Northeast States Should Not be Our Models for Handing Coronavirus: Texas Should.

Updated: Jul 14, 2020

Remember back in March when we set out to flatten the curve of the novel coronavirus? The goal wasn’t to eliminate the disease. The goal was to ensure that hospitals were not overwhelmed as they were in Northern Italy, thereby increasing the number of deaths from the virus due to inability to properly care for COVID-19 patients.

New York, Massachusetts, and New Jersey did nothing like flattening the curve. Sure, they have flat curves now—which undoubtedly has something to do with large numbers of their populations being exposed, no matter what they want to say—but they started off with spikes the likes of which were not seen anywhere else in the world—not even in Italy or Brazil. In fact, Brazil doesn’t get within leagues of New York City. At 70 deaths/million in a single day (598 deaths/8.5 million inhabitants on March 31) New York City, “boasts” the highest recorded number of deaths/million in a single day, other than Bergamo, where the data is not readily available (but which is surely more). In Brazil, this month’s coronavirus whipping boy (after, of course, the U.S.) the deadliest day is only 7 deaths/million inhabitants. Even in Sweden, which famously has had no hard lock down, the deadliest day reached only 11.5 deaths/million—higher than the U.S.’s most deadly day (7/million), but nowhere near the stratospheric lethality observed in NYC (70/million), New Jersey (59/million), New York State (51/million), and Massachusetts, (36.5/million).

In fact, if you count New York City’s probable deaths (which they want you to do), NYC is only a little behind Bergamo in death’s million. However, in Bergamo, 57% of the population has antibodies, compared to NYC’s 19%. If antibodies are indicative of who has had the disease that would mean that for about the same proportion of deaths, NYC logged approximately 1/3 the immunity benefits, meaning that the disease’s progress through NYC was up to 3x more deadly than Bergamo, despite NYC hospitals never actually being overwhelmed, where Bergamo’s unquestionably did.

And yet, now for some reason NY, NJ, and MA are being held up as the “model” states for “flattening the curve”. At the same time we are shaming states like Texas and Utah, whose deaths per million are lower than those of Germany and in the case of Utah, only slightly above Norway. On July 8th, Texas logged its deadliest day, 121 deaths—or 4.1 deaths/day/million inhabitants. On the same day, Massachusetts, at ¼ the size of Texas logged 30 deaths—one of its lowest days—and also 4.1 deaths/day/million. Literally, the exact same lethality, and yet Texas is being held up as failure, while Massachusetts is being held up as a model.

The table below puts into context the relative lethality of various places, looking at raw numbers, but also adjusting those numbers for population.









“Boggles the mind” doesn’t even begin to describe this. Is the lesson here that the rest of the country should also have maximized the number of deaths up-front, so they could claim to have flattened the curve? Was that our actual goal? To kill as many elderly people and minorities as possible, thereby providing a foundation for herd immunity such that the grandparents of the wealthy could go out and get about their business?

Below are the curves of what are currently described as “problem” states, compared to the “model” states. Which of these curves is flattened? Which is a spike? Yes, the deaths in Florida, California and Texas have seen a jump in the last few days, but again, population-adjusted these deaths are still very much in-line with Massachusetts and New Jersey even now.

“Problem” states of California, Florida, Texas, Arizona and Utah.

Versus “Model” states of Massachusetts, New York, and New Jersey. Please note the difference in scale. All of the “Problem” states’ y-axes max out at 150, where New York’s Y-axis tops out at 1250. And of course one must remember the populations. California is 39.4 million people, Texas 29, Florida , New York 19., New Jersey 8.9, Arizona 7.3, Massachusetts 6.9, and Utah, 3.2 million.


In fact, when you adjust for population, the worst days for Texas, California, and Utah are all equivalent to some of the best days for Massachusetts and New Jersey. New York, has indeed “flattened the curve” further, but again, one must at least ask the question if some of that flatness is due to a certain level of immunity within its population.

The chart below puts current and prior deaths into context in terms of population--showing the population-adjusted 7-day average deaths/day/million, to for each of these states.

Data scraped daily from since U.S. Lockdowns began to lift. Complete data set available here:

The data shown above shows the 7-day average daily deaths/million inhabitants. It includes observed data from the beginning of May through July 12th, meaning that this chart does not show the stratospheric lethality observed in NY, NJ, and MA at the beginning of April. One can also see that Massachusetts and New Jersey remain, on a population-adjusted basis, far more lethal than Texas, Utah, California, and the U.S. as a whole. It is true that Arizona has recently slightly outstripped Massachusetts and New Jersey, with an average deaths/day/million of 8 deaths/day/million, compared to Massachusetts’ 3/day/million, and New Jersey’s 4/day/million. Nonetheless, this is still in-line with where Massachusetts and New Jersey were at the beginning of June.

