Sunday, September 29, 2013

How to Avoid Defaultmageddon: Randomize ObamaCare


ObamaCare for some, 
miniature American flags for others

Things in Washington are getting pretty tense. On October 1st, the ObamaCare healthcare exchanges are set to go live. October 1st is also the date on which, unless the Congress can pass a continuing resolution for appropriations, the government will go into shutdown. Also looming on the horizon is another debt ceiling fight.

The confluence of these events has led to a substantial amount of political brinksmanship, with some Republicans demanding that ObamaCare be defunded or delayed in exchange for agreeing to a CR and/or debt ceiling increase. For their part, both Obama and most Congressional Democrats have stated that they will not negotiate over the debt ceiling, and that implementation of ObamaCare should go forward as planned.

The two sides seem irreconcilable. But as it turns out, there is a way for both sides to claim victory. Are you ready? Here it it:

Congress should delay ObamaCare AND allow it to go forward.

More specifically, Republicans and Democrats should agree to randomly split the states into two halves. In the first group, ObamaCare will go forward as planned; for the second, implementation will be delayed indefinitely.

It's important to note here that the two groupings must be truly random. I'm not proposing that ObamaCare go into effect in blue states but not in red states. It's quite possible that ObamaCare would go into effect in Mississippi but not in Minnesota. In fact, you would want a mix of red and blue states in both groups.

Randomizing implementation of ObamaCare would have several key advantages.

First, it would help us learn more about the true effects of the law. Right now, many conservatives are convinced that ObamaCare will make the healthcare system worse, whereas many liberals think it will be beneficial. You might think that, once the law does go into effect, this debate will be settled. But as the continuing debate over the effectiveness of the ARRA shows, this is not necessarily the case. If the implementation of ObamaCare is followed by higher prices, a decline in healthcare quality, or plagues of locust, those on the left can (and probably will) claim that these bad things would have happened anyway, and in fact would have been worse if not for the ACA. If prices fall, quality improves, and research reveals that eating banana splits cures cancer, then those on the right can (and probably will) claim that these good things would have happened anyway, and in fact would have been better if not for the new law.

Turning ObamaCare into the world's largest randomized controlled trial solves this problem. Instead of having to guess counterfactually about what would have happened if ObamaCare either had or hadn't been enacted, we could just compare the changes in healthcare in states with ObamaCare to those without.

Because of this, randomizing ObamaCare ought to be an acceptable compromise to both sides of this particular argument. Granted, supporters of ObamaCare will at first only get to see it enacted in half the states. But if the law turns out to be as beneficial as they claim, then the popular opposition to the law will fade, and the electorate will soon demand that the law be extended to all fifty states. Likewise, if, as conservatives claim, ObamaCare is a disaster waiting to happen, then it won't be long before those suffering under the law look at how much better their non-ObamaCare neighbors are doing, and will demand full repeal. The more certain each side is about the effects of the law, the more willing they should be to go along with this compromise.

UPDATE: In the comments, "Waffles" suggests that denying ObamaCare to half the country would be unethical

[I]t's pretty undeniable that the subsidies provided by the law are going to be a huge help to Americans who aren't as well off as yourself. There's a reason that there are pretty high ethical standards for human experimentation in the social and medical sciences, and your suggestion is way, way, way beyond that ethical line.

Of course, RCT's routinely do precisely what Waffles says is way over the line, that is, giving potentially life-saving subsidies to one group while denying them to another. The Rand Health Insurance Experiment, for example, did this, as did the more recent Oregon Medicaid RCT. In fact, one can read entire books recounting the results of this type of experiment (incidentally, my wife's job used to be doing compliance for human experimentation, and she confirmed that Waffles' comments were off base).

