we have found evidence of substantial upcoding in some NHS hospitals for dental procedures.  Afraid paper is still in production so cannot share details at this stage.  it would be hard to believe such practices would be confined to dental procedures.

Steve Birch
________________________________________
From: Anglo-American Health Policy Network [[log in to unmask]] on behalf of Max Hotopf [[log in to unmask]]
Sent: 30 June 2017 15:59
To: [log in to unmask]
Subject: Re: Using AI to interrogate big data actually works

So does anyone in the USA actually know of any organisation currently using AI to look at healthcare wastage or anything else such as diagnostics. Is there anything out there which is actually being used rather than in pilot stage in some university lab or start up?

On 30 June 2017 at 19:56, Gemmill-Toyama, Marin (CMS/CPI) <[log in to unmask]<mailto:[log in to unmask]>> wrote:
When people talk about 20%-40% fraud, they’re probably talking about 20-40% fraud, waste, and abuse. The larger proportion of that percentage is likely made up of waste and abuse (like some upcoding) – fraud is typically smaller.

However, there is currently no robust measure of healthcare fraud in the US, and I don’t think there are robust ones outside of the US either. This is because fraud is a legal definition  and requires proving an intentional deception or misrepresentation to gain benefit.

In the US, proving outright fraud requires an investigation by law enforcement, which is costly. And as I mentioned, most of the problem is really waste and abuse. The routes that Medicare and Medicaid take are to set limits on what they will pay for and when they will pay. They then set up edits to stop or suspend claims before payment, or they will recoup money after payment. In more egregious cases, they can revoke a provider’s billing privileges for a certain period of time, or they can suspend a provider’s payments for a certain amount of time.

Is AI a magic pill for fraud? I think there is a lot that AI will be able to do. But it won’t change the fact that to actually prove fraud, payers will still need to collect medical records from providers and do interviews. AI will help them better target which providers to investigate, and it will probably be able to crunch through medical records someday to look for anomalies. But there is always going to be a very large gray area because of clinical discretion.

Marin

Marin Gemmill-Toyama, Ph.D.
Director, Division of Outcomes Measurement
Center for Program Integrity
Centers for Medicare and Medicaid Services
AR-23-54
410-786-5221<tel:(410)%20786-5221>

From: Anglo-American Health Policy Network [mailto:[log in to unmask]<mailto:[log in to unmask]>] On Behalf Of Max Hotopf
Sent: Thursday, June 29, 2017 1:46 PM
To: [log in to unmask]<mailto:[log in to unmask]>
Subject: Re: Using AI to interrogate big data actually works

I was talking to a management consultant in the UK who said that in UK NHS hospitals, the coders who decide what to bill sit in a basement and are conservative types - ex-retail bankers. They don't get to talk to medical staff so they downcode. Now take private sector hospitals in the UK. Here he said there was a constant policy of upcoding.
I quite like the private sector. You need ruthlessness to cut through the professional crap and get unionised labour in uncompetitive markets to change their ways. But the private sector are buggers when it comes to claiming.

On 29 June 2017 at 18:41, Joseph White <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Maybe 20% in Miami-Dade.

Actual fraud is probably a bigger problem in the U.S. than policy-makers like to admit.  The Medicare spending slowdown around 1997 was clearly related to the anti-fraud campaign.
But yeah, 20-40% in an entire country seems too high.  That said, we should be careful about making assumptions about cultures of corruption.  In spite of Miami-Dade.  Docs and
hospital administrators are very good at deciding they "deserve" something and it's not really cheating because they are doing the lord's work.  See all the basically dishonest
billing by U.S. AMC's over the years.  In some countries "gifts" are expected from patients to physicians.  And maybe Singapore isn't corrupt in some things but it's entrepreneurial
as all hell, and entrepreneurial physicians are bad news.

