The Danes wish to know more about the coronavirus

From Martin Permin, for formatting reasons I have imposed no further indentation:

“Thanks for the excellent coverage on MR.

I lead a small team of tech workers in Copenhagen, who are donating our time and money towards building a covid-19 self-reporting tool for those citizens not (yet) in contact with health care services.

As countries shift from containment to “flatten the curve” strategies, authorities lose track of the number of non-critical cases, and to which degree people adhere to social distancing dictums. This makes it hard to predict the number if ICU beds needed a few days into the future. We’re aiming to solve this by asking all Danes for daily status updates.

Denmark is a good testing ground, but we’ll open source everything, and are building with developing countries in mind. We’re aiming to launch Monday — currently working on a green light from local health authorities.

We’re determining which data to collect. We’d love it if you’d help by asking your audience: “What daily self reported measures would you most like to see from the complete population Denmark?” (or some variation thereof).

There is of course a tradeoff between data fidelity and engagement.

What we’re considering:

  • Symptoms
    • Binary
    • Type
    • Severity
  • Whereabouts
    • Degree of social distancing
    • Hygienic measures
  • Moral
    • How concerned are you
    • Do you know anyone who’s been sick”

Are there comparable efforts to do this elsewhere?

Comments

For the category, I'd replace "Moral" with "Mental"

I think the above statistics are interesting and compelling. In addition I would consider:
- Are you quarantined (have you been exposed to someone who tested positive)
- How many tests have been conducted
- How many negative tests have come back
- geographical mapping - where are the cases by zipcode ?
- closures - geographic mapping on businesses and entities that are closed, working remotely, open etc.

You may use this idea for your app that referred to earlier in an MR post:

I have developed an app that records any contact with any individual at a distance of more than 3 to 6 feet, so that when the health department quizzes me on whom I was in contact with I can download my 3-6 feet contact list.

I am having some problems, though. Since I come in contact with people whom I do not know, I have been taking pictures of them for later facial recognition profiles.

Some people think this is a little creepy, but I tell them it is for the benefit of science, and for a prize for innovation that is available on the MR website.

The test is NOT accurate
https://www.propublica.org/article/cdc-coronavirus-covid-19-test

How many social interactions have you had today? How many of these were with someone that you had never met before?

Does that include sex?

It's been so long I can't remember. Does that count?

If you can't remember, you can qualify for another study.

If you can’t remember, you can win some Primaries.

Good one.

Wuhan Flu Over The Cuckoo's Nest.

I don't do it. The next time could be the last time.

Who is the person you met today

That you think met the most other persons who are also the persons who are likely to meet a large number of other persons

And, did you shake their hands or come in contact with them.

Donald Trump has already been identified.

"Are there comparable efforts to do this elsewhere?"

Well to put it slightly harshly, any country that creates such an app is going to get one answer immediately. The download and install rate is a measure of social cohesion and conscientiousness.

One might even imagine political rifts on who should and should not use such a thing.

Temperature

Thank you for being me to it. Fever is the only quantitative symptom.

The warden, a retired ICU RN, recently interfaced with active health professionals.

Questions/indicators: Travel overseas; interacted with an overseas traveler; moved about in area with multiple infections; elevated temperature; shortness of breath; dry cough; a rash (I haven't seen that elsewhere); etc.

I have a MD appointment Monday morning. The customary reminder call said not to come if temp is elevated and other symptoms.

Keep calm. Carry on.

Other quantitative/quantifiable items: number of positive test; number of hospitalizations; number of deaths; declines in 401K/IRA balances.

One Flu Over The Cuckoo's Nest.

Temp would be my first three choices. Others include:

*New* medications you took today (i.e. which OTCs are the most accurate predictors of looming infection).

*Increased* medication dosing (i.e. how much has your albuterol need gone up) for any PRNs you take.

How much less/more did you sleep last night than normal?

Have you traveled out of the country recently?
Is it inappropriate to ask for age and if the person had a pre-existing medical condition?
Do you live in a suburban or urban area?
Have you taken public transport in the last 4 weeks?
Did you work from home leading up to the onset of symptoms?

If you want good epidemiological data you absolutely need sex, age, and certain medical conditions such as diabetes, high blood pressure, asthma, COPD that have already been identified as leading to poorer outcomes after infection. Ideally, this should be anonymized in keeping with the EU data privacy standards.

This is March, common cold is very prevalent. How can you tell the difference? How much unnecessary burden on the testing resources will you encourage by your surveys?

Looking at cases of fever and hospitalizations for pulmonary illness -- discussed in a recent MR post -- would allow for tracking of trends that might indicate an outburst of untested illnesses that might be Covid. China supposedly did a good job but there are many who claim it was not just fake but that they hid many cases by not testing people of died of pneumonia or even who did not receive treatment. Spikes in pneumonia or other lung cases on a time series basis are one way for people to see if there are outbreaks that have not been caught.

Demographic data like breakdowns by sex and age range, proximate location in Denmark, etc

Denmark also has an excellent country wide electronic medical records system which will prove useful in epidemiological follow up.

Is it okay if we call it the Danish pox?

Too highbrow but befits the literary style evoked by your name (de plume?)

Danish pox? If that’s anything like French pox, it hurt like hell but there was sport in acquiring it.

Consider asking if they are adequately provisioned, food, medicines, etc.