But what about cases you say? Most of the media coverage has been on the massive increases in cases, even while deaths are flat or growing slightly. Let’s take a look at that. Below is a chart that shows the new cases/day/million for each of the same states.

Data scraped daily from since U.S. Lockdowns began to lift. Complete data set available here:

Here surely is room for concern, eh? Arizona has increased its daily cases/million by 10-fold between May and July. And yet, clearly, there is only a moderate increase in deaths/day/million from 4 at the beginning of May to just over 8 now. At the same time, New York, New Jersey, and Massachusetts have each decreased both their daily cases, and their 7-day averages of daily/deaths/million by 10-fold. Why is there a strong correlation in decreases in cases and deaths/million in these northeastern states, but not in these other states? How is it that these other states' observed cases are rocketing past the highest recorded levels of these three ultra-deadly states, but that their deaths remain in-line on a population-adjusted basis with NJ, NY and MA who have "flattened" their curves.

The key may well lie in that adjective, "recorded" when discussing cases. We know that up until early June we had recorded no more than 10% of actual cases. On June 25th, the CDC announced that there had been at least 20 million cases of COVID-19, based on antibody prevalence in blood samples . Because it takes roughly 2 weeks for antibodies to develop, the latest this could have been is June 10th, when there were slightly more than 2 million recorded cases. Thus, we know we had been missing large numbers of cases. The question is, are we still missing large numbers of cases? And is there variability in terms of how many cases each state is missing?

One way to answer that question is to look at the observed mortality rates by state over time.

We cannot look at the raw observed mortality rates of all deaths/all cases for each state since the beginning of the pandemic, as that would unfairly punish MA, NY, and NJ, to whom virtually no testing was available during their epidemic peaks. Furthermore, it doesn’t tell us what’s happening now that economies are a little more open and people who are at lesser risk are getting the virus. Thus, in order to get a sense for how observed mortality rates are changing over time, we’ll take one week of deaths, and divide it by the prior week's of cases. Doing the same thing for every day will give us a rolling morality rate. While the average time between being infected and dying has been shown to be 18 days, testing lags infection significantly. In NYC, an increase or decrease in cases saw a proportional increase or decrease in deaths 7 days later. This seems to hold for most other areas as well. This makes some sense, as deaths are most tightly coupled with those cases needing hospital admissions, and in most cases hospital admissions track very closely with cases—particularly in Texas. Research has demonstrated that the average hospital stay in countries other than China is 5 days, and for ICU stays, 8 days, providing additional rationale for using a 7 day delay to calculate mortality rate.

Description: This chart shows the observed mortality rate for (1 week of deaths in a given state)/(1 week of cases) each state. The goal of this chart it to try and chart the observed mortality rate for over time, linking a specific group of identified cases, to specific group of identified deaths. Because there is significant variability in death reporting on each day of the week, as well as variability in time of infection to time of death, it is necessary to count an entire week of deaths to hope to get anything like a consistent number.

Source: Data scraped daily from since U.S. Lockdowns began to lift. Complete data set available here:

This chart represents the observed mortality rate for each of these states as it changes over time. As can be seen from the chart above, there has been a significant decrease nationally in the observed mortality rate, going from roughly 6% at the end of May, to 1.5% by the middle of July. Individual “problem” states have done even better. Arizona and California both started at over 5% observed mortality rates, and have reduced them to 1.8% and 1.2% respectively. Texas and Utah both started around 2.5% observed mortality, and have pushed those numbers down to 1.2% and 0.8% respectively. While Florida has pushed its observed mortality rate from 6.1% to 0.9%.

This of course must be contrasted with New York, New Jersey, and Massachusetts. New York has also significantly reduced its observed mortality rate. Massachusetts and New Jersey on the other hand, have maintained persistently high observed mortality rates, hovering in the 10% range for both states (save for a jump to 160% for NJ, due certainly to a “catch-up” day of death reporting).

There are several possible reasons why the observed mortality rates are so different. The most cynical explanation would be that because NY, NJ, and MA have chosen to measure their success by low numbers of cases, that there is a subconscious desire not to find cases, and to test less. However, while Massachusetts is testing less, New York and New Jersey are testing more.

Source: Daily scraping worldometers, data here:

Thus, differences in testing can't explain the nearly 10-fold difference in observed mortality rates between Massachusetts, and say, Florida and Utah. Which means we must look to other alternatives. Another alternative is that Florida, Utah, Texas, Arizona and California are all simply better at finding the coronavirus cases. Again, if your metric for success is a low case number rather than a low death rate, it is not inconceivable that this might be the case.

The third possibility is that the medical treatment being received by patients in Utah, Arizona, Texas, California and Florida is better than that in Massachusetts and New Jersey. This possibility seems rather far-fetched, given what we know about the medical industry in Massachusetts.

