As usual a couple of observations before we hit the meat of the post. My friend Scott Johnson at the Powerline blog continues to be able to send questions that the state answers. And the answers are often doozies.
- They were asked what specific data they have to show that children are actually transmitting to adults. The short answer is none. They cite the research, which shows extremely limited transmission from children, especially asymptomatic children, see the study summarized below. But here is the best quote: “We do not have individual data where we link one case transmitting to another person resulting in their infection, whether it’s from an adult to another adult, even a spouse to a spouse, or a child to an adult. We can’t possibly interview every person in depth and make a subjective interpretation as to who infected who.” Bar, restaurant and gym owners, are you reading this. Wow. Just wow. What the hell is the purpose of contact tracing? The state is here saying they can’t trace transmission, period. They are saying that it would be “subjective” to say who infected whom. Wow. I just don’t know what more I could say beyond quoting those words.
- Same issue about staff being the source of LTC spread. Here we got more precise data. According to the state, 74% of the time, the first case in an LTC is a staff member, based on illness onset. So they are inferring that the staff member got infected in the community and then came into the facility. Not clear what the role of asymptomatic resident transmission may be or the role of visitors.
- Here is some interesting data on death reporting, relevant to my earlier post this week about an apparent CV-19 death. The question was how many CV-19 deaths are based on Part 1 of the death certificate, an underlying cause, and how many on Part 2, a contributing cause. After some introductory mumbo-jumbo about coding and rules, which I couldn’t follow, because it made no sense, the state said “Even if COVID-19 is listed as a secondary cause, our experience indicates almost all of the persons listed below would not have died if they were not infected with SARS-CoV-2. For the case definition that we use for COVID deaths (the CDC/CSTE case definition for deaths), it is not significant if COVID is listed in Part 1 or Part 2.” Wow again. Then we learn that as of the date of the answer at least, which seems dated based on our total number of deaths now, there are 3408 certificates with CV-19 in part 1, or as an underlying cause, and 463 as a contributing cause. Apparently they think there could be some certificates where it is on both parts. That would be inappropriate. Either a clinical condition or event was in the chain leading to the immediate cause of death, or it was just a side condition which may have made the patient somewhat more susceptible to dying. If you assume limited violation of this precept, 12% of all deaths had CV-19 merely listed as a contributing condition. Even more interesting to know is how many of the CV-19 deaths are based on test matching, not any actual clinical determination that CV-19 was actually involved in the death.
Here is a comprehensive review of the research on household transmission. (JAMA Network) The authors looked at 54 studies with 77,758 people traced in regard to household index cases. The overall secondary attack rate was 16.6%. Symptomatic index cases were far more likely to be a source of transmission to another person, 18% attack rate, compared to asymptomatic ones, at less than one percent, .7%. Transmission was more likely to adults than to children, to spouses than other household members, and in a household with one other member as opposed to three or more. Transmission from adults and children appeared equally likely, although the age of the index cases was also associated with much more likely transmission, so those two seemed inconsistent, but it also could be that different subsets of studies dealt with different issues. More severely ill people as an index case were more likely to transmit. Household transmission was more likely than family outside the household transmission or transmission to another type of close contact.
A study published by the CDC covered transmission among children in Mississippi. (CDC Paper) This was a case control study of 397 children, some of whom had a positive test and some a negative one. Attending in-person school or day care was not associated with being infected. Associating with an infected person was. Duh. Given how participants were selected or refused to participate, there is a strong potential for bias in the results. Less frequent social contacts outside the house were associated with slightly less risk of infection. Parents were asked for information about things like mask wearing, social distancing and other precautions, but not the children. While the majority of all respondents reported seeing universal mask use in schools, the report somehow suggests that children who got infected in settings with less mask use. I don’t see that in the actual data we were given. All we see are odds ratios and the analysis doesn’t appear to adjust for multiple supposed risk factors as opposed to looking at them one by one. The typical very weak CDC work, as we saw in the Kansas mask study.
Another study on contact tracing in China. (PubMed Article) 3410 people were included as contacts of 391 index cases. The overall transmission rate to a secondary case was 3.7%. Household contacts were the most common source of secondary infections. More severely people were also more likely to transmit. Children index cases were far less likely to transmit to a secondary case than were older ages.
