Saturday, November 21, 2009

Bad Medicare Data Thwarts Good Policy

In August 2006 even George Bush tried doing something right in health care. He asked his public health agencies to make pricing and payment to providers information transparent and publicly available.

Till then Medicare and the like kept such information secret or very hard to access, ostensibly to give them better negotiating leverage with providers. I think (having been in government) it's just the natural way of bureaucrats. The less information they put out there, the less vulnerable they are to any criticism, and more able to dispense favors or act arbitrarily.

But there are very good reasons to make Medicare rates and payment information freely available, especially when health reforms are such a priority:
  • Hospitals have long maintained they lose money on Medicare patients and hence need higher private insurance and "list" rates as a cross-subsidy. The media and analysts have never properly verified these claims. They and policy makers can do so with more easily available data, and compare Medicare payments with those in Europe and other countries to evaluate their fairness.
  • Doctors too complain about low Medicare rates and especially the cuts required in them by federal law. A cut of 10.6% was eliminated for 2009, and just two days back the 21.2% cut for 2010. Precisely knowing and assessing these rates will again better shape payment policy.
  • The difference in rates and average payments across provider groups and regions can identify the outliers. Thus excessive prices and inefficiencies can be curbed while studying the most cost efficient providers for propagating best practices. Atul Gawande's Jan. 26, 2009 article in the New Yorker that so impressed President Obama shows one way to do this.
  • US Medicare payments can be compared with prices charged abroad. This will highlight the achievable savings that are being studiously ignored in the current health care reforms debate. The Wall Street Journal's front page story today was of Dr. Devi Shetty's $2,000 heart bypass surgeries in India with quality and outcomes comparable to those in the US. But even top foreign providers serving medical tourists that charge $10,000 for a heart bypass offer enormous savings as they are a fifth of US prices.
Sadly though, this Medicare payment data is incomplete and difficult to find and tally. Worse, it is plain wrong. I was alerted to this problem when Prof. Jagdish Bhagwati drew my attention to World Banks' Aaditya Mattoo and R. Rathindran's 2006 article in Health Affairs (or a detailed 2005 working paper.) The authors estimated US savings of about $1.4 billion annually, if 15 exportable procedures were performed abroad on only 10% of eligible patients.

$1.4 billion? This sum is ridiculously low compared to our own calculations mentioned in our Global HealthNet website and summarized in our May 2008 WSJ Op-Ed. There were $220 billion worth of 30 "exportable" medical procedures performed in 2006, and if 25% were performed abroad, $45 billion would be saved. And this does not factor in savings due to US hospitals lowering prices due to foreign competition. Mattoo looked at only 15 procedures and used a 10% participation rate, but that doesn't come close to explaining the difference between $1.4 billion and $45 billion.

The biggest culprit I see is the flawed Medicare payment data put out by CMS and relied upon by Mattoo. It leads to average payments being heavily under-estimated. Here's how CMS has slipped up:
  • Foreign hospitals readily provide a consolidated estimate for standard procedures and a single final bill. But US treatments typically generate a flood of separate bills from providers (individual physicians, radiologists, therapists, device vendors, different hospital services, etc.) CMS inexplicably fails to list or specify all such components so researchers are very likely to miss major ones.
  • Even the figures presented seem to be wrong. For instance, they show 42,000 heart bypasses (CABG) for Medicare recipients in 2006 with average hospital payments of $22,700 (or $33,100 for complex cases) that are only 30% of the charges. But the federal HCUP database itself shows 127,000 Medicare CABG cases and says payments average 55% of charges. (It can be about 46% for Medicare that typically pays 83% of private insurer rates but that's still a lot more than 30%.) Similarly, CMS shows under $12,000 Medicare payment to hospitals for knee or hip replacements, that again seems grossly understated at 30% of charges. Among other things it looks here that CMS omitted the sizable cost of orthopedic implants.
  • To see how wrong data can skew savings calculations, consider a heart bypass that costs a total of $19,000 in a "5 star" Indian hospital including all treatment, travel and stay. If we take the US total payment to be $26,000 then the savings are $7,000 per case. But if total US payments are a more realistic $65,000 then the savings are $46,000, an over six-fold increase.
  • CMS also makes it needlessly hard to find this supposedly very open payment information online. If you have a little time, visit their website and see if and how quickly you can find this before reading further. Of all places it is tucked away in "Research, Data,..." under "Health Care Consumer Initiatives."
Remarkably, all the data needed by CMS to compile the complete and accurate payment information and make it transparent is sitting right there in its payment system servers and storage. They can also easily have it for up to a quarter or two ago, instead of it being three years old, for 2006.

The resources needed for this appear to be absurdly meager - I'd think a couple each of in house programmers, data base professionals and statisticians working for a few days. The directives have long been in place so it does not need any legislative, political or even top administrative clearance. In absence of this ready data we are paying a high price by mulling political options and policies in a vacuum. HHS or CMS shouldn't lose any more time correcting the situation.