Updated: Nov 17, 2020
The tragedy of the northeast continues to obscure our perceptions of the pandemic, and more bizarrely, to be treated as a playbook for how to manage it. Nationally, however, there is reason both for real optimism, as well as real concern (and it's not the "third wave"). The source for optimism is that 25 states have not exceeded the number of weekly deaths they experienced during the 2017-2018 flu--which was a very challenging one for U.S. hospitals. This is despite these states being in general very "open." Many more would have remained below this threshold, had they not seen significant increases in excess deaths due to their COVID responses.
This is where we find the source for concern. While many states have not experienced significant excess mortality due to coronavirus, they HAVE experienced significant excess mortality due to lockdowns and our responses to coronavirus. Specifically, on average, 42% of the excess deaths in these states are NOT due to COVID. This means for every person who dies of COVID, roughly one person who wouldn't die otherwise, dies for some other reason. Put another way, nationally, if (as of writing) our deaths/million from COVID were 701, our non-COVID excess deaths from COVID response were 270/million. These deaths are in younger, healthier populations, populations with decades to live, compared to the average COVID death, which is identical to the average age of death in the U.S. (78). Not surprisingly the hard lockdown, early-peaking states of NY, NJ, CT, MA, RI MI and LA have worse numbers, with an aggregate showing of 360 deaths/million for NON-COVID/COVID response deaths (20,000 deaths). This is in addition to the 1406 COVID deaths/million in these states (78,063). This is harm.
In some states, such as Oregon, Utah, Maine, Alaska, and Vermont, the ratio of non-COVID deaths/COVID excess deaths is much higher. Again, what this means is that in these states, significantly more people are dying due to responses to COVID than to COVID itself. In some cases like in Vermont and Oregon, a significant preponderance of the excess deaths are non-COVID deaths. This is reason for grave concern. The average age of death for COVID is 78--identical to the average age of death in the U.S. Thus, every COVID death that is prevented does not actually translate into increased life-years, on a national basis. However, the age of these lockdown/COVID response-related deaths is much younger--truly deaths that shouldn't happen.
These trends hold nationally, as well. If you remove the 7 early-peaking states of NY, NJ, CT, MA, RI, MI and LA, the U.S. has never exceeded the weekly deaths we observed during the 2017-2018 flu season. Nationally, we are also seeing one preventable, non-COVID excess death for every 2 COVID deaths. You can see that in figure 1 below. If we did not have the non-COVID excess deaths, our excess death for the other 43 states would be below the level of a standard flu season. The preferred approach of our public health officials is doing harm.
It is also important to note that during ANY flu season at peak, we see nearly 1200+ flu deaths/day (the periodicity of deaths above summer lows is almost exclusively due to flu). During the 2017-18 flu season, an additional 1400/day was added to the peak--meaning that the peak of that flu season would have seen roughly 2600 deaths/day nationally, if we had chosen to count. This is important perspective as our deaths slowly rise again--particularly as they are rising during the beginning of flu season, and in many places in step with what we would expect for deaths during this season. For better or worse, COVID seems to have beaten out the flu this year for which disease will play winter's reaper.
Source: CDC weekly deaths, relative to COVID-19, data extracted on 11/1/20.
We are currently seeing a surge in hospitalizations, which will turn into a modest surge in deaths--my estimates are a 7-day average peak of 1200/day (i.e. in line with flu deaths for a normal season), and possibly as high as 1500, though I think that is unlikely. I feel confident saying that, nationally, we will remain below the threshold of weekly deaths during the 2017-2018 flu season for the remainder of the pandemic, regardless of when and whether a vaccine is released--as long as we don't lock down again. I do not believe we will see a rise like that being seen in Europe, because we have allowed the virus to circulate much more broadly, and in so doing, have been building immunity--both antibody-mediated and T-cell mediated--which removes available virus hosts. I have detailed this ad nauseum here. Effectively, we have broken down COVID into three really bad flu seasons here in the U.S.. Only the first one, driven almost entirely by the northeast, was truly aberrant.
The reason that I feel confident saying all of this, is that if you adjust for testing, cases track exactly with hospitalizations (as you would expect), and hospitalizations are not rising at the rate we saw in the summer, and certainly not like what we saw in the spring. Figure two shows what past case numbers would look like had we been testing at this level throughout the pandemic. It approaches the problem from two different directions--and comes to nearly identical conclusions. The first approach, is to assume that we were completing the same number of tests we are performing now, throughout the epidemic. The other, assumes a consistent CFR (case fatality rate--the (number of deaths)/(number of observed cases)),which is 0.8%.
Figure 2: Normalizing Observed Cases using observed 2-week offset CFR, and normalizing for testing, with reference data of 11/20/2020
Source: The COVID tracking project. Analysis, Emily Burns. CFR calculating by averaging deaths for 7 days, divided by the cases for 7 days, 2 weeks earlier. Testing normalized cases are normalized based on testing as of 11/10/20. equation: (observed cases on date)/(tests performed on date/tests performed on reference date).
Also, there is no question that both approaches are underestimating infections by a factor of no less than 2, and by as much as 4. No less than two, because multiple conservative studies released in early summer showed at the time, that the U.S. had at least 30 million cases. One strongly suggested that there were as many as 9 million cases in March alone (the graph above shows 2.5 & 5.5 million cases in March, giving credence to this idea). We have clearly not made any significant in-roads since then in "slowing the spread," and the nominal number of cases has gone up by 4. I say as much as 4, because the real IFR (infection fatality ratio) is likely, according to the WHO, closer to 0.23% based on multiple seroprevalence studies--many of which were carried out in just these U.S. states. Early-peaking U.S. states appear to have IFR's more than double this, which likely explains some of the divergence with cases and hospitalizations at the beginning of the epidemic (there are other possible reasons, too, but I will skip those for now).
