For much of last year I used publicly available data and news to attempt to create models that demonstrated which mitigations we were doing were effective, and in what order. In addition to just understanding what was going on, and being bored and confined, I have an interest in complexity theory and the ways we learn, and don’t learn.
I encountered many problems doing so, above and beyond bad data and proving causation, including the PCR-RT false positives, the difficulty of measuring actual Vs reported Vs mandated behavior, my failure to correct for geographic spread when looking for correlations, and the non-linear nature of the spread of infections. Many measures, including disinfecting, grocery store aisle directions, 6 feet, data was just not available to even attempt to model them.
On return from Fiji in Feb 2020, I was wearing a clean NOOSH N.95 mask during travel back to the USA. At that time the outbreak at LifeCare had just been announced, but I had been watching China since December. When I traveled internationally, multiple times, in the fall of 2020, I was wearing a clean NOOSH N.95 mask all the way.
I did not start out thinking masks don’t work, and I still think that worn correctly, they are likely an effective low cost prophylactic.
During the summer of 2020, there was enough geographic variance in mask wearing that it was possible to find decent A/B comparisons, and control for most other likely important variables. In my models I have a range of “percentage wearing”, prior to late Sept, of 50% to 85%.
I’d like to say I found no difference in outcomes correlated to this variable, but I didn’t. I found a slightly worse outcome based on higher percentage of mask wearing.
After the beginning of October, there was no reason to believe the globe wasn’t masking at about 90-95%, despite widespread unsubstantiated belief that “we’re doing it, but they are not”. And save a few exceptions, the entire western hemisphere experienced the largest, by far, wave of C19 deaths. While masked to the nines.
I don’t postulate that mask wearing caused wave 3, despite the correlation. It’s possible, but seems unlikely. But there’s little evidence it helped in the least. None, really.
There are a couple of counter-arguments I want to address: It would have been worse. I can’t disprove this, but while I did find some variables that seemed dominant throughout the last 12 months, masking wasn’t one of them. And on the decay side of wave 3, there was again considerable difference in behavior between say, California and Florida, with outcomes being worse for mask states. There are few counter examples of this pattern.
Another argument, which might explain apparent success in South Korea and Japan, is that there is a critical point of mask compliance and hygiene, and anything below that point leads to spread. This is curious and I cannot prove nor disprove it.
Arguments I will not address: I am not a doctor, I am stupid, I am anti-science, I let Trump do my thinking, I don’t care if you die.
So what variables do seem to have an effect? In my admittedly crude models, in order: global travel in/out, local travel (cell phone data), population density, average age.
I gave up trying to model lockdowns. There are just too many variables to attempt to control for. Maybe they delay things, but they don ‘t seem to change outcomes unless you stay in lockdown until a cure arrives.
There are two maybe instructive populations I’m watching now. The first is Israel, who demonstrated that at about 45% vaccinated, the disease – at least in the short to the present term – drops to near zero, and effectively zero at about 60%. But the news is taking about vaccine resistant mutations now, and the Indian variant is in Israel. We shall see in 6-8 weeks if the current crop of vaccines are effective against mutations, at least in this population.
The second is my old home Nova Scotia. After successfully maintaining a basically zero infection curve for 8 months (with travel restrictions, I suspect), the “case” curve over the last 10 days is looking more like the upslope of a wave 3. And the province has already implemented a strict lockdown, which will likely have a high compliance. Note that there has been, and likely will not be any real change in travel policy, or masking behavior there.
Here’s what we might learn:
“Cases” went up as testing went up (1000/day to 10,000 day). Hospitalizations and deaths have not gone up … yet. If cases grow but health outcomes do not (in proportion), the crisis might be a data ghost. I believe the PCR-RT false positive problem resulted in massive policy mistakes based on not understanding actual cause and effect last year around most of the world. I will note that an increasing % of tests are Antigen, which seem to have a lower false positive rate than PCR-RT, and the government is making access to tests more difficult.
If testing remains constant, and cases come down in 2-3 weeks, this will argue that lockdowns are effective.
If cases and outcomes grow into a traditional wave 3, until vaccination rates reach maybe 30%, then the argument that masking and lockdowns will be even weaker. Indeed it will show what I actually believe, that we have far less control over this than we think. I have no hypothesis about how to prove/disprove the traveler/student theory which is popular in the press there now, or if C19 was moving undetected all along.
I want to write something insightful about the corrosive effect of politics on science and humanity’s decency, the apparent innate human need to find someone to blame, the ease which which people can be tricked, the complexity on non-linear systems, and our failure to learn despite apparent massive amounts of data.
But I don’t know what it would be.