Mr. Andrews:

You wrote: “It is difficult to take seriously anyone who thinks that Medicare data alone is enough to inform a reorganization of the healthcare system, particularly when more recent and broader data is available.”

That’s a nice dismissive line, but I didn’t say that, you’re putting words in my mouth. You wrote that using 2005 Medicare data was “inadequate.” I countered that it was “useful.” Perhaps I should have added a more formal disclaimer that, IMO, the 2005 Medicare data and the Dartmouth Atlas methodology is sufficient and adequate for drawing certain conclusions, sufficient for informing policy decision-making, about the regional variations in inpatient utilization rates.

(Medicare data from 2005 may or may not be adequate, good, or the gold standard for other questions; the suitability of any data for decision-making or policy guidance is always context-specific. Hall-of-Fame cost accounting professor and guru Charles T. Horngren memorably stated that “Historic data is useless for decision-making”, his point being that first you must critically examine the historic data as being representative of the expected future in which your decisions will occur.)

You purported to have newer, better data that contradicted the now commonly accepted view that inpatient (utilization and) health care expenditures are appreciably higher in McAllen than in El Paso. I’m not pre-disposed towards using just Medicate data, I’m happy to looks at yours too. But a careful review of your data instead confirms the discrepancy discovered in the Medicare data/DartmouthAtlas analysis. (And I’ll pointedly note that in your reply to my original comment you did not try to refute this principle conclusion drawn from your own data.)

I am very sure that nobody is arguing, as you put it, “that price does not matter.” It matters very much. The difference is that your position is seemingly that “billed gross charges are the price,” whereas the rest of us are arguing that “what people actually PAY is in fact the price.” In markets with 300%, 400% or 500% markups from cost to charges as your data shows in McAllen and El Paso, that’s a huge difference.

The MSRP on a car isn’t the ultimate price to consumers, the “listed price” on a house isn’t usually the final price, the sticker on an item in a flea market isn’t usually what you pay, and in health care, gross billed charges aren’t the real price either, they’re not even close.

My estimate is that 90+% of health care claim dollars are actually paid based on rates that are “negotiated” by the parties. Medicare and Medicaid negotiate with a rigid take-it-or-leave-it stance. Commercial payers, except for the very tiniest, negotiate contractual rates in advance with the hospitals and physicians in their service area. You brought up the multi-billion dollar companies that exist because of a need to negotiate list prices, after-the-fact, down to an actual agreed-to payment rate.

Under- and un-insured patients negotiate explicitly to pay less than the bills, or implicitly and unilaterally by not paying, stretching out payments, or just paying as much as they can afford. This is increasingly true as the billed amount goes up … Norm and Juanita SelfPay may pay the full $90 for a visit to the pediatrician but, one way or another, they’re not paying the full $90,000 list price for his emergency admission and CABG.

Gross billed charges aren’t TOTALLY irrelevant. There’s still a little tactical maneuvering room left there between health care providers, payers and patients.

But focusing on Billed Charges for health care policy discussions, instead of payments or costs, is like looking at Burundi in the context of global warming and ignoring the U.S. or China. It’s useful sleight-of-hand if you’re wanting to divert attention from the big picture.

“Reimbursement reform” isn’t about changing the gross charges amount on people’s bills, it’s about changing the way health care providers are reimbursed (PAID), so that provider incentives and social benefits are better aligned, and the so-called health care system is sustainable and affordable for the long-term.

I think and hope that most people remain convinced that Dr. Gawande’s New Yorker article offers extremely valuable insights into how various practice and structural differences in one health care market vs another may result in demonstrably higher costs but with no obvious improvement in outcomes or quality.

US Health care is a tremendous complex system/industry/sector. Can we all agree that trying to fundamentally re-engineer it, without understanding how it’s actually working today, is dangerous and/or unlikely to be very successful ? Discussions such as this do help us achieve that understanding.

d.d.

You start by stating “using 2005 Medicare data to inform comprehensive payment reform is inadequate.”

Well, the Dartmouth Atlas as I understand it was primarily designed to tease out regional differences in risk-adjusted utilization, societial costs, and outcomes. “Reimbursement reform” is something most people inevitably think about once they see the Dartmouth Atlas analyses, but their direct objective is more aimed at highlighting and reducing unwarranted variation in physician practice.

2005 data is perfectly usable, unless you can make a case that the US health care environment has changed enough since then that “the data lessons of 2005″ would no longer apply. I don’t see it that way, but I’m open to reasoned arguments.