That there are so many states which have never exceeded weekly deaths observed during the 2017-2018 flu season (which killed roughly 60,000 people), tells us that the U.S. writ large, has succeeded beyond the wildest dreams of epidemiologists in flattening the curve. Yes, we are currently at 60,000 hospitalizations, the same levels we saw in spring and summer, but that is still only 8% of U.S. hospital capacity (we have 900K beds, with usually around 300K available--and the most ICU beds of any country in the world, in many cases by multiples). Of note, because COVID IS more deadly than flu, the same number of deaths translates into fewer hospitalizations. During the 2017-2018 flu, the ratio of hospitalizations to deaths was 7:1. For COVID, it ranges from 3:1 (NYC--ventilation disaster, i.e. too many used too aggressively), to 6:1 (Utah). The elevated death levels are no doubt extremely taxing to healthcare professionals--particularly given how sustained it has been. But the fact that it has predominantly come in the summer should actually help tremendously to spread the burden, given the seasonally lower level of hospitalizations.
The 2017-2018 flu significantly strained many hospital systems, but hospital system strain was a given when this arrived. The initial goal was to flatten the curve so that hospitals were not overwhelmed. If even NYC was not overwhelmed, which Governor Cuomo claims to be the case (and which I don't believe for a moment), nowhere in the U.S. will be overwhelmed during the remainder of the pandemic, because no state has come anywhere near that level of hospitalizations and deaths thus far--and, no, we are not at the beginning, we are quite near the end--check out Sweden to see just how fast this can be over if we just let it. The slight increase in deaths in Sweden now shows what this looks like when it is endemic--something we are almost certainly seeing now in many northeast states, too. Note how similar these curves are, and the note same modest rises in deaths now in Sweden, New York, and New Jersey. Do also note that Sweden's population-adjusted death rate is less than 1/3 of theirs, and that Sweden's daily deaths now are population-adjusted lower than both of these states'--it just doesn't look like it because so many more people died in NY and NJ. Sweden also has the lowest mask compliance of any country in the world.
Ever since we abandoned the idea of flattening the curve, and instead moved to eradicating the disease, we have been shamed into accepting the following dogma: The young must sacrifice the "comforts" of socialization, and the imperative for income, in order that grandma might live. Now, nowhere in that calculus was the idea that the young would be expected to lay their lives on the line. The suggestion was that by taking "small" measures such as not interacting socially, not working, wearing masks, we could save the extremely elderly. The trade off was never supposed to be, "and for every two 78-year-olds we save, one person with decades left to live will die from despair, or from lack of medical treatment or diagnosis." If 230,000 deaths of people with an average age of 78 is horrifying, 100,000 people with decades to live ought to be 10-50 times more so (because that's the difference in life years lost). But just like we do really care about the opioid epidemic, we don't concern ourselves with the 100,000 people who have died, and continue to die, not from COVID, but due to the policies created to slow its spread.
To put this into greater relief, if were were to count opioid + COVID-response excess deaths like we do COVID deaths, we'd be seeing roughly 320 per day--1/3 what we are seeing now due to COVID, but of a very different demographic. In 2018 the opioid overdose epidemic took 127 lives per day, the average age of whom was 30 years old, making for an estimated national loss of life years of 3,840,000 years, compared to COVID's, literally zero, because the average age of death for COVID is the same as the average age of death in the U.S.. What if we had taken the estimated 16 trillion we have thus far spent on COVID and divided it between the two epidemics? Assuming there are roughly 100 million people with even a slightly elevated risk of death due to COVID, we spend 8 trillion, buy each of them a 50K apartment, and pay for 30K of home health services to keep them out of harm's way for a year (if they chose). We could then take the remaining 8 trillion, and spend it on the 10 million people who are addicted to opioids, which would literally allow us to spend $800K per person.
Arguing against lockdowns has been cast as selfish, because the "costs" were not in lives, but less immediate. Unfortunately, it turns out that the cost IS lives. And this is only the first batch. The excess deaths we will see over the next decade due to lack of medical attention, increased obesity, and despair will likely make these 100K excess non-COVID deaths a drop in the bucket. That doesn't even begin take into consideration the loss of a year's worth of schooling, which will be most acutely felt by the bottom 50%, who are also the most likely to be among the 60-100 million people who have gotten the virus thus far, thanks to lockdown policies that leave the poorest exposed and allow the better off to stay out of harm's way. Remind me again, which was the selfish policy?
But I digress, let's return to the state-by-state analysis of excess deaths in 2020, due to COVID and other causes. As I mentioned, the majority of states have never exceeded the highs for weekly deaths experienced during the 2017-2018 flu season. There are many more for whom that was not an extraordinary flu season, and who, thus, have exceeded weekly deaths, but by only 20-25%. This is about the same amount that states hit hard by that flu season saw their weekly deaths increase in 2018. In fact, the ONLY states that have seen truly aberrant increases in weekly deaths were in the northeast, Mid-Atlantic, Louisiana and Michigan. What these states have in common is not that they were hit earliest (Washington and California were both hit earlier), nor is it being extremely population dense--Massachusetts' cities are not significantly more densely-populated than those of California. Nor is that they have larger minority populations than other states. The southern states have far higher minority populations, and have done a far better job keeping deaths closer to this threshold. No, what they have in common, was prioritizing hospital beds above lives. In many states, including Massachusetts, New York, New Jersey, Michigan, and others, governors made it illegal for nursing homes to turn away COVID-positive patients. In other states, it may not have been explicitly stated, but doctors and hospitals were fearful of running out of capacity (which never happened), and refused to admit COVID-positive nursing home patients, thus forcing them back into their assisted living facilities. These policies, both explicit and implicit came directly from the "flatten the curve" mantra which prioritized hospital beds and lost sight of the people who would occupy them. It proved disastrous.
