A few remarks on the Oregon Medicaid study

There is a simple quotation from Josh Barro, who by the way has supported ACA.  Josh wrote:

Despite efforts to spin it to the contrary, this is bad news for advocates of the Medicaid expansion. While Medicaid is clearly good for some things, it was supposed to be good for all of the measures tracked.

Or here is Ray Fisman:

Now that the clinical results have started to come in, it’s time for liberal media types like myself to eat some humble pie. Today’s New England Journal article presents a set of findings showing that Medicaid had no effect on a set of conditions where you would expect proper health management to make a difference. There are effective treatment protocols for hypertension, cholesterol, and diabetes, yet insurance status had no effect on blood pressure, cholesterol levels, or glycated hemoglobin (a measure of diabetic blood sugar control).

Do read the rest of those posts for a more complete picture of the results, but many commentators are overlooking these rather simple upshots.

The key question here is how we should marginally revise our beliefs, or perhaps should have revised them all along (the results of this study are not actually so surprising, given other work on the efficacy of health insurance).  For instance should we revise health care policy toward greater emphasis on catastrophic care, or how about toward public health measures, or maybe cash transfers?  (I would say all three.)  One might even use this study to revise our views on what should be included in the ACA mandate, yet I haven’t heard a peep on that topic.  I am instead seeing a lot of efforts to distract our attention toward other questions.

I am sometimes reluctant to speculate about motives, but I believe there is currently a fear of stating the actual truth, given that ACA and the Medicaid expansion are coming under increasing political fire, very often involving mistruths from the Republicans I might add.

You are seeing obfuscations of reality when you encounter two particular responses to the new Medicaid results, which I have been seeing with disturbing frequency.  The first is something like “But you still buy health insurance, don’t you?”  The second is when the debate is steered into showing that Medicaid does indeed benefit poor people (which is obviously true, and was so before and after this study).

Those are both examples of running away from the idea of thinking at the margin.  A better response would run more along the lines of “The Medicaid expansion had been oversold, we now should think more along some other lines for improving our health care system.  Let’s admit that we have more of a mess on our hands than we had realized or let on.”  You don’t have to deny that Medicaid might help with long-term care problems, for instance, or advocate the abolition of Medicaid.  The real results from the new study are most likely about health insurance and health care, not so much about Medicaid per se; see Ezra’s on-target remarks.

Compare what you have seen over the last two days with the writings on the earlier phases of the Oregon study, when it seemed to be yielding a more positive picture of Medicaid.  Those earlier writings often were preparing for a coronation of this study (please do read that link) but now we are seeing hand-wringing and all sorts of talk about the study’s limitations.

For varying and useful perspectives, here are Carroll and Frakt and Megan McArdle.

Coming on the heels of the debate over Reinhart and Rogoff, I find this all sad.  If there is any cheery lesson it is that, in relative terms, macroeconomics is in better shape than we had thought!


This definitely gives socialized medicine advocates (like me) a bit to think about. But when journalists quote from an academic study, without further explanation, things like "Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years." it really bothers me. Because most people read this as "Medicaid might do something but doesn't do much in two years". NEJM is meant for a statistically-literate audience.

In other words, people conflate statistical significance and efficacy. Because that's how "significant" is used in daily language, we can't fault them. Would it be so hard for journalists to note that this doesn't prove that it doesn't work, but it just fails to significantly reject the null?

We saw this in the R-R debacle from the right. We saw this in the response to the R-R debacle from the left. And we're going to see this in the OHS study now from both sides. This is exactly the kind of evidence-based research we need, but science and statistics are governed by nuances and caveats that journalists and pols blow over with a hammer.

And we - as informed citizens or what have you - know that correlation doesn't imply causation. But we seem to think that causation implies correlation. And hence with basic logic deduce if there's no correlation, there's no causation. But a simple model can show us this just isn't true. It just means there's a third, significant, variable at play (or there's no correlation: one of the two). And journalists should emphasize the importance of finding this variable, were it to exist. In my post I also tease out a model in this situation where causation doesn't imply correlation. It's unlikely to be so simple, but informative.

I have more thoughts here: http://ashokarao.com/2013/05/02/medicalmdown/

In retrospect, I was too cynical (as I note) but it really is bothersome to see careful statistical terms like "significant" abused. Everyone who comments on this study should be required to read Wolfers and Stevenson (http://www.bloomberg.com/news/2013-05-01/six-ways-to-separate-lies-from-statistics.html).

This also, really, spells out the importance of theory and prior empirics undergirding belief in a particular association.

Very important to note responses from those on the right like McArdle are very good and don't do any injustice to the nuances. But her readership (especially for so long an article) will be selected. Will the op-eds that appear in WSJ be as careful? Will politicians be careful in Townhall meetings?

I guess I'm reacting to an old problem.

Ashok, this is an excellent point. What is the third variable upon which the core elements may pivot on ? I would rashly suggest that it is likely diet which both the paid and the unpaid ingest that is the hinge upon which many health issues swing. Perhaps when the food chain is understood as well as the Tobacco business, then we might be able to start preparing to move on to the next stage of human evolution.

Focusing our efforts more towards public health measures will help more than focusing on catastrophic care. Of course there needs to be measures implemented to help someone in case of emergencies, but the truth of the battle is that your health lies in your hands and is directly based off of what you put into your body. By teaching these methods and explaining the value they have we will decrease the demand for health care, thus causing less expenses for taxpayers and our government. Economically, it is far more feasible to teach a man to fish, rather than to give him one you caught when he is hungry.

'where you would expect proper health management to make a difference'

'Health management?' Why is it that every other 'health management' system implemented in the industrial world delivers better results, as measured by such things as life expectancy or infant mortality?

The goal of providing health care to everyone in a society is not about 'health management.'

A fascinating study, as an example, would be compare the NHS to the American system at this basic level. Which makes it conventient that the Economist reported this recently -

'LIVE long and prosper was Mr Spock's salute, but the interesting thing about Americans is that, prosperous as they may be, they do not live as long as many other nations. Males live around four years less than their Swiss counterparts, and females 5.5 years behind the Japanese. And the gap is widening. British females had a lower life expectancy than Americans in 1980; now they can expect to live 1.6 years longer. British males have gone from being roughly level to being more than two years ahead. (As we note in this week's issue, men are closing the longevity gap with women, thanks in significant part to a big drop in smoking rates. But as we note in our innovation package, despite all the medical innovation, American longevity improvements have slowed.)'


In other words, those who live in what many Europeans consider a fairly inferior health care system (a prejudice which does tend to have the advantage of being empirically based) currently live longer than Americans.

It isn't about 'health management' - it is about changing society so that everyone can be certain that they no longer need to worry about losing a job to retain health insurance, or being bankrupted by medical bills.

Which, not so coincidentally, also increases personal freedom in a Hayekian sense.

This is touching on that Karl Smith post on here a few weeks back regarding the tradeoff between background and personal risk. Of course, it's not so simple, and less background risk is a good thing.

If anything, this means the poor can move their money into (somewhat) riskier equity markets, sharing in on the big appreciation as well as reducing the risk for entrepreneurs who want to get funding.

How does not having as much debt allow investing in wall Street by the poor?

The doctors getting paid by Medicaid instead of selling off the paid bills for 5 cents on the dollar might let them drive up the price of Apple and Google shares without creating any jobs or growth. We haven't seen any hint of the rapid rise in share prices creating a boom in new businesses - the 93% just keep getting poorer and still have lots of debt to pay down or default on, so how many new widgets can the 7% carry in their pockets - they seem to be trying to empty their pockets and switch to smaller handbags and restore their cases to brief from cargo.

Besides, reduced depression will reduce the shopping to relieve depression, which might be better at reducing debt defaults, but it reduces GDP. I guess we need to know if reduced depression reduces shopping by more that $1200 to result in a net reduction in GDP.

Your comment falls nicely under the aforementioned "efforts to distract our attention toward other questions".

As you wish - but all of the other systems, to whatever extent they do or do not practice 'health management' as defined in this context, are better on basic empirical grounds (we will leave the subjective aside - most Americans having no experience of a different health care system anyways).

'Maybe 'health management' is not actually the perspective one should adopt, much less piecemeal methods which can not reflect the clear benefits found in the health care systems of all other systems in the industrial world, from completely government owned (NHS) to essentially private (Germany or France, in contrast).

In other words, is the very idea of 'health management' part of what makes the American health care system so clearly inferior to the others when based on such basic statistics as life expectancy or infant mortality (that is, what many outside of the U.S. consider a fundamental perspective of health, the alpha and the omega of what humans want from life).