Medicare-only data is perfectly usable, unless you can make a case that the “Medicare market” is so different from the markets for other patients/payers that it’s unreasonable to draw some broad generalizations from the Medicare data as a proxy for the whole. (Medciare is the ONLY nationally available proxy for the whole, it must be noted.)

Your description leaves the exact nature of some of numbers you quote here pretty vague … e.g., “hospital costs” … is that the inpatient portion of all-payor costs from Medicare Cost Reports for hospitals located in McAllen or El Paso, and what did you do to “calculate” the costs of area residents treated out-of-area as inpatients?

Nevertheless, derived from your own numbers:

Admissions per 1000 capita:
– El Paso – 99.7
– McAllen – 121.7 (22.0% higher utilization)

Days per 1000:
– El Paso – 480.9
– McAllen – 535.0 (11.3% higher)

Your assessment: “Per capita hospital utilization is similar, with a rate of .48 patient days per capita in El Paso compared to .53 patient days per capita for McAllen.” Sorry, but given these two large population bases, 11.3% higher is not “similar” … that’s a BIG difference.

Much of Dr. Gawande’s New Yorker article as I recall it dealt with utilization of outpatient, physician and other (non-inpatient hospital) services, which you leave out of this quick analysis. Let’s describe that as “anecdotal impressions from New Yorker article – not refuted.” Granted, good data to look at the ambulatory side is harder to find.

Despite your wordy-words, for the Inpatient analysis your actual data per the above entirely CORROBORATES the Dartmouth Atlas view that utilization in McAllen is appreciably higher than El Paso, and you’ve concurred that the patient demographics are remarkably similar, so that is not an explanation.

You greatly overstate the significance of Charges, and that runs counter to pretty much everyone else’s take on these wildly inflated “Manufacturers Suggested Retail Prices”, which very few people actually pay.

If a Honda Accord had a Manufacturer’s Suggested Retail Price (MSRP) of $125,000, would you leave the dealer’s chortling about your fantastic $100,000 discount, or would you leave thinking you just paid $25,000 for a solid car?

We’ll skip a longish discussion of mark-ups and price-setting in hospitals. There’s an infinite variety of invoice-by-invoice payment mechanisms, with case (DRG) rates, per diems, percent of charges, etc. But the critical point is that from a 50,000 ft perspective, actual (average annual) payment rates to hospitals have historically been essentially pegged at costs plus or minus some tolerable “profit” margin. And as is true in all quasi-free-market industries, excess profits brings more competition, insufficient profits causes competitors to leave, without external subsidies long-term providers must make an accounting profit or become extinct.

Hospital COST is much closer to what insurers, patients, and govt’l payers actually PAY from their wallets and bank accounts than “Charges”. Per your numbers, MSRP Charges in McAllen and ElPaso are 300% to 400% more than costs. Meaningless.

Per your numbers, annual all-payor INPATIENT COSTS (hospital-component only) are:
– El Paso residents – $598 per person per year
– McAllen residents – $757 per person per year, or 26.6% higher in McAllen than ElPaso.

(Actual average payments by insurers + patients + govt’l payers will be ball-park 10% to 20% higher than the hospitals’ cost numbers. Most of the difference is a cross-subsidy for the patients who can’t pay, plus, if the hospital is lucky, some left-over for profit and re-investment in facilities.)

To put that in wallet perspective, 15-20% of your normal health insurance premium is for inpatient hospital stays. They’re expensive when they happen, but most people, most years, don’t have any inpatient stays in a hospital. The other 80-85% of the premium pie, which we are not directly addressing here, goes to physicians, drugs, outpatient care, other care providers, durable medical equipment, insurance company overhead and marketing, etc.

Plus, I think you really miss the boat in discussing the “out-migration of patients” issue.

Per your data,
- El Paso is 98.8% self-sufficient for inpatient services, with only 1.2% of patients (888 cases) having to get care elsewhere.
- In contrast, in the same period for comparable demographics, 6069 or 6.6% of cases leave McAllen for care elsewhere.
- In both cases, the “left for care” cases appear “sicker” on average than the ones who stayed in-county, as measured by avg length of stay and avg charges per case. This is consistent in my expectations for out-migration in various geographic markets.
- However, because in one community we’re looking at just 1.2% of the “end of the tail of the (patient population) distribution” vs 6.6% in the other, it’d be untenable, without looking deeper into the data, to assume the larger group of “left McAllen patients” are clinically comparable to the much smaller group of “left El Paso patients.” So comparing the per-case costs or ALOS for out-migrants from McAllen vs El Paso would most likely be an inappropriate apples vs oranges test.
- We do know that, everything else being equal, people like to get their health care close to where they live. It’s a considerable inconvenience to patients, and their families, to have inpatient stays far from home.
- From that perspective, we can argue with reasonable conviction that 6.6% of McAllen inpatients didn’t leave for inpatient care because they “wanted to”, they left because they perceived they “had to.” And that’s the case for five times as many from McAllen as El Paso.