Now we are seeing another disaster unfold, as the "crush the curve" mantra is killing tens of thousands of people in support of stopping an unstoppable disease. That the only states which have not seen any real resurgence in deaths are the same states that saw early spikes ought to at least give the suggestion that this is due to population immunity--but that is consistently ignored, even though these are literally the ONLY states in the U.S. that have managed to keep deaths down, despite many states with far stricter regulations and far higher compliance with various non-pharmaceutical interventions (like masks).That the people whose lives are being sacrificed to "crush the curve" are also the people who would be least likely to get sick--in fact 55% of the U.S. is at significantly higher (3-10x) risk of dying from accidental death than they would be of dying from COVID, if they contracted it--only adds insult to injury. That their immunity would help to protect the most vulnerable makes ignoring this approach seem nearly criminal. And please don't get me started on "but they might not be immune!" This disease behaves like others. There is no meaningful evidence to suggest otherwise. If there were, it would make the quest for a vaccine a fool's errand (I will do a post on that soon enough).
Without further ado: Weekly deaths/million by state, 2017-2020, including all deaths, and non-COVID deaths for 2020. All graphs are on the same scale, to allow for easy comparison across states, the only exceptions are NY and NJ, both of which needed expanded scales. I have places the graph of Massachusetts on every graph as a point of reference. Massachusetts has been hailed as a "model" for handling the disease. That very few states have come close to it either in terms of cumulative deaths/million, or peak deaths, argues strongly against using Massachusetts as a model. But it can perhaps act as a boundary line, above which states ought to try not to go, if things get out of control. Massachusetts did not see the kind of outbreak New York saw, but it was significant and protracted, resulting in a death rate twice that of the nation at large. Despite this, hospitals in Massachusetts were not overwhelmed. Thus, states trying to manage the balance between preserving the lives and well being of the young and the old, might be able to use Massachusetts as an acceptable worst-case scenario to benchmark themselves against. This analysis is particularly useful in terms of being able to make this comparison for your state at a glance.
All data here was extracted from the CDC on November 1st. Some states have had increases since that time, but using these graphs, and comparing your state's death rates to death rates earlier in the epidemic should provide residents of those states with context to help understand the reality of their current situation. Or, if you want to feel really good about your state and your governor, just look at Massachusetts or New York.
A few other things to note about the graphs that follow. You will notice that some states see higher or lower weekly deaths. Why is this? Doesn't everyone die? Yes, they do. However, populations with healthier people who live longer will result in fewer deaths each year. Further, people don't always die where they live.
Lastly, why did I choose the 2017-18 flu? I chose this as a benchmark. We have done zero benchmarking since the start of this pandemic. The result is that numbers are reported out of context. Weekly deaths during the 2017-18 flu season provide meaningful context. Hospitals were strained but not broken during that flu. At the outset of this pandemic, we were told that it would totally overwhelm our health system. That hasn't happened. We need to understand why, and we need benchmarks help us move forward.
Alabama's weekly death tolls during COVID only exceeded the 2017-2018 flu season by 10%, and that very briefly. Additionally, had Alabama not experienced a large number of excess non-COVID deaths, that number would have remained below the threshold of the 2017-18 flu season. Since this data was extracted, Alabama's 7-day average for daily deaths has increased from 11 to 19. At peak, it was 29, so this should still keep Alabama well below the levels of that challenging season. In fact, going up now may reflect endemic rises of the disease as COVID plays this year's winter reaper in place of the flu.
The 2017-18 flu season did no impact Alaska differently than other flu seasons. Despite this low threshold, COVID deaths in Alaska never exceeded any flu season. Of greater concern, Alaska has more non-COVID deaths than COVID deaths by far--excess COVID deaths don't even register. In fact, while there are deaths from COVID, they do not show up as excess deaths, which means that these are people who would have succumbed to the ailments which predisposed them to death by COVID in any case. Alaska's death toll continues to climb in a slow, steady, largely linear fashion.
Arizona exceeded 2017-18 peak flu deaths by 30%. If however, it had not also incurred a significant amount of excess non-COVID deaths--56% of all excess deaths in fact--Arizona would only have exceeded this by 5%. Deaths have begun to rise slightly since I extracted this data, but are still well below the levels of the 2017-18 flu season.
Arkansas has kept deaths below the levels of the 2017-18 flu season, this despite 45% or nearly 1 in 2 excess deaths being due to non-COVID causes. If these non-COVID deaths had been avoided, Arkansas would have kept weekly death levels below that of a standard flu season. Currently, Arkansas' COVID deaths are leveling off, though as elsewhere in the country, they will likely rise with the endemic winter rise over the next few weeks.
California has kept deaths substantially below the 2017-2018 flu season. Had it not also incurred 40% of its total excess deaths in non-COVID deaths--nearly 11,000 people--it would not have breached regular flu levels. California has now returned to baseline death levels, and COVID deaths are still declining steadily, though cases have been rising for a month. Should these case rises translate to increases in deaths, California will have ample capacity to manage any surge. In my estimate, California will be part of the fourth wave, made up of hard-lockdown states on the west coast and far northeast coast. Or perhaps they will get the vaccine first.
Colorado early exceeded 2017-2018 flu levels by 25%. However, there are two caveats that need to be added here. First, the 2017-18 flu did not impact Colorado any differently than any other flu. States which were moderately or severely hit by that flu saw increases in weekly death rates of 25%. Beyond this, Colorado is another that has had an epidemic of non-COVID excess deaths which is directly in proportion to its COVID deaths--more than 200 people. That most of these non-COVID deaths occurred at a time distinct from Colorado's big COVID wave means that these are almost certainly 100% lockdown related. Colorado's case rate has been rising steadily for a month, with only the smallest increase in deaths. Nonetheless, Colorado's persistent restrictions make me think Colorado will be part of the lockdown-state 4th wave, even though it is a few states in from the coast.
