Updated: Oct 20, 2020
The colossal failure of 7 states to protect their most vulnerable populations hides the broad success of the rest of the U.S. in "flattening the curve." While many, including the CDC and much of the coronavirus task force, seems to have moved on from "flattening the curve" to the fanciful idea of eradicating the disease, the country as a whole signed on to the suspension of our civil liberties and destruction of our economy only as a temporary measure, and only to ensure hospitals were not overwhelmed. We have prevailed in that goal beyond the wildest dreams of epidemiologists. In March, epidemiologists said that if we could stretch this out over 40 weeks (instead of the usual 20 for flu) and keep critical cases to a "manageable" 50,000 at a time, that would be a pretty great outcome. As it stands, we have stretched the epidemic out over 33 weeks (and it is not yet over), we have never had more than 15,000 patients in the ICU (about 11% of capacity), and at no point were more than 6000 patients on ventilators (of which we have 170,000).
Despite what ought to be being hailed as an incredible triumph, we are routinely shamed by our public health officials for not doing enough. As frustrating as that is, the true insult comes when the 7 states that have performed the worst according to every possible metric are held up as examples of success. Why are we not looking at the other 43 states? States whose death rates and unemployment rates are 1/3 of these states? States whose schools are back in session? The goal of this article is to provide a glimpse of the nation's success as a whole, and which states have managed the pandemic well, using the measures that really matter--deaths, and economic carnage.
Let's start with a chart. The 7 early hot spot states, NY, NJ, CT, MA, RI, MI and LA, the "curve crushers" as we'll call them, saw 27 deaths/million/day--7 times the rest of the country, which never exceeded more than 4 deaths/million in a given day. If you look at the data, which is the real definition of "following the science," it becomes clear that these early hotspot states are the farthest thing from a model of how to "flatten the curve". Yes, there curves are flat now, but they scaled a mountain of corpses before getting onto that level ground. Not only were their daily highs 7x higher, their ultimate death tolls were 3x that of the rest of the country.
Source: The COVID Tracking Project. Rolling 7 day average daily deaths/million as of 10/10/20, "Early Hot Spot" states (NY, NJ, CT, MA, RI, MI and LA), vs. the rest of the country. Analysis and visualization provided by Emily Burns.
The chart that you usually see--which is shown below--shows all daily COVID-19 deaths for the U.S. This shows two clear humps. But what is doesn't show, and what the graph in figure 1 above does, is that that first hump was driven almost exclusively by 7 states whose populations totaled only 55.5 million. The second hump is the rest of the country, 273 million people. Yet despite representing nearly 5 times the population, this second hump yielded half the deaths of that other mountain.
The other thing you see when you tease apart the rest of the country from those 7 hot spot states is that during lockdowns, cases--of which deaths are unquestionably the most reliable indicator--were climbing steadily throughout the national lockdowns. But they were climbing much more slowly than the "model" "Curve crusher" states. The rest of the U.S. did in fact "flatten the curve," as you can see when you look at the manifest "flatness" of the two-humped curve for the "rest of the U.S" in Figure 1 above. As of now, we are counting 33 weeks of this epidemic. Currently, with each passing week our average daily death toll drops by about 50/day. Thus, last week we were at 750/deaths/day, this week 700. If it continues like this, around January 1st, we will be at around 50 deaths/day, and the epidemic will have been spread out to Dr. Rutherford's preferred 40 weeks.
As for hospital capacity, we have never come close to the 50,000/day in critical cases. The chart below overlays the same two curves from Figure 1 with hospital utilization for each group of states. You can see that at the peak in April, when NY and friends were "crushing" their curves, combined hospitalization for the "curve crushers" was about 38,000 beds. The rest of the country, again, with 5x the population, was using just 15,000 beds, for a total just shy of 60,000 beds. When the rest of the country peaked in mid-July, there were, again, around 60,000 total people hospitalized, and again, about 1/4 in critical care--this amounts to just 8% of all U.S. hospital capacity and 13% of ICU capacity (we have more ICU beds per capita than any country in the world 3x more than France and Italy, and 10x more than China).