But let me end with a quote - 'If there is any cheery lesson it is that, in relative terms, macroeconomics is in better shape than we had thought!' Well, if your measure of macroeconomic success is the worst health care system in the industrial world, and arguments about ineffectual ways to improve it (cash payments is particularly amusing, by the way - why didn't any other country with better health care ever think of that idea?) - though admittedly, this view of macroeconomics being in better shape is as uniquely American as the country's inferior health care system.

You are a scoundrel and a rogue. You should know your argument is a heap of lies and has been thoroughly debunked. http://www.forbes.com/sites/aroy/2011/11/23/the-myth-of-americans-poor-life-expectancy/

The U.S. healthcare system is well known to be near or at the top of the world in every meaningful metric- i.e. success at treating specific ailments. As the link shows, removing accidental deaths from the picture and focusing more (still not completely) on treatable events, U.S. life expectancy is tops in the world. Life expectancy happens to be a measure that is not correlated with healthcare quality among developed nations because it is mainly determined by lifestyle. I challenge you to show me evidence linking life expectancy to healthcare system quality among developed nations.

Cliff, I just have to say that your first sentence is classic. I envision you taking off your glove and smiting PA on both cheeks.

Yes, the US has the highest delays in getting health care of any developed nation, the greatest amount of rationing of health care of any developed nation. Unless you are over age 65 or can get a job with an employer that provides a great health benefit, like Microsoft, Google, Apple, IBM, but not for Wal-Mart jobs who get health care based on Medicaid rationing.

Let's remember, Oregon was the State which had the first death panels. An unelected panel of government technocrats decided who got to live or die by deciding who was worth more than the cost to treat them.

'You are a scoundrel and a rogue.'

Noted as a link below - http://www.conferenceboard.ca/hcp/details/health/life-expectancy.aspx

'‘Canada has earned a steady “B” on this indicator for nearly five decades. Japan’s impressive increase in life expectancy has set a high bar for attaining an “A” grade on average in this decade, a standard that only Switzerland has met. Moreover, Japan, which had among the lowest life expectancies in the 1960s, has been a steady “A” performer ever since. Of the five top-ranked countries in the 1970s, only Japan has been able to consistently maintain its “A.” Switzerland dropped to a “B” in the 1990s, but regained its “A” ranking in the 2000s.

The Netherlands, Norway, and Sweden, on the other hand, have all lost their top-ranking status.

The U.S. has also seen a decline in its relative performance. It earned a “C” in the 1970s and 1980s and dropped to a “D” in the 1990s, where it has remained.’

I don't know - Canadians have never struck me as scoundels or rogues, even though now, under their health care system, their life expectancy has increased to be 3 years longer than that in the U.S., since 1980.

You are a scoundrel and a rogue.

No, not really. He's just a classic internet troll.

I looked at the article you cited in Forbes.

The article cited oncology survival rates, particularly that the Americans were good at treating prostrate cancer. Most males around the world have some PSI. However American medicine has historically intervened more quickly in prostrate cancer and hence includes a healthier population, hence higher cure rates. The death rate for prostrate cancer from the entire population is about the same between European and American societies.

However, I do think that American oncology is probably best in the world. But I have also read that the British Health System has better outcomes with heart disease. And I have lived in Japan and American doctors working in Japan say Japanese gastro is much better,

"in the industrial world delivers better results, as measured by such things as life expectancy or infant mortality?"

Glad you framed it this way. Finally a little honesty. I'm sure most readers here know using infant mortality and life expectancy as measures is very poor science. I would like you to acknowledge the US does lead the world in health outcomes from medical care. You need to compare apples to apples, and, of course, the inputs as well as the results.

I certainly understand that the US leads the world in outcomes from medical care, but that this is too often money thrown away in the last 30 days of life. Dan criticizes prior' above because he goes off track looking for something effective in a ... non-surgical setting (or an alternative to life-long pill-popping).

I criticized it for being off-topic. It is not really a comment on the post at all. It just recycles some of the most worn-out talking points in favor of government healthcare, ironically in response to a post about people wanting to change the topic of discussion.

Sure, lots of other stuff matters. But the performance of Medicaid is what is actually being discussed here. Why not say something interesting about that?

Tyler offered 3 policy suggestions himself, and did limit his text to "medicaid is bad."

Paging dr. Heckman, paging dr. Heckman...

My understanding is that the Japanese in American and Japan have similarly long lives, so it would be a leap to attribute differences to "society".

I almost took you seriously until you started with the 'life longetivity' scam. Oh please... can't you do better?
A friend just had to go for a 10 minute operation (biopsy) in local anesthesie; just to find out what the problem (tumor) might be. He waited 5 weeks. Europea country; waiting times for MRI are 1 month and more.
Another friend just told me: and this is short compared to England (NHS), Canada...
How long does it take in the USA?

If you have health insurance, not that long. If you don't, never.

That's not true. You can pay out of pocket for an MRI. you can get MRI's everyday if you wanted, as long as you paid for them. You can go to any diagnostic imaging center you wanted and schedule an MRI.

so it is not "never". It may be costly, but it is not with held from you just because you don't have an insurance card.

Honestly, lots of places with MRIs are completely unable to handle cash payments, just because they've never had to do it, and see no reason to start.

This isn't neither here nor there for public versus private, though.

I didn't say cash, I said out of pocket, like paying for it on your own not though an insurance company.

The broader point is that health insurance is not the same a medical care. If you are rich enough, why even get health insurance? Just pay a doctor to do house calls, get all the scans and procedures you need, whenever you want. It's all available and no bureaucracy to stand in your way.

If you don't have health insurance, it's tougher, but it's definitely not never.

It's not just that they don't have a cash register -- they have no way of even knowing how much to charge you for an MRI. They don't know how much the product they are providing costs to the people who are getting it.

Fair enough, but generally you're not going to get in any faster paying cash, and if you have neither cash nor medical insurance, you're shit out of luck.

In England and Canada, everyone is covered. So even if you're dirt poor, and even though the wait may be longer, you'll get your MRI if you need it.

Of course you can pay cash for an MRI in your insurance does not cover it. But people without insurance frequently have less income and need to pay groceries.

Why do you think MRIs are useful? Or as useful as the high price in the US implies?

CT and MRI scans cost far less in Japan due to price controls which ironically made Japan a leader in design and manufacture of CT and MRI machines, as manufacturers fought to deliver lower cost, not lower price, but lower cost, machines. The Japanese machines require less power, less space, less skilled staff, operate faster, are easier to interpret, and have lower prices so unit volumes are higher.

By the way, I and my brother and sister have genetic driven tumors which let's me compare use of imagining. My sister had multiple MRIs to diagnose, while my primary got excited and ordered am MRI, which was ignored by the specialist who relied on ultrasound to understand the tumors - he said the MRI was overkill and not really that good at imaging tumors. I did have a study of blood flow in my neck and head, but that was just standard X-ray sequences with dye contrast injections - they used an MRI for my sister, with dye contrast injections. My sister spent several days in the Johns Hopkinton hospital after her surgeries, while I was released 32 hours after I was .admitted for the 6-8 hour surgery at Dartmouth's Mary Hitchcock.

Also, removing these tumors is elective and risky so you want one of the few teams with experience with this type of surgery, unless the tumor is compressing the vagus nerve and you can't swallow or speak clearly. My brother waited months while his employer's insurer denied approval and then cancelled the policy for all employees, and then waited months for doctors and hospitals worked with the State of Kansas to force the insurer to pay some part of the cost and charity care to pay for it, and my brother was working and had employer health benefits. His tumors were no longer elective because they were impinging on nerves. I worked for a global corporation with Medicare quality coverage - no rationing of standard elective care.

Bottom line is lots of health care is rationed in the US and when the rationing has impact on health, the only hope in the US is government intervention because charity care focuses on spreading limited resources around to as many of the needy as possible. Charity for research is different, but that money is spent mostly on basic research that benefits everyone, but by funding the research that makes the best case for connection to the specific disease.

"The goal of providing health care to everyone in a society is not about ‘health management.’"
Well, if you ask the first countries to do it (Austrian Empire for example), it was a way to maintain control, and help prevent revolutions.

"For instance should we revise health care policy toward greater emphasis on catastrophic care, or how about toward public health measures, or maybe cash transfers?"

Yes, but to get the depression-fighting benefits, shouldn't we also award more than catastrophic care to only some people by lottery that will allow those chosen few to get free placebos?

I'm curious about the benefits of insurance covering catastrophic care, but then I remember that, for a Medicaid family of four the current max income is about $32K. At that income, a $100 invoice is a catastrophe.

I guess their cable bill is a catastrophe every month then. I do wonder about the value of catastrophic coverage for this income level though, when they could probably declare bankruptcy and just not pay it.