Connecticut is one of the few states whose initial peak was higher than that of Massachusetts--72% above the peak weekly deaths it experienced during the 2017-18 flu. Since then deaths have mostly returned to baseline level. In fact, nearly ALL of the excess death that has been observed since Connecticut's initial peak has been due to non-COVID deaths. While Connecticut has seen some slight rise in COVID cases, like others of the early-peaking northeast states, these look much more like endemic level case circulation. In theory, Connecticut has as many cases now as it had in April, but the deaths tell the real story. Like most other states, Connecticut is testing 10x more than it was in April. Show me 10x cases April, then you'll have a surge to be concerned about.
D.C. like others in the northeast and mid-Atlantic saw huge initial death spikes, with more than 55% higher weekly deaths than it saw at the peak of the 2017-18 flu season--which was not trivial in D.C. Since that initial peak, deaths have leveled off, and but for non-COVID excess deaths, deaths would be at baseline levels. However, high levels of non-COVID deaths due to lockdowns keep death rates relatively high. Non-COVID deaths make up 37% of D.C.'s excess deaths in 2020 to-date.
Delaware saw a peak of 36% above peak weekly deaths during 2017-18 flu season. However, this was significantly exacerbated by non-COVID deaths, which contributed 1/3 of those deaths. Since the end of their initial surge, Delaware has seen virtually no COVID deaths, but persistently elevated non-COVID lockdown deaths.
At it's peak, Florida saw 23% more weekly deaths than during the peak of the 2017-18 flu season. If excess non-COVID, COVID-response deaths were not so high, that number would have been reduced to only 13%. Florida spent 9 weeks above that threshold--if lockdown deaths had been avoided, it would have spent only 6 weeks there. Florida is slightly above the national average in terms of non-COVID excess deaths, accounting for 37% of all excess deaths.
On September 25th, Governor DeSantis removed all statewide COVID restrictions (week 36 in the chart above). At that time, weekly excess COVID deaths were declining, and have since continued that decline. Both non-COVID and COVID excess deaths are now inline with baseline deaths for this period. National flu & ILI levels are virtually non-existent relative to seasonal expectations. If COVID acts as this year's winter reaper in lieu of flu, Florida's COVID daily deaths could gradually increase from the 385/week (55/day) now, to as much as 850/week (122/day) over the next 8 weeks, and still not exceed expected baseline deaths. It would take another 67/day to reach those same 2017-2018 peak flu levels. If flu levels do not remain suppressed, and rise along with COVID levels, things will look much worse.
Georgia has kept deaths roughly in-line with 2017-18 flu deaths, edging up to 10% more for a few months. If however, Georgia had been able to reduce its large number of non-COVID excess deaths, it would have stayed below that threshold. As it is, Georgia counts 4600+ people excess non-COVID dead. Georgia is now in spitting distance of baseline deaths. COVID deaths are slowly rising, and cases are nominally inline with July levels, but given that testing is more than double, we can assume that they are less than half--which is where deaths are now, too.
Despite Hawaii not experiencing a significant 2017-18 flu season, weekly deaths have remained below flu levels. In fact, Hawaii's deaths don't even exceed the flu deaths they experienced in early 2020. Hawaii's current death are inline with baseline deaths for this time of year--i.e. substantially below weekly deaths seen seen during a standard flu season.
Idaho briefly exceeded weekly deaths seen during the 2017-18 flu season, reaching levels 10% above those levels. However, this bears three caveats. First, the 2017-18 flu season was no different than any other flu season for Idaho. Second, Idahoans are not particularly hard hit by any flu season. Third, had Idaho not had high excess death due to non-COVID deaths--43% of all excess deaths--it would not have breached even standard flu levels. Since this data was extracted for analysis, COVID deaths have returned to slightly above the summer levels. However, since excess deaths due to non-COVID causes seem to have tapered, Idaho may still be able to remain below flu levels of weekly deaths. If not, it will be unlikely to breach the 25% increase in weekly deaths that states hit hard by that flu saw. Cases in Idaho are nominally 2x what they were in the summer, but testing has also more than doubled. That deaths are in-line with summer levels suggest that much of the increase over summer levels in observed cases is due to testing.
Illinois only exceeded 2017-18 flu season peak weekly deaths by 11%. Had a significant portion of Illinois' non-COVID deaths--which make up 37% of all excess deaths--not been logged during the same season, they would have only barely breached that threshold. At the time when this data was extracted, Illinois' 7-day average for daily deaths 43. As of yesterday, they were 69. This increase in deaths should still result in average weekly deaths staying below the 2017-18 flu season threshold. Today, Illinois reported 153 deaths, but no doubt this will be one of the "spike" days. During its May peak, Illinois saw a peak 7-day average of 120 deaths/day, with an individual high date of 191 deaths reported. Nonetheless, this only brought it to just above peak flu levels. Currently, the average is just over half that at 69. Cases appear to be far higher, but like so many other places, the increase in testing between now and May will be between 5 and 10x, explaining why significantly higher levels of "cases" show significantly lower levels of deaths. If Illinois can keep its non-COVID deaths down, it may be able to keep deaths below 2017-18 flu levels. The governor's penchant for lockdowns seems unlikely to engender this result.
Indiana has been able to keep deaths below the levels of the 2017-18 flu season, despite only experiencing a minor increase in weekly deaths over prior flu seasons. Currently, Indiana's deaths are increasing, at 22 as of extraction, up to 42 at the time of writing, but still below the peak level of 62 that was seen in the spring. As elsewhere, Indiana should be able to keep deaths below 2017-18 flu levels if it can avoid allowing non-COVID excess deaths to increase. Currently, 35% of Indiana's excess deaths are non-COVID deaths.
