The Real Results of Lockdowns: Protecting the Rich, Infecting the Poor and Minorities?

We desperately need coordinated, on-going seroprevalence testing, especially in places like Massachusetts. The reason is not so that we can provide immunity passports, or say “See, we already have herd immunity!” --that threshold is unknown at this point in any case. The reason is so that we understand the results of our public health policies, and are able to respond and improve them as we wade through the remainder of this epidemic. This is particularly important for COVID-19, given the widely divergent outcomes depending on who is infected.

My governor, the governor of Massachusetts, says he has no intention of doing widespread seroprevalence (antibody) testing in the Commonwealth. He has said he believes that is the job of the CDC. The CDC is in fact undertaking widespread anti-body testing for COVID-19, using blood samples from bloodbanks. This will give us all some very good information, but it won’t give us information that we can use RIGHT NOW, and it won’t give us information about geographic and socioeconomic distribution of cases within our own localities.

Understanding the real distribution of cases from an age, geographic, racial and socioeconomic perspective is imperative to being able to understand the results of our public health policy. Raw case data in Massachusetts, and especially in Boston, points very strongly to a deeply unequal disease burden among poorer communities and communities of color. Looking at the raw case data, it looks as though the actual results of our strong lockdown have not been to meaningfully arrest the progress of the disease, but rather to build the foundation of our herd immunity on the backs of the most vulnerable in our community--which, if true, in the long-run will increase the mortality rate we see in Massachusetts from COVID-19.

Below is a chart showing all of the neighborhoods of Boston. The chart is ranked from lowest to highest rate of infection, based on confirmed PCR tests. Fenway, at 0.4%, is the lowest, Hyde Park, at 3.1% is the highest. That is nearly an 8-fold difference in observed infection from the least exposed to the most exposed. It is imperative that we understand if this difference bears itself out in antibody studies, because it is very possible that this difference in exposure is a direct result of our public health policies. Our public health policies have directed those who can work from home to do so (usually wealthier knowledge workers), and kept essential workers (often poorer and minority) on the job and under assault by the virus. With a largely shuttered economy, the wealthy and young are nearly completed removed from the path of the disease, while the old and poor are exposed. The fact that the infection rate by neighborhood shows a strong inverse correlation with a college education (nearly a pre-requisite for the ability to work from home) is strongly suggestive that this is exactly what is happening.

Table 1: Boston COVID-19 Case Data as of May 18, 2020

Sources:

1) Boston neighborhood infection data, provided by the Boston Public Health Commission as of May 18.

2) Population, demographic, education info, United States Census, American Community Survey 5-year estimates. Extracted for each Zip code using Wolfram Alpha.

3) Analysis, Emily Burns

The following chart, which plots the over- or under-representation of infection based on population against % population with a Bachelor’s or more, puts the chart above into starker relief.

Chart 2: Geographic Breakdown of COVID-19 Infections vs. Education in Boston

Sources:

1) Boston neighborhood infection data, provided by the Boston Public Health Commission as of May 18.

2) Population, demographic, education info, United States Census, American Community Survey 5-year estimates. Extracted for each Zip code using Wolfram Alpha.

3) Analysis, Emily Burns

This data in and of itself ought to be setting off alarms among our political leaders and public health officials, concerned that we are disproportionately infecting those whom we are on-record as believing to be the most at-risk.

However, because we simply expect that minorities will fare worse, we are chalking this data up to poorer health within these communities. Our public health policy debate does not currently take into account the possibility that our response to the crisis may be directly responsible for the asymmetrical distribution of infection across communities.

Looking at the data above, if you take the percent of infected in each zip code, as well as the population demographics in that zip code, if there were even distribution across all races represented in each zip code, one would expect that 35.6% of Boston’s COVID-19 infections would be amongst African-Americans. The actual number is 38%. Thus, while we are seeing a slight over-representation, the number is not wildly off, given the locations to which the virus has been shunted. This does not make it any more just, but it does mean that the injustice can be remedied by spreading the disease burden more widely. But we won’t know unless we get good seroprevalence data, which our governor has stated he has no intention of doing.