Without coverage will they even be treated? Hospitals aren't stupid.

This also assumes that, assuming they do get care, they'll then be like, "oh ok I'll declare banckruptcy, nbd. Not problematic at all."

If I were to suddenly present you with an liability way beyond your ability to pay, would you just shrug it off because you can just go chapter 7?

Why do you think they have a $100 cable bill every month? Many have their electricity and telephone shut off because they can't pay the $100 a month those cost. Do you think they pay the cable bill even though they don't have electricity to power a TV or computer?

How else can you callously disregard the actual lives of the working poor if you can't assume all sorts of inaccurate things?

Dude, if you are paying $100/month for your cable, you are getting hosed. That you apparently do so makes me question your basic intelligence.

The big-gulp ban was not brilliant, but let's drop it in for context. Americans basically want big gulps and to live to 150 too. To make that work they need a lot of invasive medical care. The right, if they want better outcomes should at least advise against the big gulps ... as they ration the care.

I made the comment on the old thread, but there probably are serious depression-fighting effects from knowing that your catastrophic health care costs are covered.

Now, there are probably much cheaper ways of achieving that same depression-fighting effect. But just because the OHS showed one good increase in outcomes doesn't mean we should disbelieve that outcome.

First Head Start and now Medicaid. We have ZMP government programs to go with our ZMP unemployed.

I'd love to see the optimistic estimates put into the bills themselves, as triggers for automatic sunsetting of the entire program if we don't make it to the promised land. That kind of precommitment would be more persuasive and moderating of the rhetoric.


Perhaps you would like to be enrolled in my randomized controlled test in which we take persons with health insurance and have them live for two years without medical care or insurance. That way we can tell whether this is ZMP or not.

Care to join the test?

Who's withholding medical care? This is solely about medical insurance. Most of the dishonestly of this issue is the conflating of the two.

Re: Who's withholding medical care?

If you live in a market economy, and money is required to purchase medical care, and you have no money, I suppose you could say "Who's withholding medical care?", but you sound pretty silly saying it.

Who says we live in a market economy for medical care? Hospital emergency rooms are compelled to provide care and then see if they can get paid later. There are not masses of poor people dying in the streets because they sought care and had it withheld. If there are people who need catastrophic care (transplants, cancer treatments), the solution isn't "free" birth control for all.

Oregon was the State that first created death panels of unelected government technocrats who decided who was worth the cost of keeping them alive.

Boy did conservatives attack them for that.

My guess is they used the lottery rather that creating a new death panel to decide which thousand people deserved to live and condemning the remaining thousands to suffering and death.

(A list of treatments was ranked by cost-benefit, and then a line drawn at the limit of the Medicaid budget, with treatments below the line denied by Medicaid. The same was true in NH, but the list was not created in public, with the same kind of outrage over a case in NH as in Oregon, and a lot of money raised to fund futile treatment of a kid with cancer, which was denied by NH Medicaid based on the cost of care for one kid that probably would fail would cut vaccinations etc to thousands of other NH Medicaid kids.)

For what it's worth, I'm taking Dan Ariely's MOOC on Irrational Behavior right now. In the week on self-control he cites studies that show that while pre-school students do not necessarily show better cognitive and social-emotional development, they do show better life outcomes. The explanation offered is that pre-school teaches self-control.

1) A lot of people are saying the study is bad news because it "proved" that Medicaid did not affect health. No. The point estimates for almost all the outcomes measured indicated that Medicaid helped--the effect was positive. 10,000+ people is a relatively large study, but only a fraction of the beneficiaries had the conditions (e.g. diabetes, high cholesterol) that the study actually tracked. The one condition that showed a significant effect at the arbitrary .05 threshold was depression. That is also the condition that had the largest sample size. The sample size matters. Simple. Though apparently too nuanced for the mainstream media to accurately report.

2) One alternative being thrown around, giving cash, is totally infeasible. Tell me what merry group of legislators is going to propose giving poor people even more cash right now. Medicaid is the lowest cost and most efficient way to pay for health care. Why would one think that poor people would be able to negotiate better rates than state agencies with huge market clout?

True, Medicaid can probably negotiate better rates with providers than an individual. However, individuals with cash acting in their own best interests would optimize their satisfaction or health care utility. Over the long term, it would seem that providers competing for the cash of individuals would provide better value per dollar spent. Surely, less would be spent on administration.

Theoretically that is. Where is the empirical evidence to support this claim. Does any industrialized society use this type of system?

I don't know of any industrialized society using such a system. Too bad, it would be fun to see a study of health outcomes in that type of system vs Amercan or European style systems. I wonder if the gains in consumer satisfaction would translate to objective health measures.


The U.S.

It's called the Earned Income Tax Credit.

And it's widely acknowledged to be the most successful form of "welfare."

ding ding ding

I don't think I really understand. Medicaid isn't the truest form of insurance, but it does protect people from financial burdens of catastrophic events as well as expensive, chronic conditions. Giving people cash would remove the whole "insurance" part of it.

And despite the complaints from providers about the low Medicaid reimbursement rates, you can bet that they would be opposed to a cash transfer program. Not because they're afraid that empowered poor people would negotiate even lower payments than Medicaid, but because they probably just wouldn't spend the money on routine or preventative care. I realize there is a debate about whether routine and preventative care is even useful, but from the providers' perspective a cash transfer program in lieu of Medicaid would suck.

Give each person $1200 and they will be free to exercise their 2nd amendment liberty in buying a semiauto, big clips, and a 1000 rounds and Go Aurora.

Or would the NRA claim that they support background checks at gun shows to deny liberty to people getting $1200 from the government to spend more wisely than government?

Many people claim "a program for the poor is a poor program", but EITC doesn't seem to get much opposition.

Republicans in Michigan, where I'm from originally, are working hard to get rid of the state EITC. There are lots of other states where this is happening, too. There are also plenty of people who oppose EITC at the federal level. Keep in mind, converting Medicaid to cash in most cases would mean giving the beneficiaries much _more_ money than the average low-income person receives from federal EITC. Seriously, it's never going to happen.

The whole "47%" meme is built on the EITC.


First, I'm glad to see that you've actually read the study. Seems like most people here haven't looked at the paper: http://www.nejm.org/doi/pdf/10.1056/NEJMsa1212321

I'm glad you mentioned sample sizes. Indeed, depression is one of few variables that has a p-value of less than 0.05, but its not the only one. There are 26 variables measured in the study. Of those, only six have a p-value of <0.05. I've listed them below with the numbers from the study (mean value in control group vs. mean value in group with medicaid coverage; p-value)

Diagnosis of diabetes after lottery (1.1% vs. 4.9%; p<0.001)
Current use of medication for diabetes (6.4% vs. 11.8%; p=0.008)
Positive screening result for depression (30% vs. 20.9%; p=0.02)
Diagnosis of depression after lottery (4.8% vs 8.6%; p=0.04)
Health same or better vs. 1 year earlier (80.4% vs 88.2%; p=0.02)
SF-8 subscale (???) Mental health component score (44.4 vs 46.4; p=0.05)

If you increase the p-value threshold to 0.10, you include:
Current use of medication for high cholesterol level (8.5% to 12.3%; p=0.10)
Current use of medication for depression (16.8% to 22.3%; p=0.07)

If you increase you're p-threshold much higher, you're getting into numbers that can't be considered trustworthy. I'd say that all you can reliably take from the study is the following: More people got diagnosed with diabetes and started using medication for it if they had Medicaid. Fewer people with Medicaid registered as depressed from a survey, but having Medicaid made you more likely to get a diagnosis of depression (and use medication for it). People with Medicaid felt better about their health, and had better mental health. They were more likely to use medication for high cholesterol.

I'm confused as to how these items can be construed as bad. The study did NOT test all 10000 people for diabetes and only tell those with Medicaid that they have the diagnosis. It's relying on people going to their doctor, and you're more likely to go to the doc, get a diagnosis and take the medication if you have insurance. To conclude that having Medicaid causes worse health based on this data is ridiculous!

I think the relevant question should be what you can conclude from the test of a diabetes or heart medication drug after two years of an RCT experiment. I am not an expert but I would imagine, not much.

For example assume two people with the measure of interest of 100, one on medicaid and the other one not. Let's say when they get the disease the measure starts growing at 5% per year and lets assume you can reduce this growth rate in half if take medication or use medicaid to treat it. After two years their levels will be 105.1 and 110.3 which is less than 5% between them. After 20 years the difference is about 38% so could be very significant. After 30 years its over 50%. So if your level of significance is 5% you not expect the two measures to differ after 2 years. But that is not too informative. Obviously you can play with the numbers so that it's not significant even at 1%.

