Rationing Paxlovid based on race and ethnicity

The United States is currently recording over 700,000 new cases of Covid-19 per day and the number is rising rapidly.  Fortunately, vaccines are quite effective at preventing severe disease, and Pfizer’s anti-viral drug, Paxlovid is remarkably effective at preventing death and severe illness from Covid-19.  However, only 265,000 courses of Paxlovid are expected by the end of January, and Paxlovid needs to be taken early in the course of illness before symptoms are severe.  This means that supplies will be woefully inadequate during the Omicron wave.  Severe rationing seems to be inevitable. 

Should Black and Hispanic people get priority access to Paxlovid?

The New York State Department of Health recently released guidelines for Paxlovid eligibility (my italics, see formatting in original):

Oral antiviral treatment is authorized for patients who meet all the following criteria:

• Age 12 years and older weighing at least 40 kg (88 pounds) for Paxlovid, or 18 years and older for molnupiravir

• Test positive for SARS-CoV-2 on a nucleic acid amplification test or antigen test; results from an FDA-authorized home-test kit should be validated through video or photo but, if not possible, patient attestation is adequate

• Have mild to moderate COVID-19 symptoms

o Patient cannot be hospitalized due to severe or critical COVID-19

• Able to start treatment within 5 days of symptom onset

Have a medical condition or other factors that increase their risk for severe illness.

o Non-white race or Hispanic/Latino ethnicity should be considered a risk factor, as longstanding systemic health and social inequities have contributed to an increased risk of severe illness and death from COVID-19

Naturally, the inclusion of race and Hispanic ethnicity as risk factors led to moral outrage among right-wing commentators worried that Black people are about to get special treatment:

But . . . why would anyone familiar with racial politics in America think that Black and Hispanic people are likely to get preferred access to a scarce, life-saving medical treatment? 

In fact, this tweet is in bad faith.  As several people have pointed out, the guidance does not exclude White people from getting Paxlovid.  The problem is worse than this, however.  If you bother to click the links and read the guidance, it turns out that Black/Hispanic people are not automatically eligible for treatment with Paxlovid.  Race and ethnicity play a fairly modest role in determining who gets Paxlovid treatment – a role that may be justified. 

Having said this, there are serious problems with the NYDH guidance.  One problem is that NYDH does not present evidence that Black/Hispanic people are more vulnerable to Covid-19 than otherwise similar White people.  This means that the policy could be wrong on the merits – that is, it may save Black/Hispanic lives by letting a larger number of more vulnerable White people die.  The lack of evidence also makes the policy politically inflammatory and legally vulnerable.  And to top it off, it seems doubtful that the policy will do much to protect vulnerable Black/Hispanic people from omicron, and there are probably more useful steps NYDH could have taken to do this.   

In short, the policy seems to encapsulate an awful lot of what is wrong with racial politics in the United States, on both the right and the left.    

What the NYDH guidance actually says

Here is a passage from the NYDH guidance that people have ignored in their rush to be outraged:

While supplies remain low, adhere to the NYS DOH guidance on prioritization of anti-SARS-CoV-2 therapies for treatment and prevention of severe COVID-19 and prioritize therapies for people of any eligible age who are moderately to severely immunocompromised regardless of vaccination status or who are age 65 and older and not fully vaccinated with at least one risk factor for severe illness.

The document links to a risk assessment tool the NYDH has previously published, a tool that attempts to sort people by their risk of serious illness.  According to this document, the highest risk group consists of people who are 1) immunocompromised or 2) over 65, unvaccinated, and have at least one risk factor for severe illness.  Race/ethnicity is listed as a risk factor for severe illness. 

Race/ethnicity thus has a modest role in allocation decisions.  A 75-year-old unvaccinated Black woman would be placed into the same risk category as a 75-year-old unvaccinated White woman with lung disease.  However, Whites are still eligible for treatment (of course) and young, healthy Black and Hispanic people will not be getting Paxlovid anytime soon (also of course).

Is it fair to use race and ethnicity as risk factors?