This leaves one final possibility, which if the true source of this discrepancy, ought to be a source of grave concern to these states. The only other possibility is that in Massachusetts and New Jersey, the disease is still finding its way almost exclusively to the most at-risk--people with co-morbidities and the elderly. This would suggest that despite what are still quite draconian lockdown measures, the disease continues to ravage the most at-risk, while in locales with more moderate approaches the at-risk are shielded, and the disease runs through the population of those least at-risk.

It is most likely a combination of these factors. Whatever the case, it does seem at least a possibility that focusing on reducing cases rather than reducing deaths and mortality rates could be creating perverse incentives for public health officials.

But what of the documented surge in cases in these states? Given the significant increases in cases, aren’t we likely to see a significant surge in deaths? Yes, we will see (and are seeing) a surge in deaths, but it looks nothing like what we saw New York, New Jersey, or Massachusetts. It is worth noting that none of these Northeast states ever ran out of hospital capacity—despite having far, far greater levels of deaths and hospitalizations—so it seems ludicrous to expect that these other states whose cases are now surging would run out of hospital capacity when their surges are so much smaller than those of NY, NJ and MA.

It is also important to bear in mind those persistently falling observed mortality rates, and recognize that they are likely to fall further, given that the current group of infected people is composed of a much larger proportion of younger people—with average new recorded cases dropping to 35 from 65. In Massachusetts, the difference in observed mortality between 30-39 and 60-69 is 30 fold, 0.2% to 6%. And, of course the reality is even starker given the large number of cases that were and are being missed in Massachusetts, most particularly among the younger cohorts. Thus, if this shift in caseload to a younger cohort is in fact accurate, we can expect the mortality rate to continue to drop, and the deaths to climb only modestly.

Based on the currently observed mortality rates in each of these states and the U.S., given the cases logged over the next week, we would expect to see the following changes of 7 day average deaths/day/million population. The grey area indicates where a prediction is being made based on known logged cases, and expected mortality given currently observed mortality rates.

Analysis: Emily Burns. Data:

As can be seen from the chart above, if these observed mortality rates hold, the deaths/day/million for most of these states will not change markedly. Even if each of these states were to double their daily recorded cases from where they are now, it is unlikely their death tolls would even reach where NY, NJ, and MA were at the beginning of May—which is still well off those states’ highs at the middle of April. To approach those levels, they would literally have to increase their case numbers 10-20 fold from these currently elevated positions.

But what about those hospitalizations? Aren’t they spiking in these states? Let’s look into them. At the height of their surge, March 30, New York City recorded 1770 new hospitalizations in a given day and over 6000 cases. A little over 1 week later, on April 7th, they recorded their largest daily death toll—599. Roughly 1/3 of the people hospitalized a week before were dead. These numbers hold throughout NYC’s entire epidemic curve. Roughly 1/3 of people who are hospitalized end up dead a week later. Is this what we are seeing elsewhere in the country? Absolutely not.

Texas doesn’t track day-by-day new hospitalizations, only hospital bed utilization. So let’s take a look at those numbers and try to put them in context.

At the height of NY State's outbreak, on April 9th, there were more than 18,000 simultaneously hospitalized patients--about 60% of them in New York City, as New York City represented about 60% of New York State's outbreak through the end of April. Thus, in New York City at the peak, there would have been approximately 10,800 simultaneous hospitalizations. As of July 9th, 2020 in ALL of Texas, there are 9600 hospitalized patients (as of July 11th, there are 10,083). This is for ALL of Texas. Texas is 29 million people, vs. NYC’s 8.5 million people, i.e., Texas’s population is 3.4 times larger than NYC’s. For Texas to be even close to a NYC-level epidemic, it would need to have 10,800 * 3.4, or 36,000 people hospitalized. Nearly 4 times the levels we are currently seeing in Texas. To try and put this into even more context, both the Dallas and Houston Metro areas are roughly 8 million people, so each about the size of NYC. Dallas currently has 1800 active hospitalization, while Houston is at 2700. Again, compare that to similarly-sized New York City which at its peak had more 10,000 simultaneous hospitalizations.

As can be seen in the figure below, Houston's hospitals are only at half capacity, with COVID patients utilizing 28% of in-use beds--14% overall. For ICU beds, COVID patients, as elsewhere in Texas, make up a little less than 50% of ICU beds in use.


The chart below shows ICU capacity. One can see that on June 9th, before the current case surge, ICU capacity, with 15% COVID patients was already running at 89%--because that's how hospitals make money, by running as close to capacity as possible. Now the ICU is at 96% utilization. Not exactly the story that we are getting in the national news.


