The Missing Context to the COVID Story

I’ve gone down the rabbit hole. I’ve been tracking the trends surrounding COVID-19, from as many sides and sources as I can find. It’s been eye-opening in a variety of ways, most notably the realization of just how divisive, partisan, and loaded this entire subject has become. It vexes me. I’m terribly vexed.

I’ve also seen just how incomplete most of the news about this is. It seems you can’t find any sources that aren’t saying either “Record New Cases!” or “Masks Not Proven To Work.” There is context missing from the conversation. This is my attempt to take the pieces from all sides that are worth noting, strip them of their nasty baggage, and hopefully elevate the conversation. No promises, but here goes:

This is deadly serious. It is not simply the flu, not even close. Anyone who says otherwise is objectively wrong. As of now, this coronavirus has led to the death of more than 500,000 people across the globe and >130,000 in the U.S….that we know about…in roughly 4 months…with social distancing and lockdowns. Any sharp increase in cases, hospitalizations, and deaths should be taken seriously and not minimized with silly arguments about fraud, hoaxes, or conspiracy theories. Freedom does not mean you get to make whatever decision you want, regardless of the impact on others. Wear a mask — it’s easy, and the person next to you may have a child or parent with cancer. Whether you believe it’s helping or not, you might be actively putting that person in danger by refusing to wear a mask.

There is currently a real spike of cases in 13 states. But the majority of that spike is in Florida, Arizona, California, and Texas. None of these states were hit very hard early on, it came on stronger a number of weeks later. This seems to coincide fairly closely to the end of lockdowns in those places. It’s really in about 50 counties, and it is concerning. But note that the real outbreak hasn’t moved out of those counties in about a month and half.

Most of the reporting you’ll see is about the black section of the graph below, confirmed new cases. And even as we are seeing deaths increase as a result, the dotted lines at the bottom (deaths in those 4 states) show us we are not talking about the same kind of impact we saw in April. Either this virus is not as deadly today as it was in April-May, or there were far more cases early on than we imagined. My guess is a bit of both.

Maybe there’s a 4+ week lag in deaths, and we’ll see it spike drastically in the next couple weeks. Or maybe it’ll be a slight incline like we’re currently seeing, slowly decrease, and then stick around till we get a vaccine or become sickness-resistant cyborgs.

There are many ways to present the data. Sweden provides a very good example of this. There is a recent story in the New York Times that mentions Sweden as a cautionary tale. Their relaxed approach and avoidance of lockdowns resulted in 40% more deaths per million residents than in the US, 12x more than in Norway, and didn’t even reap the economic benefits of avoiding a lockdown. Sounds pretty terrible.

But why don’t we also compare how Sweden today is fairing compared to Sweden in the past? It’s an interesting question primarily because Sweden is the only developed country hit by this virus in the world (that I’m aware of) that did not mandate some form of widespread lockdown. It should provide the best comparison to previous years’ death rates by avoiding the introduction of new variables that can influence death rates (i.e. access to healthcare, isolation). Here’s how March through June of 2020 stacks up to other years:

I wouldn’t say it was a great few months for Sweden. Worst since 2007 it seems. But I’m not sure the 12x Norway story holds the same weight in light of this presentation. This, and lots more interesting Sweden data found here.

The data itself needs more context. Think about the United States alone for a moment. We have 328 million people across 50 states, all with their own unique response to the pandemic. We have a president with no interest in attempting to unify (or do anything that doesn’t benefit himself) and a Congress who can’t agree on anything. We have 6,100 hospitals reporting different forms of data to 3,100 counties, who report that data to the state level, who then report that data to the CDC. There is no system in place to guarantee consistency. Some states report hospitalizations and ICU numbers, others don’t. Some states like Florida have only been reporting positive COVID test results, not the negatives. Many individuals get tested multiple times, but most of the time each test is counted in the overall numbers. People in Mexico are (legally) coming back into California, Arizona, and Texas for medical care.

Now multiply that across the globe — developing countries, different demographics, politics, and amounts of money. It's a problem on a scale the world has never seen, so I think its ok for us to question and sharpen the data a bit.

Another thing that blew my mind a bit is the practice of reporting “assumed” positive cases. That means for each confirmed positive result from a COVID lab test, many states are reporting up to 16 additional assumed positive cases. Apparently this is common practice, as there is no real way to test everyone — and to avoid it altogether would result in a severe undercount of cases. But the water gets a bit murkier when you start adding in “probable deaths,” deaths from “COVID-like illness,” and “death with COVID” (as opposed to “death from COVID”). Finally, the water gets perfectly muddy when you announce to these cash-hemorrhaging hospitals that they’ll receive $76,975 per COVID-19 admission. Many might assume foul-play or fraud from hospitals, states, CDC, one political party or another. There may be some of that, but I don’t think that's a real factor. Coordinated efforts from that many people is nearly impossible. I think it’s a simple fact of human nature, well illustrated by this quote by Steven Levitt from Freakonomics: “An incentive is a bullet, a key: an often tiny object with astonishing power to change a situation.” Seems to me like $76k per patient is a pretty powerful bullet.

I’ll end with a clarification on the data that no one should ignore. When you dive into the reported death counts like the one below, you should know that these deaths did not necessarily occur on the day they were reported. In fact, often a large percentage of them are from more than a month ago. (CDC data found here)

This means that the last uptick in deaths you see above is not necessarily how it’s happening. The graph, reported from actual data from individual states, does not account for the actual date of death. It accounts only for the date the deaths were reported. Below is a slightly hard to understand graph that shows the extent of this phenomenon (original here). I’ve marked it up for clarification purposes, and will do my best to explain:

The official weekly report of COVID deaths is shown in the orange row on the very top. So on July 11, 4,286 COVID-19 deaths were reported. If you follow that column on the right that I’ve circled in green crayon, you’ll see the breakdown of when those 4,286 deaths actually occurred. Note that more than 3,000 of the 4,286 actually occurred in April, more than 2 months prior to the reporting date.

Here’s what the trend actually looks like based on date of death. This is directly from the CDC, yet it is not presented in anything I’ve seen reported.

None of this would be a problem if the context was understood. This is a new logistical nightmare to track, and we don’t have the systems in place to do it the way you’d expect. But it’s not understood, and we are making decisions as a country based off assumed trends that are not accurate. That, to me, is a big problem.

Conclusions

I am nervous about publishing this. I worry that by speaking in this way I will be labeled a “COVID-denier.” I am certainly not that. I also want to make it very clear that even though a few of these things are in-line with some Trump talking points, I do not believe in any way that he has done the work to understand the data. He uses those talking points because they are convenient for him politically.

This is not a political issue. Politics determines our interpretation and response, which is not the purpose of this post. Our political leaning does not change facts. My goal here is to present my current understanding of the facts, nothing more. My family and I will remain careful by wearing masks, sticking to outdoor activities, etc. There may be even more information that completely changes my mind about all of this, and I'm open to that. Either way I look forward to honest conversations ahead.

Further Reading:

• Thread that provides a better synopsis than what I just wrote, with an introduction to the idea of the virus having a natural burnout once it infects 15–20% of a local population.
• Michael Levitt — Stanford Professor quoted in the above analysis
• Potential T-cell immunity already in the population that makes the 15–20% infection burnout more plausible
• Risk levels of different activities
• Some amateur statisticians and researchers providing some of this context (ignore most of the comments on their posts from the alt-right, ultra-woke left, or straight-up trolls. The data is real though— Ethical Skeptic, Hold2
• Stats on deaths from nursing homes, shows that that 85% of total deaths in Canada and Ireland have been in nursing homes, 45% in the U.S.