Source: COVID Tracking Project
Figure 4 below shows COVID hospitalizations, ICU usage and ventilator utilization, and overall hospital capacity in the U.S.
Source: COVID Tracking Project Analysis and visualization by Emily Burns
Another way to look at this is to look at weekly deaths over time, this is the true measure of the effectiveness, or ineffectiveness of our measures.
Figure 5 above shows weekly deaths for all causes for NY and friends, from 2015 to Oct. 2020. Compare that to Figure 6 below, which is the same analysis, just for the other 44 states. For the early-peaking states, NY and the other Curve Crushers, weekly deaths more than double the weekly death tolls during the height of the 2017-2018 flu season, for every age group, except under 25. Unquestionably, these are states are the epitome of NOT "flattening the curve". If we compare Figure 5 to Figure 6, we see that in Figure 6, which shows the rest of the country, the other 44 states, for every age group except 25-44, the weekly death levels do not exceed the peak levels of the 2017-2018 flu season.
Source for Figures 5 and 6: CDC, weekly deaths by age and jurisdiction, 2015-2020, through October 1st. Visualization, Emily Burns
Now, there are several important things to note. First, for the rest of the country, there are two peaks for all of these age groups, except 25-44 (I will address this group shortly), so this still IS more deadly than the 2017-2018 flu season, which was a bad flu season (61,000 people died). But if the goal was to flatten the curve to ensure hospitals weren't overwhelmed, then this demonstrates that as a country, we have indeed been very successful in doing that--keeping at all times significantly below the weekly deaths for that flu season. One other thing to note, on average, for every hospitalization for flu, 1 in 7 people dies. For COVID, the number remains around 1 in 4. The result of this rather grim statistic is that similar numbers of deaths for COVID yield fewer hospitalizations.
The other way to look at these two figures, is to try and understand the excess deaths represented by each of them. The peak is somewhat indicative, but not incredibly helpful. The following two tables attempt to put the excess death of each group of states into perspective, and the relative lethality of COVID in each scenario.
Table 1: Comparison of Peak Incremental Weekly Deaths by Age Group in Early-Peaking States (NY, NJ, MA, MI, LA, CT), During COVID and the 2017-2018 Flu Season
Table 2: Comparison of Peak Incremental Weekly Deaths by Age Group for All Other States (44), During COVID and the 2017-2018 Flu Season
Analysis by Emily Burns, based on CDC source data above.
In these two tables, we see that not only was the majority of the U.S. was able to avoid overwhelming hospitals, but that the actual increased mortality from disease is reduced by a factor of 2-3 as well in the rest of the U.S. relative to the "curve crushers" . This is borne out further by the fact, noted in Figure 3 above, that the 6-7 hotspot states (depending on if you include Rhode Island) have nearly 3x the number of per capita deaths, at 1265 deaths/million, relative to the rest of the country, which comes in at 485 deaths/million.
Figures 5 and 6 above do bring up some very interesting questions. Most importantly, is the 25-44 age group in figure 6. The fact that this death "peak" does not at all resemble the other age groups' COVID peaks suggest that the cause of death may well be different. This is something that I will explore in a later post. But it is worth pointing out now, given that the figure is present. In fact, much of the "Rest of the U.S." graph looks very different to a typical infectious disease graph, whether it be flu, or COVID, which ought to give pause as to how many of the deaths in these spikes are due to COVID, or due to the lockdowns and other strictures designed to curb COVID.
Now that we have fairly exhaustively looked at the "Curve Crushing" states, and demonstrated that they ought not be models from a "don't kill people with COVID" , and "don't overwhelm the hospitals" perspective, let's look at them from an economic perspective. Let's call it the "Don't kill people's dreams perspective."
The table below shows all 50 states, and D.C., organized from least deadly to most--because reducing mortality must clearly be the primary public health goal of any public health policy worth its salt. But a secondary concern ought to be the impact on unemployment, as over time, that will surely impact public health. In a very real way, unemployment is indicative of how "open" a state is. It has been repeatedly suggested that "opening up" would result in massive, avoidable increases in mortality, and that because of this, we must make the trade off of astronomically high unemployment in order to prevent astronomically high mortality. This chart is an attempt to see if this is born out in the data, if in fact keeping locked down and driving up unemployment has in fact resulted in lower mortality rates. The goal of such a chart is to identify which states' policies might provide models for other states.
