Something Not Unlike Research

Psychiatric over-diagnosis and collaborative mental health care

post by Bill Gardner

I recently recommended a podcasted lecture by Dr. Allen Frances, the former chair of the American Psychiatric Association's DSM-IV (Diagnostic & Statistical Manual, version IV) task force, on the problem of over-diagnosis in psychiatry. Dr. Howard Brody has a recent post in which he criticizes a proposed revision of the DSM which would treat grief persisting longer than two weeks as a depressive disorder. He also reports an interesting rumor,

I have heard murmurings that if the APA [American Psychiatric Association] proceeds to issue the new DSM-V (apparently scheduled for next year) with this reclassification [of