Way too much interesting research today. First up, everyone is interested in children–are they susceptible, do they transmit the virus? A study from Belgium gives some more data. (Medrxiv Paper) The authors followed 84 children aged 6 months to 30 months in a day care setting. The researchers had actually begun in the fall of 2019 with a study looking at pneumonia susceptibility. The study was ongoing during the early weeks of the epidemic in Belgium and created an opportunity for testing for coronavirus. None of the children tested positive. This supports the notion that children are not likely to be significantly involved in transmission of the disease. About half the children did, however, have symptoms of the common cold, which might suggest they had active coronavirus antibodies and other immune defenses. It is going to be very interesting to disentangle the effects of existing antibodies to coronavirus.
Next up, another antibody study, this one from Milan. (Medrxiv Paper) The researchers were focused on healthy, asymptomatic adults. Antibodies against the capsule protein were tested for. A random sample of 789 blood donors aged 18 to 70 was used, from the period at the start of the outbreak forward to April 8. Blood from another 184 people who were involved in a different study was also tested. The central estimate of prevalence at what was viewed as the start of cases was 4.6%, indicating that the virus was actively circulating before cases were identified. That rose to 7.1% by the end of the study. The results led to the conclusion that only one in 20 cases was actually diagnosed by a positive infection test. The other 19 were presumably asymptomatic individuals. The authors hypothesize that younger persons with more contacts were infected first, then the disease reached older individuals and the outbreak of severe cases began. This is consistent with the hypothesis I set out a couple of weeks ago about the nature of the epidemic, that it circulated widely early on with a very low burden of illness, then hit more susceptible populations. It also appears that it is a real possibility that the stay-at-home orders exacerbated the epidemic by placing infected people with asymptomatic or mild infections in closer, more constant contact with symptomatic individuals.
The next paper may be a bit of a downer, as it deals with the typical duration of antibodies for the previously circulating strains of coronavirus antibodies. (Cor. Antibodies) This is a pretty interesting study in which ten male subjects were followed over decades to ascertain variation in antibody response to standard coronaviruses. They were tested every three months at one point and then every 6 months, but there was a multi-year gap in the study. People were regularly re-infected, as antibodies seemed to last or be effective for as little as 6 months with an average of 2 1/2 to 4 years. Interestingly, some of these subjects’ antibodies reacted strongly against the current strain. Their results also indicated that coronavirus is seasonal, with low activity in the summer months. Since antibody strength and durability may be conditioned on strength of infection, I wouldn’t be too worried yet about how durable antibodies to the current strain are. And if we all are constantly generating antibodies to other coronaviruses, and those antibodies are cross-reactive, that boosts protection. But the interplay of antibody and other immune reactions to general coronavirus strains with this one, again, is going to be important to understand.
Next up is another exercise in ascertaining the True Infection Rate, which includes all undetected cases. This is the infamous denominator problem. How do you know how much of the population has actually been infected, and how do you determine real rates of hospitalization and death, without knowing all the undetected cases. Numerous antibody and other studies suggest very high rates of undetected infections. A paper at one of my favorite sources of health care research, the National Bureau of Economic Research, looks at the problem. (Unrptd. Cases Paper) The researchers used travel patterns from areas with infections and reported rates of infection at that area and the locations traveled to. They used Iceland as a source country because of its extensive testing. They gathered information on international rates of travel to and from Iceland. Without going into all the details of their method, they found that about 6 to 24 cases were unreported for every case detected by testing.
And finally another paper on the same topic from a different source. (Medrxiv Paper) These researchers used a “backcasting” method. The bottom line here was that found across 15 countries that the true rate of infections varied widely across countries, but actual cases in total were about 18 times reported ones. One of the relationships their work suggests is that where there are more total tests, the true infection rate is likely closer to the reported one, and when a country has a higher percent of positive test results, the true infection rate is likely far higher than the reported one. For the US, they were estimating about one in 15 cases had been detected.