Add on top it the fact that some people in the sample don't get the disease at all, that some who do, still get medication but have to sell their house or borrow money and you get very low power of the test.

What we really need to see is whether Medicaid is best way to achieve the benefits it has (i.e. saving people on out-of-pocket expenses for catastrophic cases).

"I think the relevant question should be what you can conclude from the test of a diabetes or heart medication drug after two years of an RCT experiment. I am not an expert but I would imagine, not much."

You don't understand how at least some of the metrics work. Hba-1c is a measure of plasma glucose that tracks developments in blood glucose over the last few months up to the time of the test. You can see very significant variation in that metric in much less time than the duration of the study, and such differences are meaningful and important metrics when it comes to evaluating metabolic control. Hba-1c is not a metric that somehow accumulates over time - indeed the whole point of treatment is to keep it at a stable level 'relatively' close to that of non-diabetics. The farther you are from that level, the higher the risk of complications, and 2 years is plenty of time to evalute differences in treatment outcomes - they could have done that in 6 months or less if they'd have liked to.

Regarding the blood pressure metric and cholesterol level estimates, the timeframe of the study is also more than long enough to identify potential treatment effects.

Actually, considering that the study is only 2 years on and most of the subjects of the study probably were enrolled and initially evaluated over the course of several months, and it took several months to finalize their diagnosis; most have probably been treated for about a year. Now maybe we should continue the experiment for another 10 to 20 years so that the effects of these chronic diseases have a chance to manifest themselves. Then we can really evaluate whether Medicaid insurance, and its associated health care, really made a difference in people's lives.

I'm sure they picked two years to be on the safe side - and if they'd used 6 months instead it's fair enough to say that other stuff might have been going on as well (I'd probably have said some of the same things you're saying). But my point was also mainly that 2 years is plenty of time, and that the time-frame should not be a strong point of criticism (certainly not from a cost-benefit standpoint - i.e. considering what these kinds of studies cost).

The reason why the time frame is more than long enough is that we already know how Hba-1c differences tracks differences in outcomes of diabetics - the connection between microvascular complications and Hba-1c is strong, it's very well known, and it has been an established fact for years if not decades; just as the connection between blood pressure and stroke risk, and hypercholesterolemia and cardio-vascular risk is very well known. They didn't just randomly include variables in this study - these three variables are very informative in terms of prognostics.

The important point is that if the estimate after 2 years is truly indicative of the average 'lab value differences' that exist _in equilibrium_ between insured and uninsured individuals, then there isn't really a strong need for a follow-up. We know reasonably well what is likely to happen (on average) to given individuals for given lab values. It's only if the lab values change over time there's a strong need for follow-up - otherwise we'd just be confirming what we already know; that diabetics with poorer metabolic control are more likely to go blind, get a limb amputated, suffer kidney failure, experience nerve damage, etc.; that people with hypertension are more likely to die from stroke and cardiovascular disease, etc....

I'd have been happy to have seen more than a point estimate (if they only did one follow-up, again I've not yet read the study), but in this context the lab values are where the main level of uncertainty is - it's not at which outcomes might be expected from given lab values. We know that part pretty well.

The blood sugar and blood pressure numbers are getting a lot of attention here, but there are other factors to consider. These new enrollees are mostly very low income single people. They are mostly living on SNAP benefits for food, since housing eats up 50% - 70% of the income of someone making under $1,000 a month. Lean meat and fresh fruit and vegies are the best diet for these conditions, yet they are stuck in the mostly carb $3 - $4 per day diet.

Many people have had insurance for 20 years and are still having trouble with their lab values. It seems like this is a case of giving something to poor people and demanding unusually high success from them.

Let's run the counterfactual from this:

"Medicaid coverage significantly increased the probability of a diagnosis of diabetes and the use of diabetes medication, but we observed no significant effect on average glycated hemoglobin levels or on the percentage of participants with levels of 6.5% or higher."

So, from a two year study you conclude that it would have been better not to have diagnosed diabetes? And, you conclude that from a two year study that there was no significant effect on glycerated hemoglobin levels. What was the control? Do we know levels from those under private insurance that are diagnosed?

And, what is the consequence later of those who were not diagnosed?

I think we can conclude that americans are good at ignoring the advice of doctors and are getting fatter all the time.

The consequences of not getting diagnosed vs being diagnosed and ignoring the doctors recommendations are not really that different.

Re: "The consequences of not getting diagnosed vs being diagnosed and ignoring the doctors recommendations are not really that different."

???? That means medical care should not be available to people because of their income?

" That means medical care should not be available to people because of their income?"

No, it means that if you are going to ignore medical advice on your diagnosed condition, you might as well not have been diagnosed.

eno, What evidence do you have that this group differs from other groups in their propensity to respond to medical advice. There was nothing in the report that covered this, nor were there controls for other populations. Is this your imagination???

Can't speak for eno, but I don't think it is a characteristic unique to any group within the US.

It is very american however.

If group A and group B both ignore treatment advice at equally high rates, an analysis of the effectiveness of a policy that promotes giving treatment advice is probably not going to be very interesting.

My starting point is Americans do very unhealthy things across the board and any policy is going to have a very hard time getting past that.

I'd say the 6,5% limit is problematic - in Denmark roughly 90 % of diabetics are above that level, and the number is probably if anything higher in the US. A more informative limit taking better advantage of the data at hand (increasing power) would probably be if they were above or below 8.0% (the median Hba-1c for diabetics in Denmark - again, this is probably not too different from the US numbers). In my opinion you'd need really huge effects for it to be statistically significant at 6.5%.

But in terms of health it'd be easy and to some people make sense to argue that even though Hba-1c is unaffected by diagnosis, getting the diagnosis is still valuable. Diagnosis means follow-up on the disease state, which means that complications (nephropathy, polyneuropathy, retinopathy, cardiac complication, diabetic foot ulcers, ...) can be addressed - for example regular eye screening of diabetics with relatively poor compliance might still save many from going blind from their disease. Or you could argue from the point of view of the cost-efficiency guy who might argue that identifying complications in patients with poor compliance should not necessarily be a focus point because it eats up resources that could have been used elsewhere in the medical system (and that it might be cheaper to just wait with the diagnosis until they present at the ER with a stroke).

I haven't read the specific study in question yet but I've seen quite a few similar ones. Regarding the control group, the whole point is that they're presumably comparing the outcomes of relatively similar people, some of whom (semi-)randomly (with lotteries it's not completely random because it's not totally random who decides to participate in such lotteries in the first place - but it's still 'random-ish') got assigned to the insurance scheme, and others who did not.

It's time to be a little stern here, because this is a basic issue of statistical literacy.

If you fail to reject the null hypothesis, you have not really said anything about it. Might be true, might be false. All you can say is that you tried to attack it from one angle, with one set of data, and the attack failed.

You certainly haven't proved the null! This is sophomore-level stuff. I understand when the popular press gets things wrong, but can't we expect economists to appreciate the difference between a study that "did not find an effect" and one that "found no effect?"

The words "mood affiliation bias" drift through my memory, but I can't quite think why...

'It’s time to be a little stern here, because this is a basic issue of statistical literacy.'

Please - it is the authors of this blog that lecture about statistics, not commenters.

The authors of the blog claim to be bayesians. They just moved their posterior a bit from the starting point based on new information. Their priors are just different from yours.

And this kind of evidence is much more difficult for advocates of a policy. Opponents obviously love it. But advocates need to come up with an explanation why the result was found. These kinds of studies are important. Policies are put into place with expectations of results. When/if the results are not appearing you have to evaluate the policy. To carry your logic through, if decades of work never rejects a null hypothesis, we can never say anything about the policy. That is why we are all bayesians.

Fortunately, this is only one study. No need to throw out the bathwater just yet, much less the baby.

Wrparks, let me first say that I subscribe to your last sentence entirely. One paper is one paper. It's positive findings were (a) very modest and (b) indicated that Medicaid expansion was producing some good. That doesn't seem to match the tenor of much of the reporting; that's my point.

I will take issue with your second paragraph. We can certainly say things about the policy, _with a properly designed study_. We just have to select a different null hypothesis and have another go.

Notice how squirrely even Megan McArdle feels she has to be in her article, saying the program "may not make them any healthier..." It may not! Or it may! Which is roughly as much as we knew last week.

> Which is roughly as much as we knew last week.

And yet, last week we were expecting to be able to say we knew much more. Funny how getting new information works.

It may be only one study but as McCardle's article pointed out, the earlier Rand study and the previous UCSD RCT showed equally weak or non-existent results. Any fair minded scientist would have to say that the burden of proof must now be on proponents of more subsidized health care to show that the potential benefits are meaningfully LARGE and statistically significant. Otherwise, why bother running tests? Just say you believe the US have universal health care regardless of costs or benefits.