But while Waffles' objection is factually wrong, his/her objection to the ethics of RCTs is hardly unique. I'm reminded here of an old anecdote from Dr. E.E. Peacock, which I'll pass along via Marginal Revolution

One day when I was a junior medical student, a very important Boston surgeon visited the school and delivered a great treatise on a large number of patients who had undergone successful operations for vascular reconstruction.
At the end of the lecture, a young student at the back of the room timidly asked, “Do you have any controls?” Well, the great surgeon drew himself up to his full height, hit the desk, and said, “Do you mean did I not operate on half the patients?” The hall grew very quiet then. The voice at the back of the room very hesitantly replied, “Yes, that’s what I had in mind.” Then the visitor’s fist really came down as he thundered, “Of course not. That would have doomed half of them to their death.” 
God, it was quiet then, and one could scarcely hear the small voice ask, “Which half?” 

53 comments:

  1. Anonymous1:01 PM

    Would be great, if all states were similar. But California isn't Wyoming, and a random selection process could put California, Texas, new york, etc on one side. Since the markets are handled state by state, this could be pretty distorting

    ReplyDelete
  2. So stratify/match states according to some appropriate axes (income, level of inequality, red/blueness, other socioeconomic variables ...)

    ReplyDelete
  3. Anonymous1:26 PM

    Right, but the blog post was arguing it should be random.

    ReplyDelete
    Replies
    1. I meant that one could (for example) create pairs of similar states, then randomize assignment within the pairs.

      Delete
  4. This is an acceptable compromise for pundits, politicians, and economists who are wealthy enough to be largely unaffected personally by the law. However, there are millions of people for whom the implementation of this law will make a big difference in terms of personal finances and dignity. There is some controversy about the effectiveness of cost-cutting provisions and the employer mandate, but it's pretty undeniable that the subsidies provided by the law are going to be a huge help to Americans who aren't as well off as yourself. There's a reason that there are pretty high ethical standards for human experimentation in the social and medical sciences, and your suggestion is way, way, way beyond that ethical line.

    "The more certain each side is about the effects of the law, the more willing they should be to go along with this compromise."

    Not true at all. I think it's very likely that the law will benefit many people (especially those in relative poverty), and that delaying the law (even in just half the states) in an orgy of false equivalence would be a hugely immoral act. These are people we're talking about, not lab mice, and what happens to them matters. A negative outcome here is more than a data point on a spreadsheet. I'm sorry if my rhetoric is a little harsh, I'm just amazed at how detached this idea is from the real world effects it would have.

    ReplyDelete
    Replies
    1. Anonymous3:14 PM

      Let's randomize it by people instead! There will be no social cost to having my co-worker get $7,000 of subsidies a year while I struggle to pay my student loans!

      Delete
    2. Waffles,

      To be clear, you're saying that if someone proposed giving one randomly selected group of people a subsidy for healthcare while not giving it to another randomly selected group, this would be rejected as unethical?

      I take it you don't know much about randomly controlled trials.

      Delete
    3. Except the health care law is a comprehensive reorganization of the healthcare system in which the subsidies are a single part of a larger whole. This isn't randomly denying a child a second marshmallow, this is randomly denying people an entire set of institutional fixes to problems that we know to exist in the unaltered system in the name of pseudoscience (but really, ultimately, convenient politics). Also, talking down to people who are more educated than you realize is a dumb way to win over an audience. I take it you don't know much about running a platform for discussion and debate.

      Delete
    4. Also, I completely do not understand your argument that people who are more certain of the law's success should agree to reduce the number of states it is being implemented in by half. The more sure I am that the law will work, the more this seems like an unnecessary infliction of harm on the less well off to prove a point. If I'm convinced the law will be a success and improve peoples' lives, then since I care about everyone's well-being, I will want everyone to benefit from it.

      Delete
    5. Under the assumption that "the law turns out to be as beneficial as they claim," a supporter of the ACA will see grave human suffering as the mechanism behind the electorate's inevitable "demand that the law be extended to all fifty states." I see no reason for a supporter of the ACA to prefer that outcome over one in which the ACA is simply implemented in full, under the assumption that the law is expected to be a success.

      Delete
    6. Anonymous6:36 PM

      I wouldn't accuse Waffles of not knowing much about random trials if you propose something that, as others have pointed out, has such an obvious flaw as the California Wyoming problem.

      Healthcare is not a mystery, it's not some open question that hasn't already been solved by every other civilized country. It has nothing at all to do with "conservatives think" and everything to do with what vested interests want.