Table 3 shows all 50 states, ordered from least deadly to most deadly, on a per capita basis. It also shows the unemployment rate for each state (black indicates +/- 1% of national average of 7.9%, red font indicates higher than that, green lower), the population, the party affiliation of the governor (as a proxy for the likely strength and protraction of lockdowns), and the percent of the population that is white vs. black. The bolded rows indicate those states whose populations are over 3 million, as any state below that would seem to be too small to be a model for other states. Lastly, the color of the row indicates whether a state went for Trump or Clinton in the 2016 presidential election, because this is an indication of the populace's likely compliance with strong lock downs, masks, and social distancing, because for better or for worse, these measures have become politicized. Thus, Massachusetts, with a Republican governor has very high compliance with social distancing measures, and an active social distance Stazi, and as a result, its citizens are probably far more like California than say, Texas or Florida who both have Republican governors and predominantly Republican populaces. That same inversion would apply for Wisconsin and Michigan, both of whose more red populaces have pushed mightily against their Democrat governors' policies.
Table 3: Deaths/Million and Unemployment by U.S. State as of October 15, 2020
Table 3 makes it clear that the biggest driver of low death rates is a small population, and a large state. Equally clear is that the "curve crushers," NY, NJ, CT, MA, RI, LA, and MI, with their extended, draconian lockdowns, did not result in lower death rates. Rather, all but Michigan find themselves in the top-10 of deadliest states. Most also have double-digit unemployment with the exception of CT, LA, and MI. Based on this data, it seems almost impossible to argue from a data-based perspective that long, hard lockdowns are a way to reduce deaths. Couple that with the disastrous long-term consequences of high unemployment and closed schools that attend such policies, and continuing to even discuss such policies seems nearly unconscionable.
But let's look at this data from a few more perspectives, and see where the science might lead us if we take the time to look at the data. As citizens, we have been on the receiving end of a never-ending barrage of exhortations to "socially distance," "mask up," and avoid other people for more than 7 months. As noted above, there is no question that this guidance is now being responded to in a partisan manner. The one positive that comes from that, is that we can use the general political persuasion of the electorate in each state as a proxy for compliance with social distancing. If these social restrictions were indeed associated with significantly better health and/or economic impacts, we ought to be able to see that fairly clearly. The following chart overlays deaths/million (as of mid-October, '20) and unemployment rate with the political leaning of a given state.
Figure 7: Death/million from COVID 19 as of mid-October, and Unemployment rate by state, overlaid onto 2016 election results.
Sources: Election graphic. Unemployment and deaths/million as cited above.
What Figure 7 demonstrates is that being strongly blue favors both staggeringly high unemployment and mortality rates, with a few exceptions. Being strongly red is generally associated with significantly better mortality and unemployment rates. In fact, being red is only associated high unemployment in one state--Pennsylvania--which has a Democratic governor. Being red is only associated with high mortality in Louisiana and Mississippi--but as my Louisiana-born husband will say, Louisiana and Mississippi are 49th and 50th in almost everything, so perhaps this is not a surprise. What ought to be surprising (and concerning) is that a state like Massachusetts, which is top 5 in almost every possible good metric, is in this case significantly worse from both a mortality perspective, and an unemployment perspective, than both these two perennial poor performers.
Sociologists talk a lot about "natural experiments." Our federal system with 50 unique, and largely independent state governments provides a perfect such experiment. In this case more than maybe any other, we ought to be using this to our advantage to save lives, to look across the country and find the states that have best managed this pandemic. But in order to do that, you have to be willing to look at the data--something our media and our public health officials seems determined not to do.