Treat your app as a social network input tool that provides data to contagion models.

Read and gather ideas from the following:

1. The Sage Handbook of Social Network Analysis (see materials on epidemics, survey methods for network data, survey sampling in networks)

2. Social and Economics Networks by Prof. Matthew O Jackson (material on disease transmission networks)

Don't ask MR readers for ideas. You're wasting your time when it could be better spent contacting others who have more or better information.

Go to an epidemiologist for ideas on what you should gather. There are tons of literature up the wazoo on this subject.

Visit your government department of health to see how you can help. Figure out where they have gaps and what they would like to know or need.

Good initiative. Participation probably needs to be tied to a profile of reporter characteristics that permit it to be externally validated by tests results and hospital admissions.

Re: Are there comparable efforts to do this elsewhere?:

"Google on Friday said its sister company Verily is working with the federal government to build a tool to help people find coronavirus testing. 

Verily, the life sciences arm of Google parent company Alphabet, said the tool will first be introduced in the San Francisco Bay Area, where Google is based."

https://www.cnet.com/google-amp/news/alphabets-verily-is-building-a-tool-to-help-people-find-coronavirus-tests/

And surely the government and private companies won't misuse a database of everyone's social interactions, compliance behavior, hygiene habits, etc.

While Denmark's immediate land neighbors are Germany and Sweden, I suppose ferry traffic and even land excursions (automotive or rail) routinely take at least some Danes to visit the nearby Netherlands to take advantage of the legal status of cannabis products.

Apart from the respiratory health consequences of typical consumption, how significant might cannabis consumption or cannabis tourism be (or might it have been until recently) for Danes domestically?

(A recent report from the Netherlands:

https://www.reuters.com/article/us-health-coronavirus-netherlands/confirmed-dutch-coronavirus-infections-rise-by-155-to-959-two-more-deaths-idUSKBN2110OL )

(For this matter, in US jurisdictions where cannabis sales have been legalized, how's the market these days? I'm in Colorado and haven't shopped since last November, my rate of consumption began to decline late last year even before coronavirus-2019 became news. I suppose sales across the state to be in decline but have spotted no reports.)

Yea, good luck with that. If data collection is not automatic, data quality is ... problematic. This proposal has that problem in spades. Among many other issues, how tech savvy are the 75 and older demo that are most at risk.

Ask the Danes how terrible life is under socialism. Grown men ride bicycles to work rather than drive Teslas like in capitalist countries. It's very distressing.

Self-reported without any test confirmation? Horrible, horrible idea.

>Thanks for the excellent coverage on MR.

Yes! Your tireless efforts promoting (1) open borders, and (2) cramming as many people as possible into cities, have been really stellar.

Keep up the genius work, boys!

Yes, first open borders, then render borders irrelevant for hundreds of thousands of “caravaners”. Then strengthen the state’s capacities and call it libertarianism.

Has there been any attempt to report data for age while controlling for comorbidities?

What matters is the number of new cases.

The ratio (delta new cases t / delta new cases t-1) tells you what we need to know.

If that ratio > 1, we still in the exponential phase. If that ratio = 1, that’s the inflection point. Total cases will double from that point and top out.

If ratio < 1, we are in the leveling off phase.

Check out the video Alex linked to the other day on the logistic curve and coronavirus.

The more data the merrier (which techniques slow the spread would be nice to know)... but we need to know the above ratio by country or city or whatever. That’s what matters most in knowing where we are and what to expect. Obviously the mortality rate as well. My two cents anyways.

Agreed. More data, even data showing you are healthy, is data that is useful.

If you don't have p (percentage infected) , 1-p (that is, percentage of population that doesn't have it) also provides information. Sounds like a tautology, but random surveys tell you something without reaching the entire population and identifying all persons infected or those that just show up with some or all symptoms.

tell the danes
the american media uses the words fear/panic when they really mean prohibitly expensive liability/lawyers !

tell the danes
to look at todays british link
in the context of the cancelation of the london marathon

I think these questions do a better job of getting "ahead of the curve":

Symptoms over last 24 hour period:
Dry cough Yes/No
Productive cough Yes / No
Runny nose Yes /No
Night Sweats Yes / No
Sleep disturbance Yes/No
If Yes, free text entry describing symptoms
Measured oral / ear temperature above 37.5 degrees C Yes/No
If Yes then enter highest temperature in last 24 hours
Headache Yes / No
Breathing difficulties Yes /No
If yes, free text entry on description
Medicines taken for the above symptoms - active ingredient names, doses, dose frequency.
The next questions are catch up, required once and then only needed if they have occurred since the last entry
Number of people within 1.5 m in last 24 hours or since last entry
Names / descriptors of people (prepopulate with previous entered names , prompt for descriptors e.g. coffee waitress in Starbucks if name not known.)
Initial set up: Other recent close contacts in last 1-2 weeks
Tested for coronavirus (RT-PCR test) Yes / No
If Yes, dates of testing and test result for each date
Tested for recent virus infection (IgM test)
If Yes, dates of testing and test result for each date
Confirmed diagnosis of coronavirus Yes /No
If Yes, date of diagnosis and case identifier if provided by Health Authority
The app should provide the current national medical advice on when to escalate symptoms to testing or seeking medical advice, e.g. difficulty breathing, lower oxygen saturation, and where to go.

The key metrics are about testing

Comments for this post are closed