      If you really think our political process is full of good-hearted people acting in good faith, differing only in means based on their honest-to-goodness beliefs about how the world works, I don't think you're really qualified to be blogging.

      Delete
    7. This is pretty amusing. Here's a line from the Medicaid RCT document you posted, page 2:

      "Although designing a randomized study with a no-insurance control group would normally raise serious ethical concerns, recent circumstances in Oregon have created a rare opportunity to do just that."

      It seems that the principal investigators of the Oregon health care plan lottery also "don't know much about randomly controlled trials" by your definition.

      Delete
    8. Josiah, lose the attitude. Waffles is pretty much right. The RAND study started in 1971. I was in grad school in the 70's; the human-subjects regulatory regime was nothing like today's. Furthermore, the study compared various insurance co-pay levels; no participants were deprived of insurance. Harmful effects of high out-of-pocket costs were found in that study. An ethical review board would tend to reject a study that would repeat a known-harmful treatment. So the fact that the RAND study was done is itself a reason that a highly similar study may not be approved today.

      The random assignment in the Oregon study was not made by the researchers, it was done by the state government to allocate an insufficient Medicaid fund to otherwise qualified persons. As it was not done for research purposes, the lottery was not subject to ethical review.

      Also this, from the Republican Study Committee's proposed alternative to ACA:

      SEC. 442. REPEAL OF FEDERAL COORDINATING COUNCIL FOR COMPARATIVE EFFECTIVENESS RESEARCH.
      Effective on the date of the enactment of this Act, section 804 of the American Recovery and Reinvestment Act of 2009 is repealed.

      http://rsc.scalise.house.gov/solutions/rsc-betterway.htm

      I don't think your proposal is going anywhere with that crowd.

      Delete
    9. @Waffles: "Also, I completely do not understand your argument that people who are more certain of the law's success should agree to reduce the number of states it is being implemented in by half. The more sure I am that the law will work, the more this seems like an unnecessary infliction of harm on the less well off to prove a point. If I'm convinced the law will be a success and improve peoples' lives, then since I care about everyone's well-being, I will want everyone to benefit from it. "

      Well, um... what to say to this? We are in a situation of extreme division, where it is unclear *any* part of the law will get implemented, given the shutdown, the threat of default, etc. Obviously, Josiah is suggesting a way out of these impasses where each side gets something of what it wants. Now, maybe it is not a good compromise, but you really can't see why half of Obamacare followed soon by all of it might be better than none of it?

      Delete
  5. Anonymous2:10 PM

    Agreed, and not to mention that a randomized trial does not limit either side from claiming confounds.

    ReplyDelete
  6. You live in a world of abstractions. Waffles nailed it. And You.

    In case you don't know it, the Republicans have zero interest in finding realistic policy solutions to real world problems. All they are interested in is making Obama fail. Well, that and tax cuts. Have you been paying any attention to politics the last 5 years?

    Your entire framing is that the electorate will vote in their own self interest. If that were the case, the Rethugs would not have a majority in the house, or the Governor's houses in WI, MI, and OH.

    In all seriousness, this post is just pitiful.

    Noah, please hurry back. The B team is not cutting it.

    JzB

    ReplyDelete
    Replies
    1. Some of them are OK. This particular author is quite subpar, however--if I wanted to read Megan McArdle, I'd go to her site and not accept imitations.

      Delete
    2. "Noah, please hurry back. The B team is not cutting it."

      Translation: "Team B sometimes does not feed us straight Democratic Party talking points. Please hurry back so we can get that again!"

      Delete
  7. Lindsay Wilcox4:21 PM

    Wow, Josiah. I think the more "You're an idiot!" name-calling you attract, the more likely it is you're on to something. It's so juvenile it's funny!

    ReplyDelete
    Replies
    1. Anonymous6:56 PM

      Wow, Jazzbumpa. I think the more "You're juvenile!" name-calling you attract, the more likely it is you're on to something. It's so idiotic it's funny!

      Delete
  8. Before differential treatment, all subsamples do not come from the same population since 23 states and the District of Columbia have agreed to expand Medicaid under Obamacare.

