(Tl;dr: food, animals, and yourself are all carriers of disease that the average person gets ~.2x/year normally; if you include immune reaction despite non-illness, the rough base rate is probably anywhere from .02x/year to x/year and depends heavily on person; my own odds seem like ~10:1 non-COVID to COVID given a symptom, but yours are different; I'm not a doctor and don't have metis on this, maybe I say stupid things)

Now that people have been shut away for weeks, they've (reasonably) expected to stop getting colds and things. But I've now seen at least 3 cases, one with an extremely high level of quarantine, where someone still got a cold-like illness. This affords a few hypotheses:

  • Perhaps the base rate for sickness with no human contact is not as low as we thought
  • Perhaps disease spread takes more advantage of tiny amounts of contact than one would expect from a model of p(illness) amount of pathogen contacted
  • Perhaps people are terrible at quarantining

I don’t think it’s the last, and the second one is interesting but hard to investigate, so this post will be about the first hypothesis: can people easily get sick while alone?

I'm going to sketch a catalogue of the options and roughly divide these apparent illnesses into two groups: those spread by significant external amounts of pathogen vs minor internal amounts of pathogen. (Base rate estimates at the end.)

Significant external amounts of pathogen

The three main sources of these illnesses are food, fauna, and fecal.

Food poisoning can be caused by lots of different types of bacteria. For example, Campylobacter, Clostridium, Staphylococcus, and E. Coli seem to be some of the main culprits, but there are many others (and 80% of cases appear to be caused by agents we haven’t identified!). These present like "stomach flu", often for 1-6 days after 1-10 days of incubation.

Apparently a lot of the diseases dogs spread overlap with food poisoning bacteria (and even norovirus, which also presents as stomach flu).

A lot of fecal-to-oral infections appear to be from similar bacteria. (Edited for clarity) For the purposes of this post, we're ignoring diseases from other people’s fecal matter, and some of these are obviously spread person-to-person, like adenovirus. But the four listed above (can I call them the Big Four since they keep coming up?) appear to naturally occur in the lower digestive tract, so you can presumably self-contract them unless there’s some effect where you have serious immunity to the specific strains in yourself.

There are at least some others.

  • Fungal pneumonia looks pretty nasty
  • Streptococcus A is usually spread person-to-person but is also found on the skin and regularly causes food poisoning and can thus probably be acquired from oneself
  • Mold and “stuff in dirt” also seem like prominent options, though maybe mostly minor symptoms
  • Allergies are definitely minor but can mimic colds in some circumstances

In general these external infections seem mostly bacterial, since fungi are typically too weak to cause noticeable illness, and viruses mostly need other humans to reproduce unless they have animals. Not sure how many protozoans are in this group: I’d think lots, but haven’t heard of any.

Minor internal amounts of pathogen

This is the category for a bunch of weird things that may compound on each other, which you don’t really find much about in the medical literature.

The factors I see here are:

  1. Fluctuating native pathogens
  2. Variation in immune function
    1. Between people
    2. From health
    3. As calibration to ambient disease risk

As an example, you have a bunch of candida yeast on and in your body, not very deep because high temperature harms it. It is sometimes but very rarely invasive (maybe because mammals are specifically warm-blooded to fight fungi!). But in immunocompromised individuals, especially AIDS patients, suddenly invasive infections become vastly more common, and many of them have 50+% mortality. But they aren’t constant, so this must be responding to various fluctuations in candida infectiousness and immune function.

Obviously variation in immune function also applies to the external pathogen sources from the above section, but the application to native pathogens is especially interesting. Is it plausible that immune function has a wide enough range of function that internal pathogen sources may cause a significant number of flare-ups in various circumstances much less extreme than AIDS?

Here’s the pitch for plausibility.

Native pathogens are bacterial (especially the Big Four), fungal, and viral; mostly they’re on the surface of skin or mucosal membranes, but there is some penetration and humans may have a fair amount of internal viral load. We know highly-immunocompromised people regularly become infected by these native pathogens. But if normal people undergoing fluctuations in their viral load or immune function from health had these flare-ups, I don’t think they’d be very distinguishable from colds or stomach flu. To further complicate things, I think it’s a reasonable hypothesis that your body may calibrate its immune response to what seems like the ambient disease load of your environment. Thus in times of high disease load you may have immunosensitivity and “sicknesses” from large immune response to small pathogen challenge (similar to immunosensitivity from allergies), and in times of low disease load you may be “immunodesensitized” and be at higher risk of infection. Evidence for a high rate of phantom colds, or partial viral colds being replaced by self-infection, seems extremely hard to come by, and I don’t think the absence of evidence is much evidence of absence.