The city of Boston recently released the results of a targeted seroprevalence study in just those neighborhoods that are hardest hit in the chart above. The study did not look at the wealthier, better-educated, whiter neighborhoods. Had it done so, we would have known whether or not these trends in case data are representative of the real path of the disease. That study (whose raw data has not been released) showed that in those neighborhoods, the number of people exposed was actually 5 times higher than case numbers would indicate. If indeed the number is only 5x off, Boston’s case data would be the closest representation to the truth that exists worldwide. The closest any other city or state has gotten has been 10x, in New York, and 11x in Indiana. In those two states, seroprevalence data showed they were undercounting cases by “only” 90%. Boston is apparently only undercounting by 80%. Which is good news and bad news. The good news is that our case numbers are more accurate than other places. The bad news is that it means that Massachusetts really is the deadliest place in the world to get coronavirus—especially if you are a minority. This seems like something we might expect our public health officials to be trying to address by getting every kind of data they can get their hands on.

The reasons are manifold why I believe alarm bells ought to be going off in our public health officials’ heads due to what looks to be unequal geographic, ethnic, and socioeconomic exposure. It is deeply unjust to force the poorest among us to bear the immunological burden of this disease. But beyond the simple injustice, doing so may also exacerbate the lethality of the disease, increasing the real mortality rate of the disease in Massachusetts, and other locales, by pursuing hard lockdown policies that protect the rich at the expense of the poor.

Covid-19 is like a river flowing through our population. When it first came on the scene, it was racing through our population, like the Colorado river through a narrow cataract. Social distancing allows it to spread out and flow more slowly. People who are out of harm’s way are effectively on islands, and the river of infection flows around them. By forcing the closure of schools and non-essential businesses, the people on our islands are effectively children, people who can work from home (typically wealthier), and the newly-unemployed. The river of coronavirus is still flowing through the population, but now its only available (or most common) targets are essential workers and those people with whom they come into closet contact. Thus, we are in a very real way protecting the young and rich, and infecting the poor and old.

There are three primary mechanisms that preferentially infecting poorer, minority communities could be increasing the lethality of COVID-19 in Massachusetts.

First, and the only one that seems to get any play, is that minorities are at higher risk for the kinds of co-morbidities that predispose people to a worse outcome for COVID-19—obesity, diabetes, hypertension, etc.

However, it is likely that this is the most trivial of the three. Co-morbidities increase the likelihood of death from COVID-19, but by far the biggest predictor of death is age. As of May 17th, of New York City’s 15,800 some odd deaths, less than 4% were between the ages of 0-44. 25% between the ages of 0 and 65. Leaving the remaining 75% over 65. Of those 15,888 deaths, only 0.4% of them did NOT have a co-morbidity.

Table 2

Source: Screenscrape of https://www1.nyc.gov/assets/doh/downloads/pdf/imm/covid-19-daily-data-summary-deaths-05172020-1.pdf

Thus, if we want to decrease the lethality associated with COVID-19, we need to keep it away from the elderly. Shunting the disease to minority communities does the exact opposite. Minorities live in multi-generational households at twice the rate of whites in the U.S.—meaning that if you have an essential worker who is living in a multi-generational household, you are not just exposing that worker, you are exposing their family members, often older family members as well.

The third way that shunting the disease to minority communities increases the lethality of COVID-19, is that while minorities are under-represented generally in medical professions, they are over-represented in home healthcare. This means that through increasing exposure of minority communities, we are also indirectly increasing the exposure to the elderly in our population (of all races) who rely on home healthcare services. The disproportionate percentage of deaths amongst whites in Boston shown in the chart below argues strongly for this. Either African-Americans in Boston have a mortality rate that is half that of whites, or the whites who are being infected in the city are predominantly elderly, which would support the hypothesis that younger, wealthier white workers who can work from home are staying out of the way of the infection, while elderly whites who are in assisted living facilities, or rely on home healthcare, are unable to stay out of harm’s way.

