Open Database of 100M Hospital Prices | Hacker News Reader

Open Database of 100M Hospital Prices

dolthub.com

25 pointschirau12 comments

Yeah this data is going to be open until dolthub decides to change up their monetization strategy. Whenever someone posts a link to "open" data that isn't just tar/zip/CSV/etc files -- essentially locked in to some jupyter notebook platform, or data discovery platform, or big data platform, it just looks like a money grab to me.

If I cared enough to really test their resolve, I'd extract all this data and put it out in a really open form and see how they react. I don't care enough to do that though.

The word "open" is rapidly degraded. Why couldn't the title be "Explore 100M hospital price dataset on DoltHub"?

I think you may have jumped the gun on this statement. Dolthub is a bit like github for data. You can clone the entire database without so much as a login right now, see the clone drop down situated in a similar location as it appears on github. (It is 165GB)

(install dolt which is apache licensed)

dolt clone dolthub/hospital-price-transparency-v2

The data changes over time so it is possible they will stop publishing it. Also note that this data is open to begin with and they are just aggregating it, though from a quick glace doing a nice job of it. Under HHS/CMS rules hospitals have had to publish a price index since january of this year.

[EDIT] While I am critical of dolt below, to dolt's credit, it is open source and the DB is self-hostable[2], and it looks like you can use it without registering for DoltHub. That said, it doesn't look like you can self-host the thing you log into (the "remote_host"), the docs suggest that you can only login to DoltHub.

> I think you may have jumped the gun on this statement. Dolthub is a bit like github for data.

I might have, but only because of what Dolthub is right now. Microsoft didn't buy GitHub for no reason, and under Dolthub will get the bright idea to increase their revenues by restricting access to the "open" data sets at some point. The client is open source because it's essentially a loss leader for the platform -- this is like AWS open sourcing the aws or SDKs -- the point is to get you to start depending on dolt.

If this data is "open" (for some definition of "open", that evidently not everyone agrees with), I should be able to download it without logging in, without downloading dolt, knowing what dolthub is. Compare what dolthub is doing to OSM[0]. Now holding and distributing data is not free, but that is also not what dolthub is doing here (and if it was there are some other options for trying to reduce cost in that way).

> You can clone the entire database without so much as a login right now, see the clone drop down situated in a similar location as it appears on github. (It is 165GB)

Again right now you can. Making the site look like github is fine, I don't care about that -- the problem is that to do anything I don't download `git`, I download `dolt`.

> The data changes over time so it is possible they will stop publishing it. Also note that this data is open to begin with and they are just aggregating it, though from a quick glace doing a nice job of it. Under HHS/CMS rules hospitals have had to publish a price index since january of this year.

OK, so this is one thing that annoyed me in particular -- it was very hard to find the ACTUAL source of this data. They have indeed done contributing work here in pulling the sources together (the data is scattered), but why not include a list of the actual sources? After some looking, here's an example with Ascension[1]. If this was a real Github repo, I'd be able to look at the source (that's the whole point of "open" source), not just the data. I guess this makes sense if it's "open data" but not "open source".

And you know what, maybe I am over reacting, because at the end of the day if someone wants to hack on this data they'll just download dolt, then download the data set, then get going -- but after seeing this pattern like 20 times I can't help but be annoyed by it.

[0]: https://planet.openstreetmap.org/

[1]: https://healthcare.ascension.org/price-transparency/price-tr...

[2]: https://github.com/dolthub/dolt#dolt

I wonder if this will lead to lawsuits due to pricing discrepancies between insured and uninsured pricing. There's no legitimate reason to charge the uninsured more, but hospitals do it routinely.

Presumably this is because the default rate amongst uninsured patients is much higher than insured patients. You might have to fight the insurer to get your $1m, but you know for sure they have it.

Yep. Uninsured rates are set with the expectation that at best they'll get pennies on the dollar when they sell the debt to a collection agency.

And if I pay it? Charging wildly inflated prices which were not disclosed before treatment is straight up fraud.

Then you've subsidized all of the people who don't pay. If you ask they'll almost certainly give you a discount for immediate payment since it's still more than they'd get from the collection agency.

So you are saying by charging more money to a customer, it makes that customer more likely to be unable to pay. It's like a self fulfilling prophecy.

As of January 2021 it’s required for hospitals to publish a price list on their website or else pay a fine (which I think is fairly small though, like $300/day). The legislation just came into effect but it was passed in late 2019 IIRC.

When your regulatory fine is smaller than your greenskeeper bill.

Not likely. There is a lot of nuance to the way fee schedules are actually (from a legal perspective) created and applied. Having created many such fee schedules and been resposible for operations at many facilities large and small they are done in compliance with applicable laws even though it may seem byzantine. There are always a handful of fraudulent operators though.

Reply on news.ycombinator.com