    ReplyDelete
  9. Rothosen8:57 PM

    If people would quit using the word "subsidies" and use "raise tax on everyone else" perhaps the dialogue could move forward. There is no money in Washington for "subsidies." Haven't you been watching the debt ceiling debate? For these mythical "subsidies" to occur taxes have to be raised. Unless of course you're in the 'we can borrow unlimited amounts' crowd. As for "every civilized country" we are talking about Greece, Italy, Spain, England, & Japan aren't we? The final bill always comes due and we are getting to see what happens when it does.

    ReplyDelete
    Replies
    1. Yep. And we talk about "capital investment credits" or "agribusiness subsidies", let us also remember that those are covered by taxes too, OK? The bill does always come due. Thirty years of tax cuts and deregulation have come due in lowered employment, increased inequality, and hyperpartisanship. Yeah, that worked out for us.

      Delete
  10. Mexico rolled out a massive new health care policy in a randomized fashion. It avoided the ethical issues because they needed to roll out it gradually for logistical reasons anyway.
    Here are some of the studies that were performed.

    In the US, delaying it means thousands of preventable deaths (at least according to ACA supporters) making this ethically dubious.

    ReplyDelete
  11. Josiah, why limit your study to but one country. Why not expand it to the entire developed world. Countries like Britain, Western Europe, Australia (where I come from), Canada, will have Medicare-for-all and Single Payer, government run health care. Study the results these systems achieve in terms of coverage, health care and costs. Then compare those results to The Most Wonderful and Glorious Free Market No Government Health Care system in the United States. Draw conclusions from those differences, reconsider your priors,...actually better not, you might not like what you find.

    ReplyDelete
  12. Anonymous11:51 PM

    An ethical randomized trial requires clinical equipoise, i.e. a genuine uncertainty as to which arm of the study is best for the subjects. To the extent that supporters and opponents are sure of their opinions, they certainly would not see equipoise and therefore could not ethically agree to randomization. Perhaps the original proposer of this social experiment is one of the few who truly don't think they know the effect of the law. Second, in addition to the oft-noted problem of randomizing among non-standardized entities (ie the Calif-Wyoming problem), there is the problem of blinding. If people know which state they are in, and can observe effects in other states while the experiment continues, then they can change their behaviors in reaction, thus completely undermining the original design. And finally, this whole suggestion is premised on the belief that carefully constructed evidence will sway opinion and action on this topic. To date that has not been true, with large sections of the very population this will help - the non- or under-insured -- believing that the law will hurt them due to what they hear on some radio/TV/internet sources. Only time and personal experience and neighbors' anecdotes will change attitudes.

    ReplyDelete
  13. Anonymous1:18 AM

    Q: Do you think the opponents of the ACA would accept the results of this (terribly flawed) study in good faith?

    Q: Should I continue wasting my time reading this blog, if this is the level of its posts?

    Both are rhetorical questions.

    ReplyDelete
  14. I don't think Waffles is as far off base as you think. I worked for decades in clinical trials of medications. In a disease that is life-threatening or has severe morbidity, if there is already a treatment known to be beneficial, you CANNOT run a placebo-controlled trial. The Institutional Review Boards would never approve it. You have to use "standard of care" as the control. This raises the bar for new drugs, but that's appropriate.

    Health insurance is known, from lots of research, to be beneficial to people with serious illnesses. The fact that we have been (unethically) denying this benefit to millions does not make your proposed trial ethical. At least that's the way I see it.

    ReplyDelete
    Replies
    1. As the Oregon lottery document he posted notes, "designing a randomized study with a no-insurance control group would normally raise serious ethical concerns."

      Delete
    2. "Health insurance is known, from lots of research, to be beneficial to people with serious illnesses."

      This is not insurance: Insurance protects against some uncertain, unpleasant outcome. If they have a serious illness, there is no risk that is being "insured" against! It is simply a subsidy to pay for their medical bills... and of course, having someone pay some of my (already known) bills is a benefit to me! I am not necessarily against that, but let's call it what it is.

      Delete
  15. Anonymous4:09 AM

    you think the GOP will be convinced by a case study, and secondly it's been signed into law and would be affront to legal equality to have it apply to some people.