Table 3

The goal of hard lock downs is to stop the virus. If you believe you are succeeding in stopping the virus, perhaps it is acceptable to disproportionately infect those who are most at-risk of death. It seems unethical to me, but I guess that has to be the rationale of those supporting this position. However, if there is any indication—and there’s a hell of a lot more than “any”—that you are failing, and that the results of your efforts to stop the virus are instead merely slowing it, and shunting it to the portion of your population that is least likely to survive, it seems like those policies deserve to be revisited and revised.

There is a reasonable amount of suggestive data nationwide that hard, authoritarian lockdowns are increasing mortality rates, whereas softer, more communitarian approaches that spread the disease burden more generally are decreasing mortality rates. I will not say that we ought to compare NY or Massachusetts to other states, but I think there are other places that make compelling arguments that are demographically, climactically, and population-density matched, places that differ only in the political leaning—a reasonable proxy for the populace’s adherence to lockdown mandates. Below are comparisons between California and Texas, Colorado and Utah, and Wisconsin and Minnesota.

Table 4

Deaths/million is a moving figure, which is in some respects an indication of the degree of spread through the population. But there is unquestionably an element of mortality to it, as well, as the more people at higher risk (the elderly) that the disease impacts, the higher the final mortality rate will be. It cannot possibly be that lesser compliance with social-distancing and earlier opening could have decreased transmission. It seems rather that this more laissez faire approach has helped to spread the disease burden more evenly across the community, to the parts of these communities that are least likely to die, specifically, amongst the young. Nationally, 65% of our population is under 50. If the disease were primarily spreading in that population, one would expect markedly lower deaths—in Massachusetts observed mortality in the under 50 group (which is 73% of Massachusetts’ population) is 0.19%. Given that virtually no state is testing people under 65 who do not have multiple symptoms and underlying conditions, spread in this group would also go almost completely unrecorded in terms of number of confirmed cases.

One final comparison. Sweden recently released antibody tests showing that at the end of April in Stockholm, to-date its hardest hit area, antibody tests showed that 7.3% of people had antibodies, indicating prior infection. This has been cited uniformly by U.S. news outlets as evidence of a failure of Sweden’s “herd immunity” strategy. However, this merits a bit more scrutiny. In Massachusetts, we are not pursuing a herd immunity strategy (or at least we say we aren’t). Nonetheless, in the town of Brookline where antibody testing was performed in mid-May--just a few weeks after those performed in Stockholm--7% of residents were shown to have been exposed. The study results for Boston the week prior showed 12.5% of residents exposed (9.9% had antibodies) or currently (2.6% currently infected). Thus, despite having one of the earliest lockdowns in the U.S., and some of the highest levels of compliance with the lockdown, we are only a few weeks behind Sweden in terms of our levels of exposure.

At the same time, the arc of our disease has been far more deadly. Let’s say April 27th was the “end of April” when the tests in Sweden were administered. At that time, when Sweden showed 7.3% of Stockholm (and 5% nationally) with antibodies, the country’s deaths/million were 227. On May 12th, when the town of Brookline administered its antibody tests, Massachusetts had 746 deaths/million. Meaning that at roughly the same point within the disease’s arc through our population, Massachusetts was 3x as deadly as Sweden. Sweden’s chief epidemiologist is currently estimating 20% of Stockholm has been exposed and the deaths/million are 404. As of May 27th, our deaths/million were 931, with presumably less of our population exposed. Which of these is the failing strategy?

It is routinely argued that we cannot be compared to Sweden, because Sweden is almost completely white, whereas we have minorities. But the presence of minorities does not explain this discrepancy. Massachusetts is 88% white, with 931 deaths/million. Louisiana, which has also had a terrible outbreak, is 66% white, and has only 571 deaths/million.