Lastly, there’s some chance with such a large virome that chronic flare-ups may be more common than we think. Herpes is a classic. But other diseases may just be finicky: supposedly 2-10% of Campylobacter bacterial infections have sequelae or chronic presentation. One highly-immunocompromised boy was shedding influenza A from his stool for >2 months and from his respiratory secretions for >1 year, strongly hinting at some sort of chronic infection even though influenza A is supposed to be incapable of this. At least a few people now think chronic viral infections are a common cause of chronic fatigue syndrome, perhaps from mutation to non-cytolytic form or abortive infection. Probably non-cytolytic ones are too weak to cause flare-ups, but this is the kind of weird edge case that, after realizing it’s very difficult to obtain evidence against, makes me wonder if viruses do lots of weird stuff we just haven’t classified yet. (Adenovirus-36 may cause a lot of obesity, h/t Adam Scholl—that should probably make us pause and consider the state of our knowledge.)

Base rates and takeaways

This is all somewhat useless without base rates that can help us infer how likely it is that our quarantine has failed vs we’re just letting food sit out too long. Unfortunately, it’s very difficult to get base rates on what we care about, so I’m going to have to resort to a lot of hand-waving.

A sampling of those that were reported: E. Coli infections are at 0.1%/yr in the US. Supposedly food poisoning is 15%/year. Norovirus is about 5%/year, but probably most of that is person-to-person. Infections from pets are supposedly 1%/year, but most aren’t serious. For comparison, flu is usually 3-15%/year.

Stomach-flu-like symptoms will be at least the food poisoning cases plus other cases. That’s 15+%. If we relax from intensive cases to include the kind of minor confusing nausea I get once a year, I’d guess the rate goes up to about once/year based on personal experience. If there is weird immune modulation, I could see this being once per few months.

Common-cold-like symptoms don’t seem to come up aside from immune modulation, allergies, and supposedly very-low-base-rate things like strep on skin. Healthy people probably have skin virus infection rates like .001-1%. Standard partially-immunocompromised people (that just get sick more than normal) may be more like .005-5%. I’d guess weird immune modulation could take that number to 1-200% if it exists (if it were more than 2x/year I’d think people would start noticing). But there’s also allergies, mold, stuff I’ve missed, and all these exacerbated by overreaction to immune challenges. Empirically, I have ~~thrice a year that I seem like I’m just starting to get sick and it turns out to be allergies or goes away mysteriously, and I think most people probably have this less but some have it substantially more. I could imagine this rate being like 5x if my body was on high alert, and even a few times a year the symptoms mimicking an actual mild cold rather rather than a tease.

Of course, if your body can’t pick up on external disease cues very well, or it only does so with respect to actual ambient pathogen load rather than leading mental indicators like disgust integration or anxiety tracking, then probably these base rates go down. If you’re someone who gets sick a lot, they might be higher.

So I expect base rates of minor flu-like symptoms not contracted from another person to be about .1-10/year depending on person, and base rates of minor cold-like symptoms not contracted from another person to also be about .1-20/year depending on person (again, if more, we’d see that).

COVID ambient rates are (my guess) around .1-5% most places (more like 5% in the Bay). If you’ve been doing a good quarantine, your rate is probably about the rate it was when you started, e.g. I think I started around .1% and so that is probably about my current rate (I could easily be asymptomatic). So I think if I were very healthy, my model didn’t expect much mental immunomodulation, and I got cold or flu symptoms, I’d be about as likely to be seeing a COVID onset as seeing a false alarm. As it stands, being not super healthy and a little more sympathetic to various psychosomatic control theories (writ large, not just placebo), I think it’s more like ~10:1 that I’m seeing a false alarm.