NYDH treats race/ethnicity as a risk factor that can sometimes nudge a patient into a higher priority group.  Is this justified?

The answer is “yes” if

1) Our goal is to minimize the number of deaths from Covid-19,

2) Black/Hispanic people with Covid-19 are more likely to die than otherwise similar White people, and

3) We ignore practical problems:  a) legal and political constraints on policy (philosophers sometimes call moral theorizing that ignores political constraints “ideal theory”) and b) how the policy is likely to work in practice.

Minimizing deaths is widely endorsed in public health circles.  I will consider alternative policy objectives and practical problems later.

If our goal is to minimize deaths, and we are ignoring practical problems, the critical question is whether race/ethnicity is significantly predictive of severe disease among people who have Covid-19, once we control for easily identifiable risk factors.  An unvaccinated 75-year-old Black woman with no additional risk factors should have the same access to Paxlovid as an unvaccinated 75-year-old unvaccinated White woman with lung disease, provided they have the same risk of dying without Paxlovid.  On the other hand, if race/ethnicity do not predict mortality, then using race/ethnicity to allocate Paxlovid will increase the total number of deaths because lower-risk Black/Hispanic people will be prioritized over more vulnerable White people. 

The NYDH does not present evidence that Black/Hispanic people with Covid-19 are more likely to die than otherwise similar White people

Race and ethnicity could well be risk factors for severe disease, but as far as I can see, neither the CDC nor NYDH make a serious effort to show it. 

The NYDH guidance justifies using race/ethnicity as a risk factor by pointing to a CDC page that links to another page with a pretty vague discussion of racial/ethnic health disparities and Covid-19. 

Some of the causes of disparities listed by CDC, such as higher rates of occupational exposure, are not relevant to rationing Paxlovid, since decisions about who gets Paxlovid are made after people test positive.  To maximize lives saved, we would give Paxlovid to people who are most likely to die given that they are infected.  We would not give Paxlovid to people because they had a higher-than-average chance of getting infected.  (It is possible to argue that Blacks/Hispanics should be given some preference in access to Paxlovid to compensate for other inequities in health or higher risk of exposure.  I will return to this possibility.  But if our goal is to minimize lives lost, Paxlovid should be rationed according to risk of death among those with Covid-19.)

The CDC says that disparities in death rates among those with Covid-19 may reflect differences in health status.  This certainly seems plausible, but it does not tell us that Blacks/Hispanics are more likely to die of Covid-19 once we control for measurable differences in health. 

It is certainly possible that Blacks/Hispanics are more likely to die from Covid-19 even after adjusting for these risk factors, and, if so, counting race/ethnicity as a risk factor would indeed be justified.  My point is simply that NYDH needs to give us some reason to think that Blacks/Hispanics with Covid-19 are more likely to die than White people with the same health status. 

The evidence does not need to be ideal.  There is no need for a randomized trial.  Observational studies can provide relevant evidence.  We might count race/ethnicity as risk factors based on indirect evidence.  If Blacks/Hispanics are more likely to die of other respiratory diseases than similar Whites, that would count in favor of using race/ethnicity as a risk factor.  Similarly, if Blacks/Hispanics tend to get worse medical care than Whites, and there is reason to believe that this leads to higher Covid-19 death rates, that might weigh in favor of using race/ethnicity as risk factors when allocating Paxlovid (although as I will discuss in a future post, if the problem is access to timely testing, using race/ethnicity as a risk factor may not do much to fix this problem).

Unfortunately, as far as I can see, the CDC does not present relevant evidence on these points.  Although I could certainly have missed something, by simply citing the CDC, the NYDH is just handwaving.

Of course, nothing I have said here excuses the conservative critics of NYDH.  They could have rolled up their sleeves and reviewed the evidence on this point, or simply flagged this as an important question and asked NYDH to clarify why they believe that race/ethnicity is, in fact, a risk factor for developing serious disease.  Instead, they chose to misrepresent NYDH policy to score political points.

To be continued . . .