Who's the Bayesian, Tyler, Alex or both? I sorta thought one had to actually use Bayesian methods to be a Bayesian. That's like saying I side with western medicine over eastern medicine, thus I am doctor. It's amazing how 5 years ago I had trouble publishing Bayesian stuff and now it seems as everyone has always been a bayesian.

I think at this point it is safe to say that 'the authors of this blog' haven't the faintest idea of what the computation of either frequentist or Bayesian stats entails.

And you seem to be just as ignorant. In fact, Bayesian analysis suggests an _increase_ in the prior estimates of the efficacy of Medicaid. You're 'their priors are just different from your' goes to the heart of why no one listens to libertarians who claim to be Bayesians - it's just a lazy and dishonest way of freighting 'I disagree with you' with far more weight than an unsupported opinion deserves. You're free to disagree of course. But if you want anyone to think you're actually a gunslinger, you damn well better support your position with some, y'know, numbers. This 'Bayesian analysis confirms my prejudices' without without any numbers on your estimates just won't cut it.

I think the big pieces of information out there are (1) the fact that many European countries spend half as much on healthcare as the US and appear to get the same or better health outcomes, (2) the RAND study (basically little to no benefit to more healthcare), and (3) the Oregon study (same). They all seem to be pointing to the troubling conclusion that health care does not actually make people any healthier. The scientific implications are more troubling to me than the policy implications (the money was going to be wasted on something else, why not healthcare?).

For me this lowers my confidence in all medical care and in doctors compared to nurses (who can do the basic things that do seem to work, like giving vaccines and antibiotics, closing open wounds, giving pain medication, etc). It also raises my estimate of how important things like clean water, good sanitation, pest control, childhood nutrition, and genetics (e.g., impulse control, general intelligence) are to health outcomes.

You do realize that the USA subsidizes the rest of the world via their health care system don't you? Do you think it's an accident I can get brand name drugs in Greece for less than in the USA, and get treated for less using the same technology? Same in southeast Asia. The drugs and methodology in the USA are adopted, without IP protection, by the rest of the world. The anti-patent Alex T's would prefer the USA also become a medical industry pirate, but then you would not have any reason to invest in healthcare. Certainly the average government physician wage of about USD $34k in Greece and $15k in southeast Asia (not including moonlighting, which everybody does) would not be sufficient to induce people to go into medical school in the USA, unless they love helping people.

Tell me are the Greek and Southeast Asian physicians saddled with tremendous debt when the finish school or was the schooling paid for by the state? Secondly, how many physicians graduate each year from their schools? And what is the doctor patient ratios? Let's hypothesize that the US medical system creates these problems by making it a market based system and restricting the number of practitioners to keep their compensation high.

I think ant1900 has hit the proverbial nail. Money spent does not necessarily achieve better outcomes, and a recent example for me shows this anecdotally. I have a great health plan thanks to my pension. My co-pays are minimal. I injured my groin muscle area doing rehab from another injury. (This is not uncommon -- even controlled and directed exercise can lead to injuries.)

I read quite a bit about predicted recovery times and for my condition it was bleak -- 4 to 8 months of discomfort. But I wanted to see if something could be done. A full array of testing (including two sets of MRIs, two sets of x-rays, cortisone shots) did nothing. This experiment, which I thought might have produced something positive, ended up costing my insurer about $8,000.

I will now wait and see what time does. Would I have spent this money if I was underinsured? Maybe -- my wife and I have a good income.

But I can tell you that many treatments have only the most marginal effects. Catastrophic care coverage, if done properly, would lower costs significantly.

Archie Bunker wants to know if you went to the groinycologist.

I agree with Tyler. It looks like ACA will not do as much to actually improve health care as some thought and hoped. This should not be too surprising since the focus from the beginning was about how to pay for giving more people health insurance, not how to get better value for money from total health care spending. But whatever whoever thought in 2010, the issue is how do we improve the efficiency of health care spending and how would changes in ACA move us in that direction. In the long run expanding the scope of the system before improving its efficiency may still turn out to be the best, perhaps the only feasible path to reform.

I would just point out that there was a great effort--by some--to include in the ACA more research to understand what health care interventions work best, and what their cost effectiveness is. This provision remained in the legislation as PCORI, but only after the very notion of comparative effectiveness research was much maligned by its political opponents and the program was watered down at the request of legislators who nonetheless voted against the whole bill.

Yep, thinking at the margin is sorely needed. But isn't part of the challenge that the growth in government has been precisely because the wisdom of providing adaptive, flexible safety nets (in general, the equivalent to thinking at the margin) has been hijacked by the obligation to provide growing positive rights. Is one thing to say: Why not the positive right to a safety net. Is is another thing when the safety net is an escalator to heaven fueled by unobtainium.

Why would we expect access to health care to affect conditions largely determined by diet and exercise?

When has ideology ever cared about statistics?

The ACA is a political gesture. Costs, benefits and results were ignored in the rush to pass it.

The quip by Krugman is particularly silly, wrongheaded and unhelpful.

He was being flippant, but his better response is here: http://krugman.blogs.nytimes.com/2013/05/02/medicaid-nonsense/

"Somehow, conservatives think this is a big win for their opposition to universal health insurance. Why? What [the study] suggests is that the health benefits of ANY kind of health insurance are somewhat hard to identify over a two year period; so, are you about to give up your own insurance, or is your best bet that having that insurance is still a very good idea? And the financial benefits are a big part of that! Since you are going to treat your illnesses, better not to bankrupt yourself in the process, right?"

Thanks. Better, but still not particularly illuminating. He is not fairly characterizing the results of the study. It's not that the health benefits "are somewhat hard to identify." It's that the study strongly suggests there may in fact be none. And, yes, I would like to give up my low-deductible insurance in exchange for catastrophic insurance and the ability to fend for myself for non-catastrophic events.

He's speaking about the statistical significance, which is separate but related to the clinical significance. See, e.g., the comments from Austin Frakt & Aaron Carroll on the subject at The Incidental Economist. http://theincidentaleconomist.com/wordpress/additional-thoughts-on-the-new-oregon-medicaid-results/

"3) You don’t understand statistical significance!!!

I assure you I do. When your point estimate is clinically meaningful but your results are not statistically significant it usually means that the variability was larger than expected, there really was no effect, or you were underpowered to detect the difference. See (2). I can’t tell which of these are true because I don’t know if the study was powered to detect the point estimate differences they found."

Strange I don't hear calls for elimination of the tax exclusions for employer-sponsored health insurance now that we "know" that health insurance subsidy is a waste of government money.

You are not listening Alan (or have covered your ears). Eliminating this deduction (and permitting employees to take the increased wage and purchase whatever insurance they think best out of their own funds) has long been an element of conservative discussions of this issue.

I'm just as dubious of the benefits of low-deductible insurance and, given the general good health of me and my family, would be thrilled if my employer would pay me the cost of my insurance and permit me to purchase a high-deductible catastrophic policy (and pocket the savings).

Most employees would be forced to buy insurance at a higher price than the "increase" in wages, especially if you lost access to the low risk pool. Remember the government is currently SUBSIDIZING employee insurance purchase.

I think it is far more likely that a relatively few employees would be affected in the way you suggest. Certainly not most. Subsidies would be in order for them. It is one of the infuriating (to me) aspects of this effort that vast changes are advocated for what turns out to be a small number of affected parties.

"Getting rid of the tax quirk" is much different than "getting rid of employer-provided insurance."

I would like, in the long-term to get rid of employer-provided insurance. Not immediately, because it's an excellent way to get access to risk pools. And getting rid of the tax imbalance won't destroy them, either.

I can't imagine employers just turning over the difference in costs to employees, anyway. How many employers do that now if you decline their insurance?

What is striking to me is the utter logical disconnect in some of the above comments (see Prior Approval for the perfect example)- lambasting the clear results of the study, while at the same time pointing out, with great approval, that Europeans spend half as much for better health outcomes.

Then I say fine- if we want to spend half as much for pretty much the same outcomes, we should just spend half as much, right??

'lambasting the clear results of the study'

Except for all the other commenters pointing out that the results are not exactly 'clear,' in part because of the small sample size.

'while at the same time pointing out, with great approval, that Europeans spend half as much for better health outcomes'

I am not a perfect example of that, since at no point did I talk about costs, referring only to empirical data. However, it is most certainly true, empirically, that America pays significantly more for an inferior health care system.