Given the fact that these numbers are certainly going to be off by about an order of magnitude, obviously I would still monitor very closely and take thoughtful precautions that make sense both in worlds where I have COVID and don’t have COVID (don’t want to give housemates another sickness to make them more susceptible). If I hadn’t been as careful with my lockdown I’d also consider getting sick as evidence that I should tighten things up a bit where possible. Also, how textbook the symptom match with COVID was would heavily effect this estimate. Etc etc. But I hope this helps you get a handle on how to grapple with these numbers and the implicated policies.

New to LessWrong?

New Comment
15 comments, sorted by Click to highlight new comments since: Today at 9:19 AM

But the four listed above (can I call them the Big Four since they keep coming up?) appear to naturally occur in the lower digestive tract, so you can presumably self-contract them unless there’s some effect where you have serious immunity to the specific strains in yourself.

I'm not sure I understand what this is claiming. Is the idea that there's e-coli or whatever in my body, and it... like, fluctuates in how widespread it's grown or something such that sometimes my immune system is no longer handling it?

The claim is that these biota in your lower digestive tract are in containment there, but if you re-ingest them in your mouth and stomach those same strains can infect you from these parts of your body.

Those options are not exclusive - it seems likely to be some combination of all of them, including that people are bad at quarantining. And this confounds things when you try to use statistics - even if you had base rates, you'd need to know correlations in order to figure out which of them are worth intervention.

Yesterday I started sneezing and by later in the day I was incessantly blowing my nose, sneezing repeatedly, and my eyes felt puffy. The last time I was out in public (with a mask!) Was back on Veterans Day, November 11th. Yesterday was the 23rd. 12 days. Could a cold virus take that long to show symptoms? I'm at a loss in figuring out how I got a cold? From retrieving the mail after it's been delivered??

anywhere from .02x/year tox/year

What does the infinity symbol mean here, continuous infection?

Yeah, was tongue in cheek but does happen! I.e. the perpetual influenza A case mentioned

One thing I want to know is how long latent asymptomatic diseases can pass between members of a house. i.e. say 5 people all started quarantining last month, and one person had a very mild illness they barely noticed. Say the illness normally gets "processed" in about a week. Can that illness pass one by one through various people in the house, such that 5 weeks later one person gets it, and has a stronger reaction than the others?

Definitely possible, though of course it takes a few probability hits for specificity (in this case, odds of getting it the last day each time is about 1/7, and the probability of no concurrent disease is around 1/4!, so something like 1e-4.5 as likely as a "typical" spread throughout the house)

Of course, we don’t care about diseases from other people’s fecal matter, and some of these are obviously spread person-to-person, like adenovirus.

We have plenty of toilets that create a lot of areasol out of fecal matter. Giving shared toilet use, it's possible that diseases can spread.

I just meant for the purposes of this post on disease spread without being in contact with others, but I suppose it is pretty relevant in group houses where everyone is safe from outsiders but might still have a much higher rate of noncontagious spread from things like this.

Fwiw I found that sentence fairly jarring (in particular the "of course"), I think some kind of rewording about why you're not worried about it would be helpful.

(I'm personally not treating the "You're quarantined, there you don't have to be worried about the disease spreading normally within your house" as very reliable, so while it makes sense for the post to focus on other vectors, the "of course" felt weird)

Ok, I changed it to be clear it's about this post. I agree you should be worried about spread within house.

Perhaps disease spread takes more advantage of tiny amounts of contact than one would expect from a model of p(illness) ∝ amount of pathogen contacted

It would be quite tragic if different immune system strength would be more important then amount of pathogen contacted for getting infected. All those medical professionals who had an excuse to study derivation of immune system strength over tiem would suddenly lose their excuses...

Yeah. There's also things like how easy it is for your body to detect and kill (maybe infection rate is strongly sublinear in dose because of this), where the pathogen lands (if a tiny target needs to be hit and pathogen usually hits you in a single localized area, number of exposures may be more important than dose), whether a substrain is already adapted to by your immune system, etc. Nothing too crazy can be true here because of empirics but certainly mildly crazy things may be.

I've observed this phenomenon too. Quarantine errors eliminated, subjects not leaving the house, nor accepting deliveries during the incubation period. Zero human contact.

We know that retroviruses are capable of DNA integration. What if the reverse is true? What if humans can be viral proginators?