Follow-up: Single-payer health care

Comments (9)

Seems like he sent a lot of documents, and debunked the other guy (with the early 90's use of data) pretty good.

If two people have data that is different and one is over 10 years old, you have to go with the newer data, right?

Posted by BJ | June 17, 2009 1:28 PM

I'm not sure. What I found fascinating on the show -- and also in the comments below -- was the assertion that there was a lie involved as opposed to a different set of facts.

It gets to a more common problem -- evaluating data. I think we're more inclined to believe data that reinforces what we already believe, rather than consider data that does not.

And in shows and topics like this, we get inundated with a blizzard of data.

That's why I think first-person accounts of living under both -- in this case -- the Canadian and the U.S. health care system are more important.

Posted by Bob Collins | June 17, 2009 1:54 PM

It seems one can find credible data to back up any assertion. Maybe keeping it simple is best. I keep coming back to the fact that Canada makes it work on lower per-capita incomes. I hear that they are largely satisfied, but then I have no way of knowing if that is real, so there I am back to wanting data again. Sigh. Just give me at least a backup plan that doesn't rely on my increasingly strapped employer. We can fine-tune it later when we have results to look at.

Posted by John P. | June 17, 2009 3:06 PM

Thanks for the work Bob. There's lots of stuff to evaluate. In the first two pages of "medical bankruptcy" I found some assumptions that frame the facts counter to what I've witnessed ... BUT I'll do my best to be objective.

Posted by tiredboomer | June 17, 2009 4:58 PM

In the previous post, the phrase "... frame the facts counter..." should read "... frame the discussion counter ..."

Posted by tiredboomer | June 17, 2009 5:04 PM

Thanks Bob for Posting Miller’s E-mail.

We really do have to get away from this high school debate mentality in this country, it’s seriously interfering with our ability to form rational public policy. Yes, one can throw different facts and data around, but some data is more illustrative than others, and some sources are more reliable than others.

With that in mind, I have a couple preliminary observations. First, Himmelstein’s information is derived from his own published peer reviewed study that appeared in the New England Journal of Medicine. For those who don’t know, this journal is considered to be one of the most rigorous journal’s in the world, not easy to get published in. But what does “peer reviewed” actually mean? It means that a jury of researchers who are familiar with the particular subject, and unaffiliated with the researchers, ( in other words independent) reviewed the study prior to publication. The reviewers are knowledgeable about methodology, i.e. statistical models, data collection etc. They review the study and vet it in terms of its methodological soundness.

Beyond the peer review process itself there are notable characteristics of peer reviewed publications. Published studies have to adhere to a standardized format that includes: Abstract, a description of the methodology, a description of the study and why it’s done, results, summary of results, and the actual data. The idea behind this is to provide enough information for someone else to replicate the study, and see if they get the same results. It also allows for in depth critical reviews of methodology and data collection.

The peer review process doesn’t guarantee reliability but the most reliable method of study we have.

None of the stuff Miller sent comes from peer reviewed research studies. With the exception of the spread sheet, they all appear to be commercial products created for specific audiences, they are not attempts at replicable research. The Casalino document appears to be little more than a power point presentation. The only reference to methodology is the description of questionnaire that was mailed out to US and Canadian providers. This is market research methodology, and a very unreliable form of data collection. It’s very difficult to get a representative sample this way, and without knowing the exact nature of the questionnaire itself we have no way of knowing how reliable it was. The Zycher reference likewise is not a study, it’s a report issued by a private special interest research foundation. Zycher doesn’t even try to explain how he arrives at his conclusions he simply declares them. As for the so called “Sherlock” report, no author even bothered to put their name to this thing so we don’t even know who wrote it and we have no idea how the data was collected. The Sherlock report appears to have been written for one insurer, Blue Cross Blue Shield; we have no idea why the report was commissioned, or what it’s supposed to tell the client.

As for Millers’ spreadsheet, this thing isn’t even a report, it appears to be a spreadsheet of his own creation, it’s not even formatted the same way the spreadsheet downloads from HHS are. He claims the source for the data is a download from the Centers for Medicare Medicaid website. The link he provides as a source is dead, there’s nothing there. Furthermore, the data he’s referring to doesn’t appear anywhere on the site. There are no reports on administrative costs for health care containing any figures. This may be what Himmelstein meant when accused Miller of lying. Miller said that he was looking at a spreadsheet on the HHS website, or an HHS spreadsheet, it appears that if he was looking a spreadsheet at all, he was looking at his own spreadsheet. Where he got the data for that spreadsheet is anyone’s guess. Even if Miller did at some time find that data on the HHS website, who knows where that data set came from or what kinds of administrative costs were being measured or why. In other word, it’s a spreadsheet alright, but in the form provided it’s useless. I’m not sure why Medicare Medicaid would have detailed information on private insurer administrative costs anyways, or how they would collect that data. I didn’t see any publication on the website that even attempts to evaluate the subject.