Beyond this, the difference in observed mortality rates—the number of dead over the number of confirmed infections—amongst races is not a factor of 3. In fact, based on confirmed cases, the observed mortality rate in Boston (as seen in Table 3 above) is nearly twice as high amongst whites and Asians as that observed in African-Americans, and 4x as high as that observed in Hispanics. The same holds true for statewide data (Table 5 below). I.e., the data we have shows that blacks and Hispanics are less likely than whites to die from COVID-19 in Massachusetts than whites.

Table 5

Do I believe that the mortality rate for whites is actually 2-4x higher than that for minorities? No, I do not. What seems to me to be the most likely reason for the difference in the observed mortality rates is that amongst minorities, young and old are being infected—because they or their parents or partners are out doing “essential” work. While younger whites are being sheltered due to their parents being able to work from home, and keeping them on strict lock-down. It may also be due to sampling bias, with hard-hit communities of color with large amounts of essential workers being tested more, catching more of the asymptomatic infections among the young.

The following is the case and fatality data for Massachusetts as of May 25, 2020.

Table 6

Massachusetts is a particularly young state, with 73% of its citizens under 50. Which ought to mean that if COVID-19 were circulating freely in this group, that Massachusetts would be seeing an especially low number of deaths and mortality rate, perhaps like that of Utah (currently at 33 deaths/million--Iceland is 29--despite never having a formal lockdown). Even with COVID-19 presumably circulating more the most at-risk of Massachusetts' under-50 population, the crude fatality rate for those under 50 is 0.19%. This is however, only for those people who have tested positive. Boston’s seroprevalence study showed that 80% of cases went un-reported, about the same number as seen in the 18-50 population in New York State. That would mean that the real mortality rate for this group is more like 0.04%--this squares nearly exactly with the CDC's current best estimate for Infection Fatality Rate (IFR) for under 50, .05%). If 60% of the population must be exposed to this disease for us to have herd immunity, (which is not certain, it could be higher, but it is increasingly looking like it could also be much lower) that would mean that the total death toll in Massachusetts would only have been around 1500—assuming only people in those groups were effected.

But even that could be improved upon. 98.6% of all deaths in Massachusetts had co-morbidities. From the New York City data presented above, we know the proportion amongst 0-50 was 96%. We know from New York State data that the specific co-morbidities observed in the 18-44 age group pre-disposing towards death were hypertension, diabetes, and obesity. Massachusetts is a particularly healthy state, with only 22.6% of the 18-44 being obese. Additionally, both diabetes and hypertension are linked to diabetes, but mercifully, have particularly low prevalence in this age group. Nationally, diabetes in the 18-44 group is only 4.2%. Hypertension, only 7.5%, nationally within this group, with half of those cases being found amongst people with diabetes. All of these co-morbidities are observed in higher rates amongst minorities, meaning that if the disease were not running primarily through this population but through the much larger and healthier white under-50 population in Massachusetts, we might expect to see vanishingly small mortality rates and deaths in Massachusetts--like what is being seen in Utah. Would we be able to shield everyone outside of this age group, or within this age group who was at risk? No. But we couldn't possibly do worse than we are doing now.

Our lockdown approach looks to be doing the exact opposite. Especially for COVID-19, mortality rates are malleable. At the moment, it looks like we are massaging ours to make it is as large as possible, worse possibly, even than Louisiana, one of the least-healthy states in the union. All of this despite having demographics very much in our favor.

All of this brings us back to the main point of this piece. Because we have chosen not to perform widespread, ongoing antibody studies, we know neither how many of us, nor which of us is truly being infected. We have said that our goal with lockdowns is to stop the spread of the virus and protect at-risk populations. The available data say that we are failing colossally in both of these goals. We have the tools to verify if this is in fact the case. We have the tools to improve our public health outcomes and decrease the absolute number of deaths, and our population fatality rate. But for some reason, we are stubbornly refusing to use those tools.

I simply cannot understand for what possible reason we would not want to answer these questions—they are quite literally ones of life and death. We, the people, deserve answers. Asking for these answers is not subversive or “alt-right." These are reasonable questions that people throughout the Commonwealth should be asking. These are questions which our elected officials ought to be falling all over themselves trying to answer.