilities, assisted living facilities, and similar settings for older or medically vulnerable individuals
Also, in concordance with ACIP, residents in these settings". (Note that the types of facilities are not ordered under the various tiers, so hospitals are not ordered ahead of SNF or paramedics.)

So, IMO, these guidelines indicate SCC should have been delivering vaccines to LTCFs and their residents concurrently with delivery to the hospitals, not subsequently.

-------

Flu shot records are kept for health care personnel at acute care hospitals in California. In 2018-2019, for the Stanford Healthcare (Santa Clara County only) + Lucille Packard Children's Hospital, there were 16824+6243 = 23,067 health care personnel ("Total number of Health Care Personnel (HCP) vaccinated. HCP include all paid (employee) and unpaid (licensed independent practitioners, other contract personnel and adult students/trainees and volunteers) persons working in health care settings and have the potential to transmit influenza to and from other HCP and patients.")
Stanford expects to have 23,000 doses of vaccine through this week.
The total number of HCP for the 11 hospitals in SCC was 63,824. The ratio of 23,067/63,824 matches closely to the ratio of total doses in the first week & Stanford's prediction of week 2.

It would seem SCC has not allocated vaccine to anything other than to health care personnel at acute care hospitals, even though the others (LTCF, paramedics, dialysis centers) are in the same tier.

---------
For Tuoumne county, the hospital HCP is 3.9% of the county population. For SCC, it is 3.3% of the county population. For San Francisco, it is 5.5% of the county population (the most in California). For some rural counties with a hospital, it is about 0.8%. If vaccine distribution is based solely on population, the smaller this number, the more quickly vaccine gets out to the general public).

[Goose]

Thanks, g1313.  That's what I expected. 



Goose, the assertion "putting together a priority list less than a week before they needed to start administering vaccine" isn't consistent with the info in the article that g1313 linked to (which, to be fair, was posted after your posting).

I think my original post is still applicable. The post from g1313 contains the statement

“Our algorithm, that the ethicists, infectious disease experts worked on for weeks … clearly didn’t work right,” Tim Morrison, the director of the ambulatory care team, told residents at the event in a video posted online.

That is the only timeline info I see in the post fro g1313. While having people "working on" an an algorithm for weeks is better than a few days, it is still unacceptable when you have months to get ready. Most importantly, the output of that algorithm was not available until at most days before it was to be actualized. This left no opportunity to "sanity check" the results and resulted in the flapex. That is simply irresponsible. Using an algorithm that is brand new (never tested) without vetting it's results are reasonable is just plain nuts, IMHO. The algorithm undoubtedly worked exactly as designed. The results just didn't match the design goals. Allowing no time to verify the results did in fact meet the design goals and having no procedure to verify that it did is always a recipe for failure.

Irene Chen, an MIT doctoral candidate who studies the use of fair algorithms in health care, suspects this is what happened at Stanford: the formula’s designers chose variables that they believed would serve as good proxies for a given staffer’s level of covid risk. But they didn’t verify that these proxies led to sensible outcomes, or respond in a meaningful way to the community’s input when the vaccine plan came to light on Tuesday last week. “It’s not a bad thing that people had thoughts about it afterward,” says Chen. “It’s that there wasn’t a mechanism to fix it.”

One must add that if the results aren't available until 2 days before they are to be used there wasn't time for any "mechanism to fix it.”, even if it existed.

[Goose]

Thanks, g1313.  That's what I expected. 



Goose, the assertion "putting together a priority list less than a week before they needed to start administering vaccine" isn't consistent with the info in the article that g1313 linked to (which, to be fair, was posted after your posting).

I think my original post is still applicable. The post from g1313 contains the statement

“Our algorithm, that the ethicists, infectious disease experts worked on for weeks … clearly didn’t work right,” Tim Morrison, the director of the ambulatory care team, told residents at the event in a video posted online.

That is the only timeline info I see in the post fro g1313. While having people "working on" an an algorithm for weeks is better than a few days, it is still unacceptable when you have months to get ready. Most importantly, the output of that algorithm was not available until at most days before it was to be actualized. This left no opportunity to "sanity check" the results and resulted in the flapex. That is simply irresponsible. Using an algorithm that is brand new (never tested) without vetting it's results are reasonable is just plain nuts, IMHO. The algorithm undoubtedly worked exactly as designed. The results just didn't match the design goals. Allowing no time to verify the results did in fact meet the design goals and having no procedure to verify that it did is always a recipe for failure.

Irene Chen, an MIT doctoral candidate who studies the use of fair algorithms in health care, suspects this is what happened at Stanford: the formula’s designers chose variables that they believed would serve as good proxies for a given staffer’s level of covid risk. But they didn’t verify that these proxies led to sensible outcomes, or respond in a meaningful way to the community’s input when the vaccine plan came to light on Tuesday last week. “It’s not a bad thing that people had thoughts about it afterward,” says Chen. “It’s that there wasn’t a mechanism to fix it.”

One must add that if the results aren't available until 2 days before they are to be used there wasn't time for any "mechanism to fix it.”, even if it existed.

[BostonCard]

I would add that this appeared to be a case of trying to be too cute by half.  While recognizing that any rules based process (including a random draw) is by definition an algorithm it appears that they tried to do something complicated like assign points based on characteristics (like age, whether they are working on the front line, etc.) when in fact they should have just kept it simple: vaccinate front line workers seeing hospitalized patients in descending order of age first.

BC

[BostonCard]

I would add that this appeared to be a case of trying to be too cute by half.  While recognizing that any rules based process (including a random draw) is by definition an algorithm it appears that they tried to do something complicated like assign points based on characteristics (like age, whether they are working on the front line, etc.) when in fact they should have just kept it simple: vaccinate front line workers seeing hospitalized patients in descending order of age first.

BC

[Genuine Realist]

I would add that this appeared to be a case of trying to be too cute by half.  While recognizing that any rules based process (including a random draw) is by definition an algorithm it appears that they tried to do something complicated like assign points based on characteristics (like age, whether they are working on the front line, etc.) when in fact they should have just kept it simple: vaccinate front line workers seeing hospitalized patients in descending order of age first.

BC

KISS, always, if possible.

[Genuine Realist]

I would add that this appeared to be a case of trying to be too cute by half.  While recognizing that any rules based process (including a random draw) is by definition an algorithm it appears that they tried to do something complicated like assign points based on characteristics (like age, whether they are working on the front line, etc.) when in fact they should have just kept it simple: vaccinate front line workers seeing hospitalized patients in descending order of age first.

BC

KISS, always, if possible.

[Goose]

I think one fact that must be kept in mind when "prioritizing" vaccine delivery is that virtually any method of doing it is arguably imperfect. Different people have different opinions about what is "best", and there is no recognized optimum. Worse still, whatever is chosen has to be possible to implement. A perfectly fair plan that can't be implemented is useless. Speed also matters. Getting more people vaccinated sooner may be more important than who those people are exactly. It isn't going to be perfect. It is just necessary that it be "good enough".

[Goose]

I think one fact that must be kept in mind when "prioritizing" vaccine delivery is that virtually any method of doing it is arguably imperfect. Different people have different opinions about what is "best", and there is no recognized optimum. Worse still, whatever is chosen has to be possible to implement. A perfectly fair plan that can't be implemented is useless. Speed also matters. Getting more people vaccinated sooner may be more important than who those people are exactly. It isn't going to be perfect. It is just necessary that it be "good enough".