    ReplyDelete
  16. Anonymous6:05 AM

    Your assumption that the problem is Obamacare, it is not.

    the problem is whether House majority can control national policy.

    If Obama back down on this, the result will be American transition to Parliamentary system, which House is more powerful than presidency.

    ReplyDelete
    Replies
    1. Peter T12:45 PM

      And what would be wrong with that (transition to parliamentarism)? Important is accountability, that the makers of decisions have to defend them in elections. Accountability seems currently missing, due to gerrymandering, etc.

      Delete
  17. Mock Waffles all you want, but he/she was correct: This would be massively unethical. One of the key bits is informed consent. Are you seriously going to argue that 300 million people would give informed consent to this? Many don't know what the ACA says. Many more only "know" the lies that have been promulgated on the right about it (death panels, anyone?). How do people opt out of this vast experiment? How do you ensure double-blindness? How long should it run? This leaves alone the idea that the states are the appropriate unit of division anyway.

    This began as a neat what-if but your attempt to claim it can be a serious idea exposes you as, at best, naive and worst, mendacious.

    ReplyDelete
  18. Anonymous10:22 AM

    One other problem -- moral hazard. The folks and states who don't like Obamacare would be incentivized to make it fail during the trial period. I don't think this would be offset by folks and states that tried to make it great.

    ReplyDelete
  19. Anonymous11:03 AM

    In a nation where scientific method is considered a belief system, who will be appointed to interpret the results of your experiment? In a nation where evolution is rejected as atheism and the scientific consensus regarding climate change is considered liberal agitprop, interpretation of facts is nothing more than a political exercise.

    ReplyDelete
  20. The Oregon health plan lottery was not a planned experiment put together by social scientists, it was an accidental "natural experiment" due to funding/legislative constraints that allowed the investigators to simulate an RCT. As the authors of the document you linked write:

    "Although designing a randomized study with a no-insurance control group would normally raise serious ethical concerns, recent circumstances in Oregon have created a rare opportunity to do just that."

    If you're going to reference their paper in your defense, please make it clear to your readers that they do not support your ethical position.

    The RAND health insurance experiment, as the Oregon lottery investigators note, is the ONLY example of an RCT on health insurance; there have been none conducted in the 31 years since. I wonder why? Are the comments made by the Oregon lottery investigators also "off-base"?

    ReplyDelete
  21. Please explain to me how this experiment deals with basic ethical requirements for human subject research such as:

    - The right of the subject to voluntary informed consent
    - The right of the subject to end participation in research at any time
    - The right of the subject to access to information regarding research
    - The responsibility of the researcher to end the experiment if it is judged that a continuation of the experiment will result in injury, disability, or death for the subject.

    Your proposed RCT doesn't pass the Nuremberg Code.
    http://en.wikipedia.org/wiki/Nuremberg_Code

    Maybe you should specifically reference the set of ethical guidelines for human experimentation that you are using to make this judgement so that we can see your reasoning.

    ReplyDelete
    Replies
    1. Anonymous1:16 PM

      It doesn't conform to the ethical requirements for human subject research, but the ethical requirements for public policy are resumably different.

      Delete
    2. I disagree. The reason the Nuremberg Code was instituted in the first place was because the Nazi regime used state power to run unethical experiments on unwilling participants. The emphasis on individual, voluntary, informed consent is due to the horrifying lack of such under state-run human experimentation. Political scientists specialize in studying "natural experiments" (like the Oregon lottery) because there is a wide consensus that state-run random trials of public policy are unethical, especially when there is good reason to believe that the policy being tested is important to the health and well being of citizens.

      Delete
    3. The fact that the experimentation applies to a whole society rather than a small group of people should raise the importance of ethical standards of experimentation, not lower them.

      Delete
  22. Would this somehow violate Equal Protection? I'm just curious.

    ReplyDelete
  23. Anonymous4:10 PM

    The easy way to do it would be to allow those states with their own state-based exchange to proceed while any state relying on the federal exchange to 'wait until next year'.