The 2017-18 flu was only slightly worse for Iowa than a normal flu. At the time this data was extracted, Iowa was just below its 2017-18 flu levels. Since that time, it has increased from an average of 15 deaths/day to 16, thus, likely slightly breaching 2017-18 flu levels. Cases are nominally 10x more than they were during Iowa's May peak, but deaths being inline suggests that once again, differential testing is obscuring the reality.
Up until recently, Kansas had never seen any real spike in COVID deaths. Now, deaths are just barely reaching the levels week during a normal flu season--the 2017-18 flu was no different for Kansans than any other flu season. At the time this data was extracted, average daily deaths were 16, since then, they have increased to 19. Again, while this does push Kansas over standard flu levels, it does not push it over what a severe flu is--typically 25% increased at peak. Additionally, 37% of Kansas' deaths have been non-COVID deaths. Reducing these would help to keep death levels more in-line with peak flu season weekly deaths.
Kentucky's deaths have stayed substantially below peak weekly deaths of their 2017-18 flu season. Had Kentucky not experienced very high level of non-COVID related deaths, it would be challenging even to distinguish 2020 from other years. COVID deaths have been doing a steady march upwards, which, as COVID takes over for flu as winter's grim reaper, will mean that within a few weeks deaths may well be in-line with base-line levels, despite COVID deaths themselves being somewhat elevated above where they stand now. The trick, for Kentucky, as so many states will be to reduce the number of non-COVID deaths.
Louisiana was one of the hard hit early states. Still, "thanks" to a particularly severe 2017-18 flu season, deaths during their COVID peak were only 30% above peak levels for that flu season. If non-COVID excess deaths had been eliminated, the increase would only have been 15% above. Since the initial peak and hard lockdown, Louisiana saw a resurgence, though this only only briefly and slightly breached 2017-2018 flu levels. Additionally, had Louisiana not seen a high number of non-COVID excess deaths, deaths would have been below those levels. For the last 7 weeks, deaths have been flat, not surprising, given how hard Louisiana was hit initially. Given that baseline deaths are now rising with the season, flat COVID deaths should soon mean that Louisiana will soon be returning to baseline deaths--again, even though there will still be some number of endemic COVID deaths as COVID plays the grim reaper for the 2019-2020 flu season.
Maine has not exceeded weekly deaths seen in any flu season. However, Maine's ultra low COVID numbers hide a disturbing increase in non-COVID deaths. Maine has one of the highest proportions of excess deaths due to non-COVID causes--63%. Fully twice as many people have died in Maine from COVID responses, as from COVID itself. Maine is also on my list of 4th wavers, as hard lockdown states that, like much of Europe in the summer had nearly fully suppressed the virus. Again, perhaps they will indeed be saved by the vaccine, perhaps not. But a 2:1 ration of non-COVID excess deaths to COVID deaths is concerning regardless.
Maryland is another of the first-wavers. Like most first-wavers, it was hit hard. However, despite this, deaths only exceeded flu deaths by 25%. The 2017-18 flu did not hit Maryland any harder than a normal flu, making this even more impressive. Had Maryland not seen a large number of non-COVID excess deaths, weekly deaths would only have been around 12% above standard flu levels. Since its initial peak, Maryland's excess deaths have been split evenly between COVID and COVID-response deaths. Deaths have remained flat since the beginning of July, despite two subsequent case "waves," which failed to have any impact on deaths. The current wave of cases is roughly the same size as that seen in July. Given that testing is surely multiple times higher, this wave, too will likely result in no noticeable increase in deaths. Maryland, like many other states looks slated to be at baseline deaths within a few weeks.
Massachusetts, like the tri-state area stands as a cautionary tale. Peak deaths exceeded peak deaths during the 2017-2018 flu by more than 75%. Since its initial peak, deaths have remained inline with baseline deaths. COVID deaths are inching back up, but, here again, this looks to be COVID playing the grim reaper this year instead of flu, as baseline deaths are also inching up, resulting in no variance from prior years. Like most other hard hit northeast states, governor Baker is insistent on taking credit for the flat line, and ignoring the spike. Massachusetts has recently re-enacted curfews. Cases are up, in theory to the same levels that they were in April, but testing in Massachusetts has increased 6-fold since August alone. Cases are up 12x. Deaths? They are flat. For months, Massachusetts had been doing one of the worst jobs in the nation at finding cases. The same level of cases in Massachusetts, would yield 10x the number of deaths of say, a Utah, or and Idaho. Now it seems that Massachusetts is catching up to the testing that other states are doing. Unfortunately for its residents (I am one), this has been used to further curtail our lives, livelihoods and liberties, to provide fear and fodder for educators so that our children might learn. Someday soon, we hope that governor Baker (or his masters, Dr. Scott Gottlieb, and Dr. Ashish Jha) will learn about T-cell cross reactivity, and realize just because he massacred our elderly, does not mean the rest of us must be sacrificed to hide the horror of his misstep. (Also, I would not be nearly so mad if he would just acknowledge he made a mistake, and remove the limits on household mixing, outdoor masking, and masking of children).
Just to make the point of how little impact governor Baker's actions have likely had--notice the similarity in the curves. Sweden never locked down, Massachusetts has never unlocked. Sweden has the lowest mask compliance in the world, Massachusetts has to have the highest, and put it in place on May 5th (way on the downside of their curve). Sweden is 1.5x the population of MA. Note how much lower the peak deaths are in Sweden. Note that Sweden, having a slight uptick in deaths now, still has half the daily deaths (population-adjusted) that Massachusetts has had since July. Massachusetts deaths/million is nearly 3x Sweden's. Governor Baker's genius, or just herd immunity built on the backs of the most vulnerable, and most likely to die?