I don’t know whether or not Miller was lying, what I do know is that I wouldn’t assign any value to the “data” he uses to support his claims. Now it’s possible that Miller and Himmelstein have gone round this before, and Himmelstein has examined Miller’s methodology in the past. on the other hand Miller’s claim that only 10-12% of our health care dollars is going to administrative cost is so wildly contrary to most of the other data being used he may have just triggered a powerful if inappropriate reaction of indignation from Himmelstein.

Miller claims he has issues with Himmelstein’s methodology. However, presumably, Miller would point to the sources he provides as reliable examples of methodology, if so his judgment regarding methodology is seriously unreliable.

Posted by Paul | June 18, 2009 9:44 AM

Addendum to my previous post:

I didn't have time for more than a cursory examination of Miller's submissions. I took another look at the Casolino slide show and there are some interesting things there. Some of the data comes from what may be a peer reviewed journal of sorts: Casalino, et.al, “What Does It Cost Physician Practices to Interact with Health Insurance Plans?” Health Affairs Web Exclusive 5/14/09". Health Affairs may be an industry journal, I don't know how closely they adhere to the peer review model. Some journals are more rigorous than others, as noted the New England Journal of Medicine is recognized world wide as a rigorous journal.

You have to pay to see the whole article so I can't evaluate the methodology except to point out that this mailed survey method is problematic, it could be that the authors compensated for this but there's no way of knowing without seeing their methods.

Having said all that, here's what's interesting: This is the first reference in Miller's e-mail. The issue at hand was whether or not Miller's claim that only 10-12% of our health dollars go to administrative costs is a valid claim. This study even it's complete form (assuming the various summaries one can find are correct) offers no evidence of any kind to support Miller's claim. It only attempts to asses physician's costs, associated with payers. It makes no attempt at all at any comprehensive measurement of administrative costs throughout the industry. And, even in it's limited evaluation, it warns that these costs are the fastest rising costs within the sector and that compared with the Canadian system they far greater than they should be. In other words, on the face of this study actually contradicts Miller's assertion that we should ignore the problem of administrative costs and focus on the other "90%" of spending. The Casolino study actually provides some nice evidence in favor of single payer.

Posted by Paul | June 19, 2009 10:03 AM

//The Casolino study actually provides some nice evidence in favor of single payer.

I'm sorry, let me rephrase that. The Casolino study is more supportive of single payer, it compares the US and Canadian systems and finds that the Canadian system is less expensive and more efficient. The validity of the study one way or another still needs to be evaluated.

Posted by Paul | June 19, 2009 10:08 AM

I am Doug Sherlock and the author of the Plan Management Navigator that Paul discusses. We are very small so that if there are any questions about our work, I am always available to discuss it. That is the reason that we did not mention an author. No one commissioned us to write this, and we provide it free of charge to anyone.

Paul notes that peer review helps to assure the quality of the reports because, "a jury of researchers who are familiar with the particular subject, and unaffiliated with the researchers, ( in other words independent) reviewed the study prior to publication." We are not in an academic setting but the materials that we publish subject to similarly rigorous scrutiny. The source of our data is responses to very detailed surveys submitted confidentially to us in exchange for a report that helps the plans determine how well they are performing. Because the resulting analysis is primarily used by firms that participate in our studies and others who license our benchmarks, if they were not reliable then the deeply knowledgeable health plans would not participate or use them. This year, fifty health plans serving 39 million members are participating in our studies. Health plans serving 1 in 2.7 insured people used our 2008 editions of our studies.

We are not sponsored by the Blue Cross Blue Shield Association or any other organization though we have from time to time accepted consulting assignments from various firms and industry groups. Blue Cross Blue Shield plans operate independently of one another, and 21 of the 39 independent Blue licensees participated in our study this year. Paul's comment seemed to me to imply that Blue plans are one entity or share a common operating platform but they are separate organizations.

Finally, I know that citation does not constitute peer review, but I was honored that Dr. Himmelstein thought well enough of us to cite us in a paper he published in the New England Journal of Medicine concerning the Costs of Health Care Administration in the United States and Canada in 2003.

We certainly invite any questions that anyone might have about our research. Our website, which has some interesting summary content on health plan administrative costs is www.sherlockco.com.

Posted by Douglas B. Sherlock | August 15, 2009 12:55 PM