[Leary]

One of the key findings coming out of China is that CT scans can outperform reverse-transcription polymerase chain reaction (RT-PCR) lab tests in diagnosing the Coronavirus[1].

[1]https://www.itnonline.com/content/ct-provides-best-diagnosis...

If the CDC's current tests are flawed or too few, then the CDC should provide guidelines to health providers to quickly diagnose potential patients using CT scans.

[riahi]

As someone who interprets CTs, the findings described in that paper are totally nonspecific and just indicate a pulmonary infectious/inflammatory process. The takeaway I have is if the CT is totally clear, COVID is unlikely. Otherwise, you need confirmatory testing to figure out what is going on with the lungs.

[mnm1]

A CT scan is fairly standard procedure for diagnosing many things. Why do doctors and hospitals need CDC permission to use it to test? This seems like a bad movie where the CDC is not doing much due to political pressure. Why can't doctors act in their patients' best interest on their own?

[semi-extrinsic]

A CT has a pretty high radiation dose. As far as I understand, for every 2000 people given a chest CT we expect 1 person to get a fatal form of cancer due to the CT.

Compared to the natural incidence of getting a fatal cancer in your lifetime (about 400 in 2000), this is a very low. And for serious conditions, a CT can literally be a life saver.

But if you apply chest CTs to screen for a disease with (let's say) a 0.1% fatality rate, the side effects of your screening increases the fatality rate by 50%.

[anonsivalley652]

It's ~29000 deaths / 70M CT scans per year.

That's 414 micromorts, which is almost exactly as risky as skydiving once on average.

https://en.wikipedia.org/wiki/Micromort

Source: US NCI 2007

[semi-extrinsic]

Yeah, 1 in 2000 is 500 micromorts, somewhere in the 400-500 micromorts region.

[jonas21]

The wikipedia article says skydiving is 8 micromorts per jump. So more like skydiving 50 times?

[medymed]

I don’t think biologic math always works combinatorially like multiple sky dives.

This sometime troubles me with radiation dosing comparisons. Another is time distribution, when a radiologic imaging study is equated with exposure from X amount of time in an airplane at high altitude. The time frames are different—- drinking 1 liter of water in an hour is usually safe, and drinking 50 liters of water is fine spaced out over a longer time period—-but not 50 liters over an hour. Radiation doses are often quite time dense.

[antognini]

Yes, the rule of thumb I know is that you can get about 25 CT scans in your lifetime before having a significantly increased risk of cancer. By contrast you can get about 25,000 chest X-rays.

[eganist]

From what we can tell, SARS-CoV-2 has a 2% mortality rate among those diagnosed. That seems like a pretty decent trade in odds if it accelerates early stage treatment.

[scottlocklin]

Considering you just read an article on how the CDC has told people to not test people who aren't obviously going to die, you may want to revise that figure downward some out of sample bias.

[kalipsosu]

2% within confirmed cases. There is a part of the iceberg which we cannot see because mild cases will not seek care. Actual mortality rate is far lower than that. You should check mortality rate for outside wuhan population. It is 0.08%

[ciceryadam]

Italy's 52 deaths vs 1900 cases is higher than 2% - https://www.theguardian.com/world/2020/mar/02/italys-coronav...

[eganist]

If you want to see the outcome of a country's health department ignoring this outbreak for political purposes, look no further than Iran.

https://www.newyorker.com/news/our-columnists/how-iran-becam...

[kalipsosu]

CT changes did not occur before disease progression. In one study it occurs after 3-5 days after being admitted to hospital.(I could not find the link right now sorry.) Therefore cannot be use for 'quickly diagnosing'