The same holds true for the rate of increase. In the same New York Times article cited above, it is noted that a sign of things “cooling down” in New York was when active hospitalizations increased by only 200—the lowest daily increase since the start of the pandemic. Again, adjusted for population, that number would need to be 1.5 times that size, or 300 to reach the same level in Texas. Since the end of June, Texas has in fact had many increases in this range, but rather than being their smallest increases, they represent their larges increases—once again proving that the worst days in Texas in the pandemic, are better than the best days in NYC.

Source: COVID-19 Hospitalizations by TSA

Now let’s look a little at utilization. Texas has a little over 50,000 staffed beds. Currently, COVID-19 patients are taking up less 20% of those, just 9600. There remain more than 10,000 general beds, and more than 1000 ICU beds available—15% of their capacity. Standard open ICU capacity is 30%-- so using up half of the available ICU capacity doesn’t seem bad in the teeth of a pandemic. New York City has approximately 23,000 total hospital beds, which means that in the teeth of their surge, COVID-19 patients were utilizing nearly 47% of all available hospital capacity.

The graph below shows the change in deaths, cases and hospitalizations in Texas over time. Several things are worth noting here. First, while observed cases are clearly tightly tied to hospitalizations, they are not tightly linked with deaths—which is fantastic.

Now, let’s try and put this into context relative to NYC. In NYC, at the height of their outbreak, we saw the following pattern: ¼ of identified cases were hospitalized that day. 1 week after, 1/3 of those hospitalized died. Working back from Texas’s case numbers, if the same pattern held true for the cases observed on 7/1 (8300), we would expect 680 deaths today. Yesterday we saw 120, today, it looks like we’ll be back below 100. Applying the same to Florida, back on July 4th, we saw 11,500 cases. If the same pattern as NYC were holding for Florida, then we would expect 11,500/4/3 or 958 deaths one week later, on July 11th. What did we actually see? 95 deaths.

Let's talk a little more about Florida, given that yesterday Florida had over 15,000 cases. On April 6th, New York City logged its largest single day case count--6378. That same day, approximately 27%, or 1700+ of those cases were hospitalized (also the highest number). 1 week later, 570 people died--an observed fatality rate of nearly 10%. Again, July 4th was Florida's highest day of reported cases prior to yesterday. On that day, Florida logged 11,500 cases. That same day, Florida logged its highest number of new hospitalizations--341, or roughly 3% of identified cases. 1 week later, Florida logged 95 deaths, for an observed case fatality rate of approximately 0.8%. This means that in NYC, the hospitalization rate for observed cases was 9 times higher than that seen in Florida, and the observed mortality rate around 12 times higher. By the same token, when we look at hospitalizations, Florida's highest daily new hospitalizations was 435, on July 7th--compare this to New York City's 1700+, and remember that at at 21.4 million people, Florida is 2.5 times the size of New York City. An equivalent number of hospitalizations in NYC would be 170--numbers that New York City was seeing at the end of May, when again, they were being vaunted for "flattening the curve".

I point all of this out, not to minimize the public health challenges of coronavirus in these surging states. There is no question that maintaining normalcy is not something you “happen” into in this case. But up until this surge in cases and the attendant media attention, I think it is very hard to argue that Texas, Arizona, Utah, and Florida haven’t handled the coronavirus pandemic extremely well—far, far better than NY, MA and NJ.

There is no doubt that given the current surge of cases, some of these states need to do some level re-trenching. But the level of re-trenching that is being foisted on them is excessive. Why not instead of broad shutdowns and masks, prohibiting counter service at bars as Sweden has done? How about being honest about whose really at risk, and giving some explicit guidance to young people about their need to avoid elders if they are going out in groups? Or even better, undertake strategies to help families living in multi-generational homes protect themselves the way that many other wealthier groups are doing? Texas is only close to hospital capacity because they were basically trying to resume a pre-pandemic level of normalcy, not just in life, but in their hospitals, allowing for routine and elective care. That is an admirable goal, and there is no question that we need to make sure that as we wade through the rest of this pandemic, we ensure hospitals are open and available to people for “elective” procedures—many of which are life-saving, if not in the immediate term.

Texas, in particular, I believe is proving itself to be a model for the rest of the United States. Texas looks an awful lot like the U.S. as a whole with a large and extremely diverse population, several mega-cities, sprawling suburbs, and many small towns. And with this make-up, and a very light touch, Texas was managing to keep the pandemic in check, doing their best (and a far better job than NY, MA, and NJ) protecting their elderly and at-risk, yet allowing those who re less at-risk to make their own decisions and live their lives. All this with bars open, and no mask requirements, and hospitals on a non-pandemic footing. My sincere hope is that Texas doesn't pause for too long and will resume their bold course, that Texas, Florida, Utah and Arizona do not give up on this complicated dance where they are learning to pair living life with the need to preserve lives.

260 views0 comments