Michigan saw a 50% increase in weekly deaths over the peak of the 2017-18 flu season. Had it not also had a large number of non-COVID excess deaths, it might have been able to cut that to 25%. Since the end of June, Michigan has had only slight increases in COVID deaths--increases that may well be linked to COVID's role this year as the grim reaper, rather than flu. What is deeply disturbing is that Michigan has ever-growing and extremely meaningful number of non-COVID excess deaths, deaths which are almost certainly linked to the iron fist governor Whitmer has brought down on her constituents. 48% of Michigan's excess deaths are non-COVID deaths--1 in 2, or 6136 people. Were it not for Michigan's non-COVID deaths which currently dwarf its COVID excess deaths, Michigan's death rate now would be indistinguishable from other years.
Minnesota had kept deaths below levels of any flu season (2017-2018 was not remarkable in Minnesota). Further, they had managed to keep non-COVID excess deaths to a minimum, only 24%. Even when I downloaded this, I felt that Minnesota was ripe for the 4th wave. It looks like it may get in on the tail end of the 3rd. At the time that data was extracted, COVID deaths were averaging 16/day. At the spring peak, they were 24. Now they are up to 32, which I imagine puts them right in line with their peak weekly deaths during flu season. Now, I would argue, that it is not unreasonable for Minnesota to go above that threshold, given that COVID is worse that the flu, meaning equivalent deaths yield fewer hospitalizations. In this analysis, many other states that have had worse 2017-18 flus are "benefiting," by having high numbers of deaths for their COVID deaths to cross. Simply assuming a moderately bad, "bad" flu season, which I define as a 10-25% increase in deaths at peak, would mean that Minnesota could easily more than double daily deaths from where they are now, and still be well in company with those who have managed to flatten (the original goal), but not crush the curve.
As if to make my point for me, Mississippi. Mississippi had an extremely severe 2017-2018 flu season, thus, it is not unexpected that it would have a severe COVID season (the factors predisposing patients for death in either one are the same). However, Mississippi's peak COVID deaths were only 10% above the weekly deaths for the 2017-2018 flu season. Compare this to Minnesota directly above, which clearly has a much "flatter" COVID curve, but because the comparison reference point is lower, Minnesota much stay flatter to achieve, by percentage, the same result--hence why I argued that 25% above normal flu ought to be a reasonable goal, given that this is not an unreasonable result in a bad flu season. Mississippi has also had a reasonably large number of non-COVID excess deaths--32%, or 1 non-COVID death for every 2 COVID deaths. These deaths did help to push Mississippi over the flu threshold, and had they been able to limit them, they may well have been much closer to these flu-season levels of mortality. Mississippi remains largely flat, with COVID deaths at roughly half their peak, and excess non-COVID deaths diminishing. These, coupled with rising expected seasonal deaths may mean that Mississippi is soon very close to baseline seasonal death levels--even if COVID deaths do increase due to COVID standing in for the grim reaper this year, in lieu of flu.
Missouri has yet to exceed deaths seen during the 2017-18 flu season. Recent surges since this data was extracted have gotten it closer, but likely not exceeded it. At the time of extract, the peak average deaths was 34, it jumped as high as 40, but is now back to 36. Missouri has also seen a sizable number of excess non-COVID deaths, 2163, or 43%. These numbers appear to be going down, which will help Missouri stay below the 2017-18 flu threshold.
Like many healthy, active Western states, Montana has very little variation in weekly deaths throughout the year, with flu seasons not impacting them greatly. As a result, the 2017-18 flu was no different to any other flu season in Montana. Despite this very narrow range within which to work, Montana has only exceeded flu deaths by 15%. At the time of data extraction, Montana's deaths/day were 9, they jumped to 12 after, and have since receded. Montana does also have a high level of non-COVID excess death--36%. Were this limited, Montana would likely be below standard flu season weekly deaths.
Nebraska has managed to stay below peak weekly deaths of the 2017-18 flu season. At the time of data extraction, Nebraska's deaths looked to be rising. This rise has held steady, and cases are now beginning to stabilize and decline. These two factors mean that Nebraska will likely not exceed peak weekly deaths for the 2017-18 flu season. Recently, Nebraska's non-COVID excess deaths have also started to rise. If ameliorated this would help the state to stay below these previously high mortality periods.
Nevada exceeded weekly deaths of the 2017-18 flu season by 10%. Had they not also experienced a large number of non-COVID excess deaths, they would have been below this threshold. Since the point at which I extracted this data, deaths have increased two-fold, but this should not put Nevada anywhere near the flu threshold. Additionally, both deaths and cases are now declining. While case numbers were in theory higher than the summer peak, deaths never reached even half those levels. Once again showing that comparing case numbers across time is unreliable.
New Hampshire only barely flirted with its 2017-2018 high for weekly deaths--this despite that flu season being no different for New Hampshire than prior seasons. Since June, deaths have been in-line with baseline levels. A massive purported case surge in the last month has had virtually no impact on deaths. New Hampshire is largely open, and has been since mid-summer, so it may avoid being part of the fourth wave.
New Jersey is another one of the early peakers that sent COVID-positive patients back to nursing homes, and it shows. Peak weekly deaths in New Jersey are 150% (2.5x) higher than all deaths during any prior flu season. It is also one of the few states that makes Massachusetts look good. New Jersey is another one of the early peaking states that now has a putative case surge equal in size to its April outbreak, but which has resulted in zero increase in deaths. It is one of the more frustrating parts of this pandemic that despite being testing-obsessed, people who ought to know can conflate these two numbers, knowing the massive difference in testing. Again, New Jersey is testing at at least 10x the rate it was testing in April. Show me 10x the cases we have now, and I'll show you the same number of deaths. No, New Jersey is basking in the red glow of its April massacre, and continues to see no resurgence in deaths. However, like governors Baker and Cuomo, as if punishing those citizens still living, Governor Murphy continues to re-impose strictures on his population, despite deaths remaining at seasonal baseline levels. One again, we pray that someone tells him about T-cells. Now that Trump is, in theory, no longer a barrier to rational thought, perhaps someone could try?