    ReplyDelete
  24. I find it shocking that anybody is even debating this ridiculous idea. The author is suggesting that whenever a political minority loses an issue of public policy, that the policy should be implemented only for half the population, randomly selected.

    By this logic, only randomly selected citizens should be required to pay taxes. And those taxes should be used only to build schools and roads for some of the people, again randomly selected. Then all government programs opposed by conservatives should be cut by the percentage of the population they represent --- about 30% --- so long as the citizens benefiting from those programs are randomly selected.

    And you call yourself a serious thinker?


    ReplyDelete
    Replies
    1. "The author is suggesting that whenever a political minority loses an issue of public policy..."

      No, on this one, polls show that a political *majority* lost on public policy:
      http://www.huffingtonpost.com/2013/06/27/obamacare-approval_n_3511401.html

      Maybe they are deceived, but lets not lie about the situation!

      Delete
    2. Hey Mr Unwilling to Compromise,

      Have fun with the shutdown, hope you enjoy!

      Just think of this, if it goes on long enough, people might start to figure out how to live with less government. Wonder how long it takes, I'm genuinely curious.

      Delete
  25. Wonks Anonymous6:13 PM

    The best randomization in health care story I've heard is one where a bunch of doctors are already performing a procedure which hasn't been vetted by experiment. One doctor plans to do said vetting, but finds that he would be prohibited from experimentally giving & withholding a treatment to further science, although it's perfectly okay for him to perform/withhold treatment in the same exact cases simply at his own discretion! I wish I could find the link now.

    ReplyDelete
  26. Fun idea, but there's another big problem with it that will become a partisan (though valid) reason to claim failure/victory: people will move from state to state. If I'm poor and can make it to a state with a subsidy, I will do so (not unlike immigration from developing nations to developed ones), which then impacts the economic and health metrics for both states.

    ReplyDelete
  27. Anonymous11:33 AM

    "Instead of having to guess counterfactually about what would have happened if ObamaCare either had or hadn't been enacted, we could just compare the changes in healthcare in states with ObamaCare to those without."

    The fundamental problem is that any changes in any direction can always be chalked up to some other factor. For example, if Obamacare existed in Idaho but not Montana, you might see a lot of sick people move to Idaho to take advantage of Obamacare, while the healthy people moved to Montana to avoid paying for it. Or you could argue that Colorado doesn't have the same issues with the Navajo nation that New Mexico has, so that's why they turned out better. And so on - basically, people wanting to argue the issue for non-rational reasons can always find a seemingly rational reason to take their position.

    ReplyDelete
  28. Anonymous5:25 PM

    This one of the most stupid post 1) There is some evidence, based on some model that the subject of the trial is expected to be effective. 2) Everyone start with small scale trial and third and most important one, if new procedure/model/drug whatever seems to be more effective than the previous procedure/model/drug then it will get ADOPTED.

    Every evidence from every where in the planet suggest that the current system is cruel, expensive, inefficient and stupid and obamacare is going to be better but not the current state of the art. We have the small scale study in MA, we have the various European experiences and we know that the current US system is unbelievable cruel and expensive. Why should we subject half of country to such cruelty because some idiots have trouble letting go of their failed ideology and consider real data.

    ReplyDelete
  29. The Oregon medicaid experiment was only allowed because Oregon didn't have enough money to provide for all. It was a natural experiment, not a randomized controlled trial. Denying medicaid to half of the people in state X randomly obviously violated clinical equipose. The Rand study did not do this- it tested various types of insurance, in a situation where the benefit wasn't as clear cut.

    ReplyDelete
  30. Congratulations. You have just invented the HealthCare equivalent of reinstating the randomised draft.

    ReplyDelete
  31. I fully support the idea of a randomized trial... at last some genuine empirical measurements in Economics. Buck up lads, you will be a science yet.

    There are a few talking about government subsidies. Of course, no such thing exists, there are not government subsidies and never will be. What exists is government wealth transfer, where money is taken by force from one group of people and given to another group of people... in order to make an accurate empirical trial it is absolutely critical that in such a test, no wealth transfer can cross between the control group and the other group. The test must be exactly balance sheet neutral. Other than that caveat, great idea.

    ReplyDelete