Against the bleak backdrop of a dismal 2020 and a worsening pandemic, November finally gifted the world some good news: the Pfizer and Moderna covid-19 vaccines were shockingly effective. We celebrated, a bit. Then the bickering began. How are we to distribute these vaccines? Here in the U.S., we have our experts on the Advisory Committee on Immunization Practices (ACIP) to give the states guidance on how to distribute these precious vaccines, and they have spoken. Their recommendation to vaccinate health care workers and nursing home residents and staff first is not unreasonable. However, I would argue it is unwise, and will lead to many thousands of unnecessary deaths; and that every state should be engaged in a vigorous conversation as to whether these guidelines should be followed. Allow me to explain.
First of all, the disclosures. As a family physician, and not someone who pursued training in infectious disease/epidemiology, or even faintly considered a masters degree in public health while in medical school, I claim no expertise in this matter. I am just the nerdy kid in the back of the room, jumping up and down, saying, “Please let’s think more about this!” Now, if we could pretend that while pursuing my Economics degree in the musty halls of Princeton University, I was studying at the elbows of faculty giants like Alan Blinder and Ben Bernanke rather than living in a beery haze of poker chips and basketball games, I might have a bit more credibility in what I want to say, which is this: we have lost sight of the concept of opportunity cost in managing this pandemic. Opportunity cost – put simply, the loss of benefit from a given choice not being made – has been ignored at our peril throughout this pandemic. We may have opted to shutter schools, close parks, and curtail businesses for noble reasons (and I endorsed many), but we often over-estimated the benefits and under-estimated the costs of their closures. So, now is the time to make up for all our errors in judgment, and really get this vaccine deployment right.
What is our mission?
The ACIP goals as stated are complicated: “to maximize benefits and minimize harms; promote justice; mitigate health inequities; and promote transparency.” That’s all very nice. However, in the short term, we have a problem: scarcity. Best estimates imply that we will have about 40 million doses of the Pfizer and Moderna (presuming imminent approval) vaccines to dispense in the initial round of shipments. Both are two-shot series, with boosters 3 and 4 weeks apart, respectively. Optimistic reports suggested we might have another 60 million in January, but concerns are already mounting about these next rounds. Demand greatly exceeds supply, in any case. Even if only 60% of those offered accept a vaccine, with some 21 million health care workers, 2 million care home residents, another 51 million elderly over the age of 65, and at least another 50 million essential workers not in the preceding groups, we have far more arms than we do injections.
2500 deaths. Every day. I am all for battling injustice and promoting social equality. Right now, though, I am mostly for keeping people alive. That is a mission I can stand for with this vaccine deployment. Want to help people of color? Vaccinate the people of color most likely to die. Let’s not get too fancy here.
It’s fair to disagree with my choice of mission. If your goal is social equity, or re-opening schools with minimal risks to teachers, or jump-starting the economy as rapidly as possible, you might choose to reduce transmission rather than severe illness and death, and focus on immunizing essential workers. We lack much data on whether these vaccines will be very effective in this regard; only the AstraZeneca/Oxford trial checked for asymptomatic cases, and their results were modest (a 28% reduction), but it’s reasonable to think the Pfizer and Moderna vaccines would prevent the majority of all infections. If your goal is to keep hospitals open and fully functional in their life-saving capacity, and avoiding the sort of nightmare headlines from the Mayo Clinic that we saw last month, then prioritizing health care workers makes sense.
However, I will observe these headlines appeared the week before Thanksgiving, at precisely the peak of the case curve in Minnesota. The Mayo Clinic has a staff numbering in the tens of thousands. Some elective surgeries were canceled. An estimated 93% of those staff infections were obtained in the community rather than the hospital; I doubt the other 7% were mired in feelings of betrayal that their choice to pursue a career tending the ill had left them with an illness. The Mayo Clinic did not crumble to the ground; or, if it did, no one bothered to write another article about it once cases began their decline in the region. Yes, the stress in these situations is real, and there is presumably some modest, impossible-to-estimate collateral loss of life in a community when hospital staffing is stretched thin. But, again, let’s not get too fancy here. In the long term, herd immunity is the goal and every health care worker and essential worker of every stripe should be offered a vaccine. Right now, though, every day counts – and the goal should be saving lives.