New Mexico has been another one that has managed to keep deaths below weekly levels seen in any given flu season (2017-18 did not hit it harder than other flus). It does have disturbingly high non-COVID excess mortality, but that has recently declined to negligible levels. This is a good thing, because deaths have recently started to surge--something that I would have predicted based on the combination of hard lockdown--as evidenced both by high unemployment rates, and high non-COVID excess deaths, or lockdown deaths, and low overall deaths/million. Since extracting this data New Mexico has increased it's daily average deaths by 3 fold. Still, given a reasonable buffer, it might be able to stay below the flu bar. Though as with other states that have low excess mortality during flu season, that seems like an arbitrarily challenging bar here. It seems that even with this current surge, New Mexico may be able to keep daily deaths at a level of +25% of a standard flu season. Again, that would put it on the same level as those states who had significant increases in mortality during the 2017-18 flu season.
New York State, excluding New York City
New York State is, not surprisingly worse than any other state in the nation-- more than 100%, 2x, the total deaths (not just flu, all deaths, of which flu are only a relatively small portion) during any given flu season. Since June, deaths have been at baseline levels, except, not surprisingly non-COVID excess deaths, which recently, after months, and months of lockdown are becoming a significant part of deaths. It is also worth noting that while New York has only 20% of its excess deaths as non-COVID deaths, the very large denominator of overall deaths hides the fact that this is still a significant number. There are 2700+ New Yorkers (outside of the city) who have died for reasons other than COVID.
New York City
Please note the change in scale. For all of the other states, I have used a max of 400. Even for New York state above, I used 450, New Jersey I used 600. But for New York City, I had to use 1200. If I used this scale on every other state, the jump in COVID deaths would barely register, even in hard hit states. You can see that even Massachusetts is dwarfed by what happened in New York City. New York City weekly deaths exceeded 5x the number of weekly deaths experienced during any flu season. Again, that's not just flu deaths, that's 5x all deaths. That doesn't even count the non-COVID excess deaths, which again, are not at all insignificant, nearly 6000 people in fact, died from non-COVID causes, or COVID-response-related causes. Let me just try to put this into perspective. The same number of New Yorkers (only those in the city) died from non-COVID related causes as total people died in Louisiana for COVID-related causes. You might recall, Louisiana wasn't exactly a star performer when it comes to COVID mortality. In fact, New York City, has more excess deaths from NOT-COVID, than 38 states have that died from COVID. If we add New York City and New York State together, the number goes to 42. .
Again, I wouldn't be so harsh, except that I am very, very upset that New York is continually pointed to as a model for how to handle COVID. It is not. Not unless you want a large number of your community to die from and with COVID, and a significant number to die from the restrictions put in place to stop the spread of a disease that had already burned through your population like a California wildfire through wine country during a drought. Please, spare us Cuomo, unless he's willing to acknowledge the roll that immunity plays in his community's currently flat curve.
North Carolina's data was incomplete, only going through 8/29/20--but nothing has materially changed since then, so we can still use this analysis. North Carolina has never exceeded weekly deaths seen during the 2017-2018 flu season. This despite having a large number of non-COVID excess dead through out (1325, or 33%). Deaths have remained steady since this data was downloaded. North Carolina does have unusually low cumulative deaths, and a troublingly high number of excess non-COVID deaths, which speaks to a level of lockdown that might be suppressing cases. I would peg North Carolina as a 4th waver. however, the fact that they have had a fairly consistent, non-negligible level of deaths (roughly 30/day) makes me think the virus may be circulating well there, and they are managing to do a good job keeping it away from the at-risk populations. I hope that is the case, and that they can continue with their exemplary trajectory. Unfortunately North Carolina's governor appears to blow with the political winds, and despite admirable handling heretofore that has maintained deaths substantially below flu levels, he is now re-introducing restrictions.
North Dakota is the only 2nd or 3rd wave state to see anything that looked remotely like the early peakers. North Dakota saw a 50% increase over all deaths during a standard flu season (probably 70% now). However, it is still a little different from these other states, and fortunately, also looks to be subsiding. It looks to have been a brief spike, as opposed to a sustained peak (mostly, I'm sure, for lack of hosts). But it is still significant. At its peak hospitalizations, which look to have passed about a week ago, North Dakota had 400 people hospitalized. This is equivalent to 550/million, the same number that Massachusetts saw at its peak. It's peak 7-day average deaths looks to be around 15 (which it looks like it just hit, and which is 2x what it was when I extracted this data). Population adjusted, this works out to being equivalent to 134 deaths in Massachusetts. The same number of hospitalizations in Massachusetts yielded 174 deaths at peak (for a 7-day average). The discrepancy here is almost certainly due to improved care--it is certainly NOT due to the virus getting less lethal--it's still the angel of death for a certain small part of our communities. North Dakota, like the early peaking states, should not be an example, but a cautionary tale. I am all for being as open as possible, but I do also think that there are certain situations, and more importantly certain people, that require extra scrutiny. I hope that we can look at North Dakota and identify the lessons in caution that can be applied, without deciding to shut it all down. North Dakota has managed an extremely low unemployment rate, and up until a few short weeks ago, an extremely low death rate. There is a balance to be had here. We won't be able to identify the lessons that will help us achieve that balance, unless we look at the examples we have and learn from them, that includes looking at what went wrong in places like North Dakota and New York, as well as what's going right in Florida, North Carolina, New Hampshire and Utah.