On fraud in U.S.:

http://healthaffairs.org/healthpolicybriefs/brief_pdfs/healthpolicybrief_72.pdf

I never believe one "success" story.

cheers,
Joe








On Thu, Jun 29, 2017 at 1:13 PM, Michael Gusmano <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Right. I guess I am interested in more subtle differences and disagreements. I may be wrong-- and the data may show this -- but based on the literature I reviewed it is hard for me to believe that you get to 20-40% of health care spending in any country on clearly fraudulent practices alone. Eventually this is going to touch on real disputes about the goals of medicine, on the meaning of clinical utility and what counts as evidence.

M

Sent from my iPhone

On Jun 29, 2017, at 1:07 PM, Max Hotopf <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Interesting. Well, Ted Minkinow at Fullerton said that what he called fraud ranged from a doc who claimed three days in a row for a v expensive drug treatment for very high blood pressure for the same 27 year old woman (total $80,000) through to over-treatment in the sense that you could compare the activity of two clinics with similar patient profiles and see that one was 50pc more than another, through to doctors prescribing a course of ten pills, when all that was needed was four.
I was talking to a senior manager in a large Dutch statutory insurer which recently detected what he said was clear cut fraud and said there was still a long internal debate as to whether to report it to the authorities. I thought it was revealing that a forward-looking insurer should still be conflicted over something like this.

On 29 June 2017 at 17:57, Michael Gusmano <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear Max,

Thanks very much for sharing this. I am interested in reading more about these efforts. A lot of my recent work has focused on fights among clinicians, patient advocacy groups, biotech companies, insurance companies and government officials about the meaning of terms like "waste," "clinical utility," and "value." Although the application of health services research to shape coverage and payment policies seems to be far more accepted outside the US than within, there enormous "unexplained" small area variation in every country. To what extent has the research on efforts to use "AI" (which often looks to me like algorithms applied to large datasets) addressed the disagreements about what such systems produce?

Particularly with regard to the United States I am convinced by the claim that prices, not volume, are the big problem. At the same time I am sure that there are lots of routine interventions of dubious value. Nevertheless, I am interested to know how those attempting to develop and employ these systems are going to confront the inevitable disagreements about recommended treatments. To what extent will these systems build in room for clinician or patient autonomy? To what extent should they?

Best,
Michael
________________________________
From: Anglo-American Health Policy Network [[log in to unmask]<mailto:[log in to unmask]>] on behalf of Max Hotopf [[log in to unmask]<mailto:[log in to unmask]>]
Sent: Thursday, June 29, 2017 12:32 PM
To: [log in to unmask]<mailto:[log in to unmask]>
Subject: Using AI to interrogate big data actually works
I thought you guys might be interested in what I think is the first tangible example of how AI can cut healthcare costs. And I thought we could start a thread on the subject.

There is a chain of private clinics called Fullerton in SE Asia which also works for big insurers and their corporate clients.  Fullerton has a JV with Microsoft and a good data set from Singapore.
It claims that it has found fraud/wastage rates of 20% in Singapore and that some of its large corporate accounts have seen their bills fall by 10-15% in a year (impressive given double digit medical inflation). It found much outright fraud plus a lot of overprescription and overtreatment.
I kind of thought that is slightly obscure outfit was doing this then there would be many JVs with big US payors and so on all producing similar results. But that doesn't seem to be the case.
Incidentally, Fullerton also claims it can predict heart attacks with 70pc accuracy within two years with the system. I maunder on about this here. <https://www.healthcarebusinessinternational.com/fraud-analytics-set-to-reshape-private-healthcare/>
If squeaky clean Singapore has 20pc wastage, then the figures for more corrupt systems would be higher. Academic studies put wastage at 25-33pc for the USA and anecdotes suggest China would be 40-50% plus, as they pour pills down patients' throats.
Any thoughts? I also came across Zebra Med, an Israeli start up which claims its AI software can already interpret images better than radiologists (the world's most highly paid doctors). Happy days...