Principle #1: Give Life-Saving Vaccines to the People Most Likely to Die
This is my greatest beef with the current plan as endorsed by the ACIP and CDC. We might only have 19 million doses in our first round. I will not quibble with allocating two million to care home residents, as they have been responsible for roughly 40% of deaths in the U.S. I also won’t argue with administering to roughly another two million staff for these homes, since we cannot be sure that nursing home residents, the frailest, least robust individuals in our country, will mount an effective immune response to these vaccines. However, giving the next 15 million doses to health care workers does not compute by any calculus valuing lives saved. I know the arguments. They put themselves at risk. They are needed, healthy, for hospitals to function smoothly. They tend to roll up their sleeves more willingly than average Americans. The problem: they don’t die very often. Approximately 0.5% of US deaths from covid-19 have been health care workers. Approximately 80% of US deaths from covid-19 have been from those over 65. 80%. Versus 0.5%. How is this even a question?
Obviously, I am not the first to ring this bell; an ER physician and ICU doctor have published essays in recent days raising this point. There are the ethics: is it really right for me, as a 51 year old healthy, immensely-privileged white guy, to be given a vaccine in two weeks, while my 87 year old father living in our multi-generational household might wait another 2-3 months? We know “primum non nocere,” but what about, “primum non rapere” – “first do not steal?” More important, though, are the pragmatics: if we agree my father is on the order of 160X more likely to die of covid-19 than me, does it make sense to give me the vaccine first? If I truly cannot transmit the virus to him after immunization, okay – we have slightly dropped the odds that he is exposed to SARS-CoV-2. I’m hardly his only threat, though. What about my daughters, attending in-person school – not eligible for this round? My wife? His caregivers? His massage therapist? Let’s keep this simple: give the man his shot.
Some back-of-the-napkin calculations might explain my sense of urgency. Let’s say our present pace of about 2500 deaths per day continues to be the norm through January. Let’s also presume it will be February at the earliest before we have adequate supply to vaccinate every American over 65 who wants a vaccine. Add in the surprisingly positive data released on the performance of the Pfizer vaccine just 11-12 days after the first shot — in the time period before the booster, that first dose appeared to 80+% effective against symptomatic disease. (One week after the booster dose, efficacy rose to the much-quoted 95%.) The bottom line: it is possible that everyone who gets a shot in the next week will be 80% less likely to get a case of covid-19 by New Years Day.
Now let’s think about how people are dying of this disease. Roughly, those 2500 deaths break down as:
500 deaths age 65-74
500 deaths age 75 and older
1000 deaths in long term care settings (not included in above #s)
The appeal of vaccinating those in nursing homes is obvious – that number might drop to 200 per day within weeks. There are about 21 million Americans over 75 living in the community. This should be our target for every other available vaccine dose right now. At 80% efficacy, we could save 400 deaths per average January day if, much as the UK has done, we moved those initial shots earmarked for Health Care Workers to the very elderly. Even if we vaccinated 10 million of the most covid-facing medical staff – say, Emergency Dept and covid Unit maintenance staff, CNAs, nurses, clerks, physicians and other medical personnel – and pushed the other 10 million out, immediately, into the community, that could be 200 deaths saved every day from cases that would have been acquired in January. If the elderly would have been left unvaccinated through January, even if cases drop in half for February, that’s still 100 averted deaths on an average February day. It’s the gift that keeps giving! Health Care Workers? At a 0.5% proportion of deaths, that might be 10-15 deaths saved a day through January. The opportunity cost of moving all Health Care Workers to the front of the line ahead of the elderly easily could equate to over 5000 lives lost unnecessarily, from cases in January alone.
I am not the only one to notice this. The venerable Marc Lipsitch and a team of colleagues recently released their modeling data, which concluded that vaccinating the elderly would maximize lives saved. Per an interview with Scientific American: “‘Almost no matter what, you get the same answer,’ says Harvard epidemiologist Marc Lipsitch.
Vaccinate the elderly first to prevent deaths, he says, and then move on to other, healthier groups or the general population.” I generally feel a lot better about my opinion when I can hide behind Marc Lipsitch’s apron.
Now, given the observation that people of color in America die at a similar rate to a white cohort ten years older, I think it’s reasonable to consider lowering the target age for these populations. Doing so indirectly via prioritizing essential workers might be great for fans of social justice, but immunizing younger people of color rather than older people of color in order to tread lightly on people’s sensibilities is just another way of letting people of color die in disproportionate numbers from covid-19. As per a thorough study on the risk factors for dying of covid-19 in the UK, the idea is to get the vaccine to the people on the right side of this chart. Public health figures can communicate openly about their strategy and save lives.