Ohio has managed to keep deaths substantially below the 2017-18 flu levels (partly because it was very hard on Ohio), and admirably flat. However, it does look to me at potentially ripe for 4th wave status. As elsewhere, there are three things that tell me this might be the case. First, high unemployment usually means that a society is not really open, and thus that the virus it not circulating that freely. Second, low overall death--it's just hard to avoid. Third, high excess non-COVID deaths, in Ohio's case, that number being 3600, or for every two COVID deaths, there is one non-COVID death. Again, to me these are all signs of a society where the virus is not circulating to any great degree. Looking at case numbers shows a recent case spike, but as you may be guessed by now, I'm not a big believer in cases. Deaths remain flat. What is promising is that Ohio has significant room before it gets anywhere close to 2017-18 flu levels of deaths. If it were able to reduce its non-COVID excess deaths, that ceiling would be even higher.
Oklahoma has kept deaths substantially below the levels of the 2017-18 flu. Additionally, it has kept non-COVID excess deaths very low. Normally, I would say that Oklahoma was a 4th wave candidate. However, its very low unemployment rate, and low excess non-COVID dead suggest a well-functioning society where the virus' transmission is not greatly limited, except where it comes to vulnerable people. Further, the point at which I extracted this actually represented near high in terms of daily deaths--and yet death levels are almost in-line with baseline deaths. COVID deaths have increased by about 30% since I downloaded this data, but a 30% increase in that level of excess would still keep deaths substantially below 2017-2018 weekly death levels.
Oregon has kept weekly deaths substantially below flu levels. In fact, if it weren't for a significant number of non-COVID excess dead, it would be nearly impossible to distinguish 2020 deaths in Oregon from prior years. I do believe the combination of extremely low COVID deaths, coupled with high unemployment, high excess non-COVID dead point to a society where the virus has been largely suppressed, which I believe makes Oregon a 4th wave candidate. It has seen a large case surge recently, but it is not large enough to make much of a difference. Perhaps Oregon will escape and make it to the vaccine. Whatever the case, it can certainly sustain significantly higher numbers of COVID deaths before even reaching the flu threshold for weekly deaths.
Pennsylvania is another early-peaking state. However, Pennsylvania has done far better than many of its peers, relative to the 2017-18 flu. Pennsylvania only increased weekly deaths by 20% over weekly deaths at the peak of the 2017-18 flu season. This is despite a significant initial death surge. Like many other early-peaking northeast states, Pennsylvania has seen a massive cases surge over the last six weeks followed by virtually no increase in deaths. High unemployment tells us that the governor there is continuing to keep things locked down, despite ample evidence that a significant case surge now will not result in the same sort of challenges as the spring.
One of the early peakers, Rhode island's peak deaths were 50% above 2017-18 flu levels. The 2017-18 flu in Rhode Island was not significantly more severe than most, but Rhode Island does appear to be typically fairly hard hit by flu as evidenced by a rather pronounced decrease in deaths during the summer. Like many small states, Rhode Island has more of a needle like spike--kind of like the mirror image of North Dakota. Rhode Island has kept non-COVID deaths to a minimum (as in, lower than zero), which is to its credit. Rhode Island, like all early-peaking northeast states has seen deaths at baseline levels since July. At the time I downloaded this data, Rhode Island's case levels had "spiked" theoretically to the same level as the spring. But, as with most other northeastern states, deaths stayed flat, 10x below spring levels. Since that time cases have doubled again, once again without any concomitant increase in deaths. As so many other places, Rhode Island can experience a significant surge in COVID deaths, and still maintain baseline death levels, as COVID substitutes for this winter's flu (which as noted earlier, it has nearly fully displaced for the time being).
South Carolina experienced a summer surge that brought its weekly deaths to 18% above levels seen during the 2017-18 flu season. Like many other states, excess non-COVID deaths were responsible for pushing it over that threshold. Since that time, South Carolina has brought deaths down to nearly baseline levels. Since this data was extracted, COVID deaths hav decreased by another 33%, surely bringing South Carolina back to baseline territory. As elsewhere, even if South Carolina were to see a mild surge in COVID deaths, it could still stay will within baseline seasonal levels as COVID displaces flu deaths this year.
Like many large, small population states for a long time, South Dakota had managed to keep deaths below levels seen during flu season. It has since jumped to 10% above a standard flu season. The 2017-18 flu did not his South Dakota harder than other flu seasons, thus, this is simply above baseline flu levels, which still seems like a pretty good outcome. South Dakota's excess non-COVID deaths are amongst the lowest in the nation, which should help it to keep excess deaths to a minimum. Since extracting this data, South Dakota's COVID deaths/million have risen to 680. COVID deaths have doubled to 15/day, which means that likely this chart in a few weeks' time will look like North Dakota's. Hospitalizations and cases look like they may just have peaked and begun to recede, we will see. Hospitalization levels in South Dakota are now at the levels seen in Massachusetts, though population-adjusted, peak deaths are 40% lower. It is possible that this increases, as those now hospitalized die. However, South Carolina reporting of deaths and hospitalizations appear to be in near lockstep. Perhaps that will continue, perhaps not. If South Dakota is able to keep hospitalizations and deaths at this level and then drive them back down, it will have managed to stay within the lanes of the Massachusetts worst-case scenario. I hope they are able to do so, and again, that they do not resort to hard lockdowns to deal with problem, an easy answer during what is now, I'm sure a trying time in South Dakota, most particularly amongst family members of those impacted, and first responders.
Tennessee has managed to keep weekly deaths below the levels seen during its 2017-18 flu season. This is despite having very high levels of non-COVID deaths. In Tennessee, more than half of its excess deaths have been due to non-COVID cases. Tennessee is currently experiencing a mild case surge--30% above where it was when I extracted this data. If they are able to reduce their non-COVID deaths, they will be able to keep deaths below the levels seen during the 2017-18 flu season--which for Tennessee was not severe (which I am defining here as 25% above normal levels of death seen during a standard flu season).