But there's no reason that we can't do such an analysis. Probably the most useful analysis is to compare states with similar sizes, populations, and demographics, but different political leanings to see if being more blue, or more red, i.e. more or less likely to follow social distancing, more ore likely to open the economy, is more or less likely to result in lower mortality or lower unemployment. Let's say, Wisconsin and Minnesota--demographically nearly identical, though Wisconsin has a higher African American population. "Redder" Wisconsin has both a lower mortality rate (243/million) and a lower unemployment rate (6.2%) than Minnesota (383, 7.4%). Colorado and Utah. Ruby red Utah, which never had a formal lockdown has a mortality rate less than half that of Colorado, and an unemployment rate that is 33% lower (155/362, 4.1% to 6.2%). Or, North Carolina and South Carolina. In this case, the mortality rate is almost identical, but the unemployment rate is lower in redder South Carolina. Continuing to pursue this argument, you could pit California vs. Texas. Both are mega states with significant minority populations. Texas has been one of the most open states in the U.S., while California has retained one of the longest and strongest lockdowns, and has had mask mandates in place from the very earliest point of the epidemic. Texas does have a higher mortality rate at the moment, of 579/million compared to California's 414, but given that California has not really opened up, it is unlikely that they will stay there. The cost in an increase in the unemployment rate--which over the long-term will translate to higher mortality, as well as exacerbating the state's already massive inequalities--is unquestionable. At 11.4%, California is nearly double the unemployment rate of Texas, which clocks in at 6.8%.
The point is, this is the kind of analysis we ought to be performing, finding states that have best managed the dance between keeping both mortality and unemployment rates low, then, ideally, looking to those states among the top-performers that have had the least intrusive policies when it comes to restricting civil liberties. Given this approach, it would seem like Utah, with its 155 deaths/million, 4.1% unemployment would be an example of how to manage coronavirus in the U.S., particularly as it has allowed municipalities to function as they see fit, some electing mask mandates, others not, and never having imposed a formal stay-at-home order. Instead, our national media and public health officials continue to focus on the "success" of states like New York and Massachusetts with their 1300-1800/million death rates (i.e. 10x that of Utah), and double-digit unemployment.
Another area where the media and public health officials seem determined to stick their heads in the sand and ignore the data, is on the racial inequality associated with coronavirus outcomes. The prevailing narrative is that more minorities die of COVID-19, ergo, COVID-19 is more deadly for minorities. It is simply assumed that this is due to long-standing health inequalities that result in lower overall health of minorities. However, before one makes that assumption, one ought first answer the question, "Are minorities being infected at higher rates, too?". According to the CDC, the answer is "yes." The following figure is taken from the CDC's Web site.
Figure 7: COVID-19 Cases, Hospitalizations and Deaths by Race/Ethnicity
For every minority group above, you can see that if death rates are higher, case rates are also higher--in some cases significantly higher even than the death rates. In the case of African Americans, their case rates are 2.6x higher, but their death rates are only 2.1x higher. That actually translates to a lower case fatality rate that that observed for whites. The same is true for Hispanics. They have a slightly higher death rate (1.1x), but they are infected at nearly 3x the rate. Let's dig into this a little bit more.
Table 4 shows the aggregated case and death data for all observed cases and recorded deaths with race data. Looking at Table 4 below, it starts to look as though there is a significant under-representation of whites in cases, relative to Blacks and Hispanics. Table 5 shows just exactly how great the disparity is, and what political persuasions exacerbate this trend.
Table 4: COVID-19 Cases & Deaths by Race and by 2016 State Electoral Elections, as of 9/14/20
Source: KFF analysis of data from The COVID Tracking Project and publicly available state websites. Data retrieved on September 14, 2020. Total State Population Distribution by Race/Ethnicity based on KFF analysis of 2018 American Community Survey.
Table 5 below shows that, yes, minorities have higher death rates relative to whites, but as with the CDC data shown in figure 5 above, the cases for each ethnic group are also higher, in some cases, as in the case of Hispanic population, significantly higher.
Table 5: Population-adjusted Multiple of White Infections & Deaths for Differing Races
Source: KFF, as above
But what is particularly interesting about Tables 4 and 5 above, is how much greater the discrepancy between races is in blue states than in red states. Notice that in blue states, cases amongst African Americans are more than 30% higher than in African American's in red states. Amongst Hispanics, the difference is even greater, with Hispanics in blue states 3.4x more likely to be infected than whites, vs. 2.3x in red states. This would seem to lend credence to the idea that protracted lockdowns protect the rich, and infect the poor, leaving minorities and the poor in states pursuing such policies at a greater disadvantage to minorities with more lenient public health policies.