The point of having broad health care available to essentially all members of a society is that the society itself is healthier, when using empirical measures, data measures which now have time spans stretching over decades. I have no idea why a limited test is considered more relevant than the experience of literally hundreds of millions of other humans, in a variety of different health care systems (cheaper ones, as you pointed out, though that was never my point).

In other words, the rest of the world seems to understand the 'macroeconomics' much better than the U.S. - which is the only industrial nation that continues to sink further in the ratings concerning life expectancy -

'Canada has earned a steady “B” on this indicator for nearly five decades. Japan’s impressive increase in life expectancy has set a high bar for attaining an “A” grade on average in this decade, a standard that only Switzerland has met. Moreover, Japan, which had among the lowest life expectancies in the 1960s, has been a steady “A” performer ever since. Of the five top-ranked countries in the 1970s, only Japan has been able to consistently maintain its “A.” Switzerland dropped to a “B” in the 1990s, but regained its “A” ranking in the 2000s.

The Netherlands, Norway, and Sweden, on the other hand, have all lost their top-ranking status.

The U.S. has also seen a decline in its relative performance. It earned a “C” in the 1970s and 1980s and dropped to a “D” in the 1990s, where it has remained.'


So basically, you should take all your conclusions and turn them on their head, because the U.S., using real data as I linked to above, has the best outcomes in the world. So why isn't the rest of the world learning anything from the U.S.? Europeans spend less and get less.

What best outcomes? Seriously - you can quote Forbes, and I can quote the Conference Board in Canada, using international comparisons. One of those sources shows that life expectancy in Canada has increased compared to the U.S,. while the other source claims special circumstances to show why the statistics in America are somehow skewed.

Really, what are you going to believe - international comparisons of normed life expectancy rates or special pleading for why the U.S. isn't as bad as the ongoing data indicates?

I don't have a horse in this race, I might add - I have no need to concern myself with American health care, and actually have a guilty feeling observing the ongoing trainwreck from afar. A fact that fills me with a sense of something other than satisfaction, a feeling that expats living in civilized industrial nations understand, even if they can't explain it to their American compatriots, try as they might. As demonstrated, again, in this comment thread.

It isn't about 'health management.' another one of those peculiar American concerns, it is about ensuring a basic level of health care for every member of society. The macroeconomic benefits which can be seen when looking at other industrial societies covering literally hundreds of millions of humans, with empirically better results, regardless of the details of the health care models they use.

Breaking apart a big group into sub groups isn't "special pleading." Otherwise you land right in Simpson's Paradox.

It requires a few more neurons to understand, yes.

'Breaking apart a big group into sub groups isn’t “special pleading."'

So, the entire rest of the world is somehow different from America? Because that was the point of the Forbes article, assuming you read the link, of course.

Yes, america is a very unhealthy place compared to much of europe. We eat poorer, exercise less, and generally are proud of this it appears.

More of us actually need health management and education to prevent the problems more than we need health care improvements. By the time we get to the health care portion, the damage is too great.

Apparently as an expat, you have been gone too long. Much of our health wounds are self-inflicted.

Also, I think one can easily infer what the authors of the study thought of the results by the fact that their publication was safely postponed from last Summer.

As to disproving that you can't get good health care with medicaid; Ezra is wrong the case is not disproven. 10,000 patients whose doctors know they are being watched closely may get much better care than typical medicaid enrollees. Doctors are human and are more diligent when they know they are closely watched.

i dont know why you keep bringing up reinhart & rogoff.
this has nothing to do with that.

What I find amusing is that most of the commenters downplaying the study and fretting about how opponents of more health care will simplify and abuse the results don't note all the supporters who were trumpeting the early results of the uncompleted study as showing The Science Has Spoken! (see the Reason article that Tyler linked to.) And even the Obama Administration was quick to feature those early results. That puts a special burden on supporters to rethink their position and even to tolerate opposition distortions. After all, a purely sober discussion will not erase the propaganda benefits of the earlier extreme triumphalism. So this more than turnabout is fair play.

I never though I would say this but you guys need to go into a major depression for a while.

When you recover you might have a different take on a program that was found to "decreas[e] the probability of a positive screening for depression (−9.15 percentage points; 95% confidence interval, −16.70 to −1.60; P=0.02)."

I don't like the way (some) people here are laughing that off, but if the point of Medicaid is to improve depression, there are probably much cheaper ways of doing it.

More guns (suicides)?

Nope. Same suicides in UK way fewer guns.

Access to health care is a meaningless discussion when access is available. What isn't available it the ability or skills of individuals to care for themselves. The middle class isn't healthy because they have insurance, it is healthy because the same skills that gets them to the middle class, impulse control, caring for issues when they are small, good habits, etc., will also keep them healthier.

Here we have universal and free health care, but we still have people who fall off due to lack of basic life skills. If you have diabetes, you need impulse control, high levels of discipline. The local mental health folks find that 90% if their resources are consumed by a very small number of individuals, so they are changing their focus from crisis management to setting up systems where these individuals are helped to take their medicines, which then avoids the 3 week hospital stays a couple of times a year to get stabilized.

Exceptions occur, so having some hard and fast solution that would apply to everyone is silly. Government is a blunt instrument, so we would expect silliness. And statistical measures of health results are in my opinion of limited value. By definition a health event is an outlier, and the numbers of individuals in a population that have any specific health event is small, so you start getting into statistically insignificant measures. An individual needs this, the question is whether they are getting it or not. We are not talking about vaccines and sewage handling.

Yes. Also, note that the skills required for economic success in a bureaucratic modern society are also the skills necessary to receive and benefit from modern bureaucratic medicine.

Health problems keeps a lot of people out of the middle class. I personally know a number of intelligent, honest, hard working people who are poor because their health problems prevent them from showing their potential to the labor force. And since they're poor, they cannot afford proper treatment. The worst is when a person is too sick to work at all, and cannot recieve disability benefits (and Medicaid) because they don't have the money to get their conditions properly documented to qualify for disability. I've seen firsthand it bringing multiple people to suicide ideation.

Long term vs. Short Term.


Apples and Oranges.

Any other errors here?

A few thoughts:

1. General comparisons of the US to other countries ("The US spends more on healthcare than Country X but has worse outcomes. Therefore, it must be the fault of the US health care system.") are meaningless. "American" is a nationality, not an ethnicity. Valid comparisons take these ethnic differences into account (for example, compare Norwegians to Norwegian-Americans, Japanese to Japanese-Americans, Africans to African-Americans, Mexicans to Mexican-Americans, etc.) In almost every case, the hyphenated Americans come out on top.

2. Think that countries with socialized healthcare systems do it better? Like, maybe, the UK? Well, survival rates show that the cancer care in the US is a lot better than in the UK:

Cancer Survival Rates
USA vs. England
90.5% Breast Cancer 78.5%
69.9% Bowel Cancer 51.6%
66.3% Prostate Cancer 44.8%
62.9% All Cancers (Female) 52.7%
66.3% All Cancers (Male) 44.8%.

3. Compliance is essential. Seeing a physician on a regular basis to care for your diabetes or hypertension, doesn't do much good if you don't take your medicine and follow his or her lifestyle advice (lose weight, stop smoking, etc.).
Look at this:

At the end of the eighteen-month trial, the doctors told me that the treatment was successful and that I was the only patient in the protocol who had always taken the interferon as prescribed. Everyone else in the study had skipped the medication numerous times – hardly surprising, given the unpleasantness involved. (Lack of medical compliance is, in fact, a very common problem.)

Cancer survival rates are meaningless. Very early detection of cancers means treatment and reporting for many cancers that would not have killed you. Prostate and breast cancer are good examples of cancers that are often non-fatal (at least for the 5year survival period) if left untreated.. BTW, the UK is not the whole world of government sponsored health care.


So the U.S. is better at detecting these cancers early?

You say most would survive with no treatment, but in the UK over 20% of breast cancer patients die. Over 50% of prostate cancer patients die! Most would survive with no treatment??

Results are comparable with other countries.

Be careful. Better detection can improve survival rates while not changing outcomes one whit.

I also want some more data about "50% of prostate cancer patients die." If you are an elderly man in America you probably have prostate cancer, and you probably aren't going to die from it. Dying of prostate cancer is incredibly rare even if you have it.

I said very early diagnosis inflates the number of cancers and improves reported survival rates that are based on a specified time frame (often 5 years). PSA testing is not so routine in the UK and fewer "benign" prostate cancers are recorded and treated. Cancer screening is proving to be largely ineffective at reducing mortality for many cancers. Mammography has similar issues to PSA testing for men. (Note: today the American Urological Association has formally withdrawn the recommendation for PSA screening for men under 55).

Seems like this study is a good illustration of the difference between "not guilty beyond a reasonable doubt" and "innocent." An acquittal is not a delcaration of innocence.