Principle #2: Don’t Give Life-Saving Vaccines to People Who Are Already Immune
I think this one got caught in the political wash-out. Legitimate Infectious Disease/Epidemiology types seem reluctant to embrace immunity acquired from infection, perhaps to avoid the risk of aligning themselves with the White House/Great Barrington Declaration enthusiasm for Herd Immunity via natural infection. To endorse delaying vaccination, given that we do not know how enduring or protective prior infection will be in the months ahead, makes Public Health types nervous. I get it. I also get that every single vaccine we give to someone who is unlikely to benefit from it is a potential waste of a scarce resource.
Paul Romer, whose Nobel Prize in Economics suggests he got more mileage out of his studies in The Dismal Science than I did, waded into this subject with a Twitter post extolling the value of testing people for antibodies prior to vaccination. It generated an energetic conversation on multiple threads. One of Dr Lipsitch’s collaborators, Dan Larremore, joined in to voice support, as one part of the aforementioned pre-print makes the case that delaying vaccines for those who test positive for covid-19 antibodies will save lives early in vaccine deployment.
I agree with the gist of Dr Romer’s take as it spooled out in other threads — that we cannot wait for scientific certainty to make decisions in a setting like this. Only a handful of re-infections have been identified, 8 months after the initial wave of infections, implying protection is enduring for most. Arguably the most comprehensive analysis of B- and T-cell immunity in the blood of those recovered from infection found evidence for broad protection even 6 months later. While we have no means to test millions of Americans instantly for neutralizing antibodies, we can at least check for what is probably an excellent correlate for immunity — antibodies to the spike proteins — which should persist rather reliably for at least a year. Perfect? No. Adequate? Yes. You can walk into Krogers and walk out 15 minutes later with results for $25. I think the US government can secure a better deal than that. While the limited Pfizer data on the performance of their vaccine amongst those with a positive antibody test (to the less-useful N-protein, I will add) is rather inscrutable, what they published is quite clear: at their pre-determined end point of all infections beyond the first week post-booster, there was exactly one infection in the vaccine group and one infection in the (slightly larger) placebo group. Short-term benefit of vaccinating those with a positive antibody test per this data? Zero, indicating that immunity acquired from infection did the job.
Let’s return to that napkin again. Best estimates would hold that the 16 million identified cases of covid-19 in the US probably amount to more like 60 million or even 100 million actual cases; let’s split the middle and settle on 80 million total cases. Of the 64 million without a history of PCR positive test, if we estimate that a rapid antibody test to spike proteins correctly identifies 80% of these, we have 51 million more Americans with probable immunity. That’s 20% of people we can exclude from the first waves of vaccination with very little cost or effort. Carrying on with a fairly reasonable assumption that incidence rates do not vary greatly among age groups, that’s an extra 4 million Americans we can vaccinate from the first batch of 20 million doses. Roughly speaking, if 90% of our “skip first round” group has vaccine-like immunity from prior infection, and only 10-20% of our “vaccinate first round” group has it (from either ever-controversial pre-existing T-cell immunity or being a false negative antibody test) then we have succeeded in increasing the chances of those 4 million vaccines saving a life by five-fold or more.
Principle #3: Send Life-Saving Vaccines to the Places Where People are Most Likely to Die
My patients might not appreciate my advocacy for this position, but it stands to reason that if forced to allocate a scarce vaccine to different regions in the midst of a peaking pandemic, it makes sense to focus on those regions where people are most likely to die. Hawaii’s case load is stable, and we are averaging 1 death/day per million residents over the past week; while a state like Kansas is seeing a rise in cases and is already averaging 14 deaths/day per million. In the immediate future, those vaccines are 14X more likely to save a life in Kansas than Hawaii.
Assuming we might not want to start a civil war with vaccine allotments, there is still room for efficiency of allocation within states. Los Angeles County is averaging 55 deaths per day, while San Francisco County, at about 8% the population, is at 0.4. If current trends hold, that would imply a vaccine administered to an elder in LA is about 11 times more likely to save a life than one in San Francisco. I don’t want northern California to secede from their statehood. I do, however, want to save as many lives as possible with this vaccine. My napkin-back tells me we might have enough vaccine right now to immunize 50-60% of those Los Angelenos who will otherwise die next month, preventing about 700 deaths. Devote those vaccines to 10 million Northern Californians instead and we might prevent 60 deaths. Expand this regional math to all 50 states and the numbers could be considerable.