But the true test really is, how deadly is the disease? Table 6 makes it clear that blue states' hard lockdown policies are worse across the board for all races. The average case fatality rate is 3x worse for every single race in blue states than for red states (inline with what we have seen elsewhere in this article). This once again gives the lie to the idea that extremely restrictive lockdown policies provide protection to anyone.
Table 6: Case Fatality Ratio by Race for COVID-19 Cases & Deaths for which Race Data is Available.
Source: KFF, as above
The other thing that you might notice from Table 6 above, is that the CFR for whites and Asians is actually higher than that of both Blacks and Hispanics. This goes against the prevailing narrative being peddled by Public Health Officials. But if you look more closely at those claims about "increased mortality amongst minorities," every single one of them points to the deaths amongst minorities relative to their populations, not relative to their rates of infection. Looking at death rates relative to infection rates shows that minority case fatality rates are not, in fact, higher. It shows rather, that they are lower, in the case of hispanics, substantially lower. Why would that be? Well, the one thing that we know impacts COVID-19 mortality more than anything else is age, and particularly age coupled to multiple serious co-morbidities. If the only Asians and whites who are being infected (and tested) live in nursing homes, then the white and Asian mortality rates are going to be very high. By the same token, if large numbers of younger Hispanic and Black workers are exposed due working in "essential" jobs, then their mortality rates will be lower.
This would mean that lock-down reliant policies for mitigating the spread of COVID-19 are DIRECTLY responsible for increasing the exposure of minorities to COVID-19, and hence driving the unequal mortality rates we have seen in this country, by driving unequal infection rates across races--but probably more importantly, across classes. What's more, our public health officials are aware that this is happening. When you couple these facts with the fact that minorities are twice as likely as whites in the U.S. to live in multi-generational homes, this kind of public "health" policy starts to look more like, well, a genocide. Let's say "no" to genocide.
Now, do I really think that our public health officials are trying to wage a genocide? No I don't. Why, then, are they pursuing policies that are clearly having asymmetrical impacts on minority groups? I believe this is because they have taken up "following the science" as a mantra, not as a practice. Their mantra has become "flatten the curve" and they are just going to keep bludgeoning it until it gets flatter than was ever intended, no matter what the consequences. In truth, it is clear that the goals of our public health officials have morphed from trying to "flatten the curve," to trying to eradicate the disease. In trying to achieve this ever-receding goal, our public health officials appear willing to accept ever higher death tolls amongst minorities, the poor and the elderly, simply as unfortunate collateral damage. It's almost as if they have said that we have to accept this as part of the greater good of eradicating COVID-10. I wonder if the groups who are being sacrificed are OK with it? Or if they might perhaps like a little more evenly shared disease burden?
We must being following the science. Heretofore, this phrase has been used as a mantra to provide cover for our politicians. I am fearful now that it is being used as a shield by our public health officials to spare them from criticism for past efforts.
If, instead of using "Follow the science" as a slogan, mantra, or shield, our politicians and public health officials were to take up this phrase and use it as a tool, we might have some success. The wonderful thing about science, is that it is self-correcting. If you perform an experiment, and it proves your hypothesis wrong, then you adjust your hypothesis, and try another experiment, or ideally, several. We have the benefit of having had these experiments performed for us--all we need do is look to the data. If, on the other hand, you follow the science down the wrong path, and it leads you to death and destruction and you keep going down the same path again, and again, because you are certain you're right, you're not a scientist. If you're doing it because you think it's a reasonable sacrifice to get someone out of office, God help you.
Let's stop holding up the New York massacre as a model. Let's look instead to states that are having success. Let's start looking at the data, let's start following the science down a better path, or better yet, using it as the powerful tool it is to cut a new path. There are many options open to us. Let's stop insisting on the one that has created the greatest man-made public health catastrophe in American history. Let's choose a public health policy that takes account of the public it is meant to support. It's time for American Public Health officials to stop blaming the failures of their policies on the American Public. We are not the problem, your policies are.