So...interesting to note that the study actually recorded substantial improvements in all 3 measures for which Fisman claims there was no effect. The study was small enough that you can't make statisically supported conclusions and at 95% confidence interval. So you can't go around saying the study proves Medicaid is effective and be intellectually honest. But to use the study to support the opposite conclusion of no effect is even more so.

From the paper itself:
"Nonetheless, our power to detect changes in health was limited by the relatively small numbers of patients with these conditions; indeed, the only condition in which we detected improvements was depression, which was by far the most prevalent of the four conditions examined. The 95% confidence intervals for many of the estimates of effects on individual physical health measures were wide enough to include changes that would be considered clinically significant — such as a 7.16-percentage-point reduction in the prevalence of hypertension. Moreover, although we did not find a significant change in glycated hemoglobin levels, the point estimate of the decrease we observed is consistent with that which would be expected on the basis of our estimated increase in the use of medication for diabetes"

Yeah this study really doesn't say what people seem to think it says.

What else is new?

I think that one of the things that this study did was to separate the economists from the statisticians.

Kelly Ayotte being in the news and one of my Senators is on record opposed to background checks as not useful for reducing murder, but what is needed is better mental health care to prevent gun violence.

The Oregon study proves that expanded Medicaid improves mental health. Obamacare will present a natural experiment it seems of half the population seeing expanded Medicaid, and half not, so we can see if
1) Medicaid improves mental health on a larger scale as the two year result suggests
2) the NRA and Ayotte are correct that improvements in mental health will reduce gun violence

Did the Oregon study look at the rate of gun violence between the two populations in the study group??

If expanded Medicaid works better than gun control as used in the rest of the world to reduce gun violence (even the Bush administration believed in gun control IN IRAQ to reduce gun violence) then shouldn't the NRA supporting Republicans support Obamacare's Medicaid expansion to reduce gun violence?

Actually it was critics of Obamacare that claimed it would 'ruin healthcare' or 'put the gov't between you and your doctor'. If that was the case, one would expect that people who entered Obamacare (or a portion of Obamacare) would have different health outcomes from those who didn't. Little noteable difference indicates that a major criticism of Obamacare was wrong.

Words here are a bit deceptive. "Healthcare reform", strictly speaking, would be reform of actual health care. Looking at the laws around what doctors and nurses are allowed to do, how research results are disseminated, what medical schools teach, etc. But that's not what people mean when they talk about healthcare reform, what they are really talking about is better described as "Health financing reform" or "Healthcare Payments reform". But let's face it, we don't speak German and "Obamafinancingreformofhealthcare" simply doesn't roll off very well.

So we see the actual impact on healthcare itself seems mildly positive, which is nice but not really required for health care financing reform. If tomorrow your employer announced he found a way to lower the cost of your health insurance 5% but there would be absolutely no difference in what's covered, participating docs, copays etc. you'd be happy. But the employees actual health wouldn't be expected to change except in the very mild way that improved finances releaves a bit of stress which has some mind-body benefits.

What really should be measured here is not health outcomes but financial ones. Are hospitals that treat Medicare patients doing less cost shifting to people with private insurance or soaking the poor saps who are actually willing to pay cash for hospital services ($250 aspirin coming up!)? Do Medicare patients have a lower risk of financial distress from healthcare? Check, they do. Does this cause Medicare patients to go hog wild getting excessive treatments thereby creating costs that didn't exist before? It would seem it doesn't.

Yes it is worth asking is it better to just give a poor person a cash stipend rather than Medicare coverage. But this neglects the fact that if the new Medicare patients didn't radically change the amount of treatment they were getting, then either they were paying for it directly or someone else was paying for it. If they were paying for it directly then coverage is effectively giving them more cash. If someone else was paying for it then a good chunck of the benefit of the expansion is to unload the burden from them. A real look at the costs.v.benefits of the program then must incorporate those players as well.

My hunch is that on balance the expansion was a net benefit you have:

1. Minor health benefits to those who got the expansion.
2. Real finanical benefits to those who got the expansion.
3. A decrease in uncompesenated care which benefits health suppliers who write off some amount of care to the poor.
4. A real decrease in cost shifting, esp. onto those with coverage or those paying OOP for healthcare which helps move towards what many conservatives say they want to see in the health system, more honest pricing so consumers and the market can make more use of the information contained in price signals.


"Tyler Cowen links to a couple of writers today who say the recent study of the Oregon Medicaid experiment was bad news for Medicaid fans because it showed that Medicaid coverage had no effect on most of the medical conditions that were studied. His summary:

[start Tyler quote] Do read the rest of those posts for a more complete picture of the results, but many commentators are overlooking these rather simple upshots. [end Tyler quote]

It's disappointing to keep reading this stuff, because it's flatly not true. There are three main takeaways from the study:

There were positive results on some measures (depression, financial security, rates of diagnosis).
There were also positive results on all of the other measures studied (blood pressure, diabetes, cholesterol).
But the size of the study was too small to determine if the positive results in #2 were real. This says nothing at all about Medicaid. It just says that, unfortunately, the experiment was too small to be definitive."

Some awfully hard spinning by Drum on point # 2 Barry. Wouldn't the fairer characterization be: "We cannot state with any confidence that there were positive results for any of the other measures studied."

As I pointed out above, Medicaid is mainly about healthcare financing rather than health care itself. Imagine two poor people who double over in a mall with a heart attack. One has Medicaid and the other has no insurance. Would we really expect the outcomes to be all that different? In both cases people will call 911. In both cases cops or paramedics will come to do cpr. In both cases they will be brought to an ER.

Many of the positive measures (BP, diabetes, cholesterol) that were found to be 'too small' all happen to share the following:

1. They tend to track improved health over the long term. Better BP today than you had a month ago is unlikely to make that much of a difference to your help, but better BP over the next ten years is.

2. The types of health care costs and outcomes they target tend to be much longer term. Avoiding massive bypass operations ten years from now saves a huge amount of money as well as lost resources (reduced work, etc)

Yes, but doesn't my point still stand? In other words, "We cannot state with any confidence that these indicators, which if they moved in a positive direction would save a huge amount of money, in fact moved in a positive direction."

When you say 'save a huge amount of money' are you talking from the standpoint of healthcare finance reform or healthcare reform? In terms of healthcare yes it would save a huge amount of money if we found a way to have less cancer, diabetes, heart disease etc. Aside from a magic pill that instantly cured those things, though, any such policy change would probably require decades to really notice. For example, you could probably demonstrate that the costs of lung cancer today are many billions of dollars cheaper because of all the anti-smoking efforts we made from the 1980-2000's.

If you mean healthcare financing reform then the study isn't really looking in the right area. As I pointed out you would need to look in other areas too like unreimbursed healthcare incurred by hospitals, less cost shifting onto the private sector etc. None of those things would be captured by this study.

Poor folks got poor ways. It's not at all surprising that standard measures of health either don't budge or barely budge even if the poor are observed and treated. That's because you can't treat the underlying problem (for now at least) with medicines. And that's nothing new.

I'm a lawyer and represent many large and long standing American manufacturing companies. In the course of representing them in latent disease cases I've reviewed decades of medical surveillance reports (in one case beginning in 1917) and have seen the various training and education programs the companies established. Many of the people hired early on (admittedly cheap labor hired for that very reason) were shockingly poor and illiteracy rates sometimes exceeded 50%. Not perhaps surprisingly one of the earliest issues the companies faced was the spread of lice (and other parasites) along with bacterial and viral pathogens throughout the workforce with the attendant loss to productivity. What the companies uniformly realized was that treatment alone didn't help much and so they initiated training and education programs that we'd find shocking today (and which would probably get the companies sued).

In the old records of those companies you find training courses on personal hygiene including sexual hygiene, how to safely prepare food, what to do when you give birth at home, the perils of alcohol and the like. From there the programs expanded (these from one of the great oil companies born at Spindletop) to things like "How to be a good Husband", "How often should I bathe?" and my favorite "Why it's important to have Books in the Home". They even offered after-work remedial reading courses.

Early in my career I was lucky to get to talk to some of the old guys (by then in their 80s) who'd started working in the 1920s as engineers or management. They were the ones who told me "poor folks got poor ways". From their perspective the best thing the company ever did was not paying good wages but instead, as one put it, "civilizing the worker". They had training courses on marriage, on how to dress, on how to balance a check book and even how to dance at a square dance. By the 50s they had a pamphlet called "Your Son may be College Material" and a driver's ed course for wives. No doubt it was self-serving but the company's efforts to eliminate the causes of workplace downtime and inefficiencies by changing behaviors was a success and in the bargain changed lives for the better.