There are two logical explanations for that major discrepancy in out-migration:
(1) some significant clinical service is not available from McAllen’s medical community, although it is available in El Paso, or
(2) there’s economic triaging going on in McAllen.

These two options are not mutually exclusive, some services potentially might not be available in McAllen … for anybody … because they don’t at least break even financially, and no provider/group/hospital/govt is willing to subsidize that particular money-losing operation. (Reformers please note, that’s what happens with market-based solutions.)

So … guys with the data … what does the data tell us about the payor-mix of the “leave McAllens” vs the “leave El Pasos” ? Is there comparable out-migration of Commercial and Medicare patients, or is it predominently “under-funded” patients who have to leave McAllen for care?

Also, where do the McAllen out-migraters go? That is, what kinds of hospitals do those patients end up at, for-profit, or not-for-profit/govt? In El Paso, apparently the county hospital serves as a safety net. Do we have significantly more out-migration in McAllen because there’s no equivalent safety net public hospital there? (I’m not from Texas, I don’t have any personal knowledge of these markets.)

To re-cap, per your own numbers:
(1) All-payor cases/1000 in McAllen is 22% higher than El Paso ;
(2) All-payor days/1000 in McAllen is 11.3% higher ;
(3) Inpatient hospital costs per person per year in McAllen are 26.6% higher ;
(4) You present no data on non-inpatient utilization or costs (80% of the health care pie), which a reading of the the New Yorker article suggests would have an even bigger difference than we see on the inpatient side ; and
(5) the out-migration data suggests than the McAllen market may have higher levels of financial triaging limiting access to, or the supply of, significant medical services.

Your data does not refute the Dartmouth analysis nor Dr. Gawande’s New Yorker article, it instead reinforces both.

IMO.
———————-
Aside to Mr. Puestow: would you happen to have any insights from your experience about the relative pricing of commercial insurance premiums in El Paso vs McAllen that would tend to show that, from an actuarial point of view, utilization and pricing in McAllen and El Paso are pretty similar, or fairly different?

Maybe somebody can search out the monthly premium rates paid by the county School Board in El Paso vs McAllen, demographically those would probably be pretty similar and the data should be in the public domain.
———————-

Always good to support policy analyses with data!!! Let the debate continue.

Douglas Dame
another data guy

I am fascinated by the enthusiastic efforts in this posting, and in others, to cast doubt on Dr. Gawande’s brilliant New Yorker article, and on the results from the Dartmouth Atlas. (Disclaimer: I teach in the Dartmouth economics department and work with the Atlas group.)

First of all, it’s not just the Dartmouth Atlas that presents a problem for the authors, it’s the official Medicare per capita measures of spending. According to the CMS website, 2007 per capita Part A (hospital) expenditures were 49% higher in McAllen (Hildago County) than in El Paso County, while 2007 Part B (physician) payments were a whopping 112% higher in McAllen. (http://www.cms.hhs.gov/MedicareAdvtgSpecRateStats/05_FFS_Data.asp)

Second, the authors also find that McAllen hospital costs are much more: $569 million compared to just $454 million for El Paso, and for a similar population base. (They also report patient charges which, unlike Medicare payments, are imaginary – no one actually pays them.) They try to explain away these differences by reference to a remarkable exodus of the sickest patients from McAllen to hospitals outside the county.

Unfortunately, this does not let McAllen off the hook. If McAllen is shipping the sickest patients to lower-cost neighboring hospitals, then any hospital patients who still remain in McAllen should be both fewer in number and cheaper. But they’re certainly not.

Let’s ask a different question – why are so many sick patients migrating to other hospitals, particularly if McAllen’s hospitals are so new and high-tech? More intriguingly, how are they getting to those other hospitals? Could this migration be why some McAllen hospitals spend about 6 times as much per patient for ambulance services than El Paso – hundreds of thousands of dollars per year? (See the 2008 Atlas data for Texas; http://www.dartmouthatlas.org). Have the authors stumbled across yet another reason why McAllen is so expensive – that patients are sent in ambulances half-way across Texas, with Medicare footing the bill?

I have had numerous conversations discussing the New Yorker article, and find my own opinions being informed by the “facts” as they were presented. This rebuttal provides valuable and fresh perspective that needs to be injected into the national debate.

Thanks for taking the time to research and write this detailed rebuttal.

It is both valuable and appreciated!

Lyle Beasley