Principle #4: Give Life-Saving Vaccines As Fast as Possible to As Many People as Possible
The logistics around mass distribution of an ultra-cold storage, two-dose vaccine are challenging enough. Trying to keep the vaccines in storage for 3 (Pfizer) or 4 (Moderna) weeks to maintain perfect allocation for a booster shot adds to this challenge, and, more importantly, cuts in half the number of vaccine doses available right now. The US government recommends this practice, perhaps to avoid any appearance of legitimizing a one-dose regimen. However, we should be beyond appearances at this point. Former FDA head (and current Pfizer board member — but now is not the time for a Medical-Industrial Complex rant), Scott Gottlieb, is not a fan of holding back doses: “I don’t think we should be holding onto supply now, anticipating something goes wrong that’s going to cause a lot of other challenges. We should be taking some risk.”
From my first napkin: every additional 10 million doses in December to people over 75 who want a vaccine equates to 400 lives saved per day in January. The Pfizer trials ranged in their booster dose between 3 and 6 weeks. The perfect being the enemy of the good here, let’s trust efficacy won’t fall off a cliff if we have to wait a bit to deliver boosters. The messaging can be simple: “We expect the first dose to help within two weeks, but the booster will bring better and longer lasting immunity.”
Back when the field of medicine was young and fresh — 2019, maybe — the “Limitations” section in every published study allowed space for the authors to consider weaknesses in their data or ways in which it might not translate outside of their work. At its best, this discussion would help those with less expertise in the field better grasp the ways in which the study results might not apply smoothly to one’s clinical practice. In the era of covid-19, and the apparent race to publish and hold sway over the crowd, it strikes me that most “Limitations” sections only review the most obvious and superficial weaknesses of the study, rather than real flaws in study design or potential biases in assessing the data. With that in mind, I want to be sure to concede the potential flaws in my arguments.
First of all, I grant the irony in arguing to consider the opportunity cost of not distributing vaccines first to the elderly without really trying to quantify the opportunity cost of not giving them to health care and other key workers. Lives saved through better-staffed hospitals and reduced community transmission are just too hard to estimate, although I do assume them to be dwarfed by lives saved by prioritizing the elderly. Taking this argument further, perhaps nursing homes are not the ideal top priority. Given that the average new arrival to a nursing home only survives five months per this decades-long study; and that in the best of times, without a pandemic, nursing homes hold some 25% of the nation’s annual deaths for those over 65; dropping nursing homes on the priority list might seem cold or heartless, but not entirely irrational. I can’t argue if your moral philosophy favors a small chance of saving a young life over a greater chance of saving an elder life; but I will observe that your approach will not do much to empty ICUs next month.
The other obvious caveat is that, if these vaccines roll out with unexpected great speed — on the order of 150M doses by January — my concerns are largely misplaced. The opportunity cost of waiting to vaccinate those at highest risk shrinks as the anticipated delay shortens. (I will ask, however: have you seen anything in this country over the past 10 months to make you think a massive roll-out involving a brand new technology with über-complex supply chains is going to go smoothly? I didn’t think so.) Likewise, if vaccine hesitancy ends up being an even greater issue than anticipated, vaccine allocation will be less important than vaccine acquiescence. What’s more, if our current wave suddenly evaporates, choosing a focus on essential workers and breaking transmission chains will be far less costly than if the pandemic sustains its current pace another month or two. Finally, there’s a cost to adding in layers of complexity to a vaccine deployment. Adding an antibody test adds a small element of time and expense, but perhaps a larger element of mixed or confusing messages: “They sent me home without a vaccine because they said I was already immune.” Asking state officials to pick their regions for preferential vaccine allotment, or physicians to submit lists of highest risk patients, has a cost measured in time and stress, and could undermine efficiency and trust in the process. Deviating from an initial roll-out centered in nursing homes and hospitals, places that are equipped to distribute a vaccine and already contain the people eligible for the vaccine, creates the need to bring people from the community in to hospitals, clinics or pharmacies with only a week or two to prepare. I understand the point of view that we could end up best off just following the plan we have — the not-so-ancient proverb, “a shot in the arm is better than two in ultra-cold storage.”
Just because I can acknowledge the limitations of my argument does not mean I will stop jumping up and down to raise my concerns! I am sure that Hawaii is not unique in still forming its final plan to distribute the boxes of Pfizer vaccines beginning to land on our shores. Across the Pacific on the US mainland, the hospitalization curves are much higher, and so, too, are the stakes, in getting this roll-out done right. I defer to those experts in the world of epidemiology to parse out the details of how best to deploy these vaccines to maximize their benefit. My hope is that they are guided by one simple principle:
“Get the vaccine out to the people it is most likely to help while it is most likely to help them.” I would like to see this chapter of the US covid-19 response have a happy ending. Then I’ll sit back down again.