Look up any of the big mortality studies on the huge plants in East Texas or Southwest Louisiana and you'll see reductions, relative to others in the county (or parish) or state, in heart disease, stroke and diabetes well beyond what the healthy worker effect would predict. My hypothesis is that it's not because there was a plant doctor on site to salve burns, flush metal specks from eyes and perform annual CBCs and prostate exams; nor because they had great insurance that allowed them to see any doctor they wanted. It's because they were, like many who go through the military, given the gift of ordered, which is to say, disciplined lives. Those leading such lives get examined when they ought, take their medicines as prescribed and exercise as directed.

Thus my take is that it wasn't just the $$$ that lifted so many into the middle class - it was also the effort to change behaviors, to eliminate or prevent "poor ways". Likewise, until we address the "poor ways" of poor folk we're unlikely to significantly affect their health status however much money we spend on healthcare.

So. America, generally described as the richest society in human history by Americans, has 'poor ways'- well, that is a novel explanation for why American health care is so poor.

We are not as wealthy as we think we are.

Re: "Likewise, until we address the “poor ways” of poor folk we’re unlikely to significantly affect their health status however much money we spend on healthcare"

Isn't your presumption that introducing the poor population to doctors and the medical system will not address the poor ways of poor folks. Introducing persons to doctors and healthcare providers DOES change poor ways; you even say so in your piece, if you look carefully, at your comments about the factory doctor.

This is just pointing and spluttering by you. The factory doctor treating the employee and his family in the company town is a very different dynamic. I'm as fond of feudalism as any bigoted reactionary, but I'm guessing most folks here are not.

The basic point remains: unless the government's social workers are going to prod people out the door for their daily walks, cook their meals, stub out their cigarettes, and ration their alcohol, all socialized medical care will get you is ... more consumption of socialized medical care.

Whether we should provide all these wonderful things as a moral obligation is a different issue; probably, we should. But it's Econ 101 that any attempts to socialize the cost away from the consumer will lead to distortions, manifested in things like riskier behavior and over-consumption.

Anti, The post very clearly stated that the intervention by the factory doctor changed the behaviour of the employees. You may call that sputtering, but you are ignoring intervention. If you want to be rational, you should be comparing the intervention and the non-intervention state in terms of physician influence. As someone noted above, even if you intervene on diabetes, for example, the blood sugar may not go down--even in middle class groups--but it may not go up, either. So, step back, think.

Can we talk? Thanatos is describing a much different interaction than we have today. His descriptions were of men of worth being patriarchal and civilizing the masses over several generations. Our society lacks the will to be quite so demanding of our precious citizens. No judgment, no standards. A company doctor today is influential in getting a pain killer prescription and starting the paper trail so you can claim disability some day, not telling the dirty masses to man up and act like a decent person. God forbid we hurt anyone's feelings.

+180, this is unbelievably spot on

The Incidental Economist(s) link to your piece favorably and expand with some thoughts of their own


They've other posts on this study, as well.

How can you ignore the fact that depression dropped by 30% !!!! isn't that huge. Think about this, how can someone who is poor and gets Medicaid lift themselves up and move up the economic ladder, if they are depressed?

Some other links, courtesy of Brad DeLong:


Aaron Carroll covers concepts like power, sample size, etc.


The commenters cover the estimates and such.

Oh, it looks like Tyler left off some things from Josh Barro

"Still, the “I knew Obamacare was a waste of money” reactions are misplaced. The financial effect is a big deal. Having Medicaid reduces your likelihood of facing medical expenses that exceed 30 percent of your income by 80 percent. This reflects that Medicaid is, in large part, a redistributive income-support program, which is desirable given persistently high unemployment and widening wage gaps. The government should be taking steps to support people who are being left behind by the increasingly asymmetrical economy, and this is one of them."

"The depression finding is important too. McArdle is skeptical, since depression went way down without anti-depressant use going up much. “Does the mere fact of knowing you have Medicaid make you less depressed?” she asks. It’s plausible to me that the answer to that is “yes,” since financial strain can be a cause of depression."

Heck, here's the rest of the article:

"I think it makes more sense to read the study as an indictment of medical insurance broadly, rather than just one of Medicaid. If Medicaid is spending lots of money without clear improvements in physical health to show for it, isn’t it likely that much more expensive private insurance -- which we can’t do a random-assignment study of -- is doing the same? The study is yet another argument for “some health care for all, but not too much.”

And the study tells us one thing that should make us quite hopeful: Individuals seem to possess a lot of information about which treatments are the most useful for improving their health. People without health insurance consume less health care but get physical health outcomes that can’t easily be distinguished from those on Medicaid; that suggests they are prioritizing and getting the care that is really important.

The discovery mechanism that’s used now is unacceptable: We’re forcing people to ask themselves the question, “Is this medical treatment so important that it’s worth going bankrupt over?” But maybe we can come up with other structures that divine the same information in a less punitive way.

All of this makes me wish even more that conservatives had been productive partners in health reform rather than trolls. If conservatives want a consumer-directed redesign of the U.S. health-care system that forces patients to pay at the margin more often for care -- in order to reveal what treatments are useful -- they could have gotten it as part of the health-care overhaul. They just also had to agree to include the progressive fiscal reforms that liberals wanted: ensuring universal coverage and transferring money toward poor people who can’t keep up with the rapidly rising cost of health care.

They didn’t, partly because Republicans care more about not spending money on poor people and not changing programs that old people like than they do about making the health-care system more efficient. They also would have entered a political minefield. Nobody likes being told they’re getting too much health care. Just look at the bipartisan political outcry at the U.S. Preventive Services Task Force recommendation that women start receiving mammograms at 50 rather than 40.

I’d love to see this study shift the conversation on health reform toward how to use government and market mechanisms to ensure that money goes toward care that is actually useful. Obamacare includes a number of policies toward this end that could surely be improved, including by increased consumer direction. But given the right’s consistent non-constructive approach toward health policy, I’m not holding my breath."

I agree with Tyler that these results have very little to do with Medicaid, but rather with the efficacy of medical insurance in general. I think the takeaway is that there is very little relationship between health and medical insurance. The same study found no relationship between consumption of medical services and health. We will learn more as the study continues (they don’t have enough to start looking at mortality yet), but we do have enough to conclude that, as long as we insist on using insurance as our tool to improve health, we will have to spend vast sums to move the needle a tiny bit.

Zen and the Art of Motorcycle Maintenance had a great disquisition on how, when we encounter a problem, we jump too quickly to remedies. In his case, confronted by the problem “motorcycle won’t go”, he had the carburetor disassembled before finding that he was out of gas.

There is a similar effect in education. The great secret of the educational establishment is that there is a very small school effect. The vast majority of variance in school achievement is accounted for by other factors, the strongest being mother’s education level. Yet our response to failure is to redouble our misguided efforts, throwing good money after bad.

The drug war is another obvious example.

I thought of Ivan Illich:

Many students, especially those who are poor, intuitively know what the schools do for them. They school them to confuse process and substance. Once these become blurred, a new logic is assumed: the more treatment there is, the better are the results; or, escalation leads to success. The pupil is thereby “schooled” to confuse teaching with learning, grade advancement with education, a diploma with competence, and fluency with the ability to say something new. His imagination is “schooled” to accept service in place of value. Medical treatment is mistaken for health care, social work for the improvement of community life, police protection for safety, military poise for national security, the rat race for productive work. Health, learning, dignity, independence, and creative endeavour are defined as little more than the performance of the institutions which claim to serve these ends, and their improvement is made to depend on allocating more resources to the management of hospitals, schools, and other agencies in question. Ivan Illich Deschooling Society(1973: 9)

You really don't know anything about statistical analysis, do you? But I'm more than willing to be convinced otherwise: Tell me, if you want to go with the p-value approach, how many coin flips would it take to show a coin biased 51% in favor of heads if you went with p=0.05?

You making such an authoritative pronouncement and all, I'm sure you can tell me in seconds (it's an extremely simple problem.)

Tell you what, instead of cutting medicaid, let's get rid of the tax exclusion for employer provided health insurance, and stop giving members of congress health insurance while we're at it. How does that sound?

'Zen and the Art of Motorcycle Maintenance had a great disquisition on how, when we encounter a problem, we jump too quickly to remedies.'

If there is one thing certain in this debate, the problem of the inferior American health care has nothing to do with jumping too quickly, or finding remedies.

How about something like ... when people are diagnosed with hypertension, diabetes, etc, direct them toward workshops entitled "If you have condition X, and actually want to do something about to to improve your quality of life, free public workshops are available to learn and share knowledge on lifestyle decisions and medical strategies to effectively managing your condition" , in 5 words or less, if possible.

Comments for this post are closed