More contact tracing, this time in Switzerland. (SSRN Article) 219 households had infection testing and antibody surveys done among 302 household members and 69 other close contacts. 57% of household members were infected versus 19% of the other contacts. Having more contacts outside the home was actually associated with less risk. The authors interpreted this as people working outside the home, and therefore spending less time at home had less risk. So once again, let’s encourage people to stay at home, really makes sense. And the risk was lower in larger households, which is also fascinating. And here is one interesting nugget, buried as usual. Mask wearing in public was not associated with a lower risk of being infected.
This research from the National Bureau of Economic Research focuses on the economic impact of eliminating in-person schooling. (NBER Study) The paper demonstrates the substantial negative effect from parents being unable to work or being less productive. There is also an impact from children missing real education.
And no post would be complete without some mask research. (AIP Article) These was a physics-oriented paper, examining the dynamics of air and particle flow around a mask. The authors used a surgical mask and a model to study airflow with and without a mask. It is good to see the bias right off the bat, the first paragraph says “wearing a mask has proven to be an effective method of protection in this pandemic.” I believe there are about a thousand case curves that beg to differ. (and as you will see later in this lengthy summary, the authors own findings beg to differ.) It says both exhalation and inhalation are prevented. So now we know what the outcome of the study will likely be, kind of pointless to plod on, but I am not easily deterred. These are obviously not clinicians (not even sure what kind of engineers they are), for they next discuss the putative process of infection and discuss how the consequences of lung infection require mechanical ventilation. Uhhh, last time I checked the new guidelines are to avoid ventilation if at all possible, and have you heard anything lately about a ventilator shortage?
But I digress. Next up we get pure speculation on viral dose, a particularly under-researched area, but the estimate here seems quite a bit higher than what better research suggests. And realistically, given replication capabilities, even a few particles can cause a massive infection quite quickly in a susceptible person. A quick review of prior research to understand airflows around mask follows. And then we get to the nub of the method. Oh look, it’s a model, but not just any model, its a “computational” model. Yes, why try to study the effects of a real human wearing a real mask under real life conditions and inhalation or exhalation of real aerosols when you can just build a computational model. And let’s model the mask with a perfect face-seal, because that is very like real life as well (in fairness to the authors, who I am only half-mocking, they note this in their limitations section).
Remember what I have said about models, they only tell you what you tell them to tell you. Sound complicated? It’s simple, models have to have inputs and parameters and formulas and those are given to them by humans. The mask here was modeled as having a 65% filtration efficiency, which basically means that the concentration of a particular undesirable particle is reduced by 65% behind the mask compared to the general atmosphere in the vicinity. Not particularly good, but also likely means larger droplets would be blocked much more efficiently than smaller ones and that 65% is an average as far as I can tell from the description of the model. There is no reason to think that small amounts of virus on a lot of smaller aerosol particles are less infectious than a lot of virus on much fewer large droplets. To state the obvious, they found that airflow is substantially different with a mask on than without. Yes, and? Ahhh, that is where it gets interesting because it turns out that those alterations may make transmission more likely.
Under this model, the mask tends to even airflow out across the mask. Masks slow particle speed which if they pass through the mask, can alter trajectory and landing spot in nasal versus mouth passages. The effect varies with particle size, the largest being most affected. Unfortunately it does not appear that flow was simulated over a very extended period as would be typical in real life, and it doesn’t appear that what happens to particles initially trapped in the mask was identified. But one finding was that a mask may make it more likely that particles enter the nose and less likely that they reach the lungs. That is a big so what, since if an infection starts in the nose, it can easily spread to the lungs. Masks also, according to this model, may cause more particle deposition on the face, especially around the eyes. Hmm, anyway that face deposition could be a problem?
All in all, this study, if accurately modeling short-term airflows, suggests why masks not only are likely to be ineffective in blocking virus-carrying particles much of the time, but may lead to circumstances favoring transmission. And I still would like a more realistic study over a longer time, because all those particles initially entrapped in these virus collection devices, ultimately either stay or are pushed out or pulled through the mask. That would be really important to understand.