Thursday, April 17, 2014

A Dumb Data Dump

While I was at ACP’s annual scientific meeting in Orlando last week, the big news, unrelated to the meeting itself, was CMS’s decision to release a massive amount of data on what Medicare paid out in 2012 to each U.S. physician. 

What is now commonly referred to as the CMS “data dump” created quite a stir—reactions from ACP members ranged from angst (“Why did this happen?  How will the data be used? What will my patients think?”)  to acceptance (“What’s the big deal? I don’t mind if the public knows how much Medicare paid me, I have nothing to hide”).   The first thing that many attendees did was to download their own data, something made easy by a tool created by the New York Times to look up a physician by their last name and town (or zip code) and then extract the data from CMS’s files (working directly with the CMS data file itself is a ponderous process).   Most I spoke to thought their numbers looked “about right” to them, although a few said there were inaccuracies in their own profiles.

Did the data dump show that any particular physician was guilty of fraud and abuse?  Did it show that any particular physician was over-charging the program?  Or that that a particular physician had ordered too many tests, drugs, and procedures?  Did it provide any useful information about the quality of care provided by each physician?  Or about the kinds of patients each physician treated and how sick they were?  The answers are: no, no, no, no, and no!

It was just a raw data dump, showing how much Medicare paid out to each physician for some of their Medicare patients (patients enrolled in Medicare Advantage plans were not included), not how much physicians actually took home from Medicare (because the data did not subtract overhead, like the costs incurred by oncologists for chemotherapy drugs dispensed in their offices).   Because the data included only what it was paid out per physician, without any context or adjustment for expenses, case mix, or quality, it is simply impossible to draw conclusions from the data about the appropriateness of the care provided by any particular physician.  It certainly is not possible to conclude that any particular physician, including the outliers who received the most total dollars from Medicare, were guilty of fraud and abuse—only a court proceeding can prove a violation of law.

Yet much of the attention in the press was initially directed at Medicare's top-paid doctors, as Fox News called them the, “344 physicians who took in at least $3 million apiece for a total of nearly $1.5 billion.”   Follow-up press coverage found that in some cases the high payments that were assigned to a single physician actually reflected payments to thousands of them.   The New York Times, in an article titled “The Medicare Data’s Pitfalls,” reported on “Dr. Jean M. Malouin, a family medicine physician at the University of Michigan Health Systems, [who] shows up as one of the top Medicare billers in the country, collecting payments of $7.58 million in 2012 for more than 207,000 patients. But Dr. Malouin directs a Medicare project that involves 1,600 primary care physicians, who each receive a small payment each month. Those payments are funneled through Dr. Malouin. The doctor’s situation is described in a website that the hospital set up on Wednesday to help explain the data to the public.”

The same article quotes ACP’s new President, Dr. Dave Fleming, on the limitations of the data. “One concern is that this is a huge data dump, and a lot of interpretation is occurring without the data actually being analyzed, with exposure of physicians who have been paid huge amounts of money. I understand the implications, but there may be very legitimate reasons as to why.”

This isn’t to say that some of the top paid doctors don’t have some explaining to do, and at least one of them is being investigated by authorities for potential fraud and abuse.  And patients might legitimately want to ask why their physician is an outlier.  Further analysis will likely show that some outliers can’t be justified because of differences in case mix or other legitimate factors.  But it is a leap too far to assume from the raw data dump that a doctor is guilty of behaving badly, never mind criminally.

The data is most useful in analyzing trends and outliers, by specialty and region, to inform public policy.  Bloomberg News has a nice chart that shows the Medicare pay-outs by specialty and not surprisingly, internal medicine as the specialty that received the most Medicare dollars, more than $8.7 billion, because there are more internists treating more Medicare patients than other specialties.  The average Medicare payment per internal medicine physician, though, was only $95,466.  The top four specialties, with average payments of more than $300,000 per doctor, were hematology/oncology, radiation oncology, medical oncology, and ophthalmology.  Yet the data dump amounts for oncologists and ophthalmologists include Medicare payments for the drugs (chemotherapy, and medications for macular degeneration) physicians in these specialties typically purchase and dispense in their offices, with most of the money going to the drug manufacturers, not the doctors.  Medicare limits physicians to a 6 percent mark-up on the drugs they buy, which some suggest may itself create an incentive for physicians to prescribe the most expensive drugs.

“Doctors make a markup when they buy a drug and then use it” writes New York Times reporters Andrew Pollack and Reed Abelson . “Medicare is supposed to pay 6 percent over the average price of the drug. That percent represents a larger number of dollars for an expensive drug than for a cheap one. Retina specialists talk of colleagues who earn huge amounts of frequent flier miles by buying Lucentis using credit cards.”

Transparency is here to stay, and the American College of Physicians believes that the public has a right to know where their taxpayer dollars and premiums are going, including how much their physicians are receiving.   Such data can help shine the light on why there are variations between and within specialties, regions and individual physicians in what they are receiving from Medicare and other payers, leading to policies to address variations that are not justified.  In the right context, such data can help inform consumers about their choice of physician—if combined with reliable data on quality, outcomes, patient experiences with the care provided,  and adjusted for case mix and overhead.  Physicians should have the ability to review the data before it is released, and get corrections if it is inaccurate.

Most importantly, when raw data are released to the public on how much physicians are paid, or on the quality of care they provide, it needs to be accompanied with explanations on the usefulness and limitations of the information available.  Medicare could, for instance, have released the data to the public with an explanation that it only showed the amount paid out to individual doctors, before overhead, not how much they actually took home;  that it did not adjust for differences in the patient population being treated; that in some cases, if the billings of multiple physicians were assigned to a single physician; that physicians were not given the opportunity to review the data for accuracy and context (and seek corrections) before it went out, that the data included only traditional Medicare and not Medicare Advantage, that some specialties treat more Medicare patients than others and will therefore have higher average billings, and that the data are most useful for research on trends to inform public policy, not for making a judgment about an individual doctor's charges or quality or fraud and abuse. 

By not doing so, Medicare made the data less useful than it could have been, unfairly tarred some physicians, and mislead the public.  The government needs a smarter approach to transparency than just dumping raw data without context on the public and physicians.

Today’s questions: what do you think of Medicare’s data dump?  Have your patients asked you about it? What are you telling them?  Have you looked at your own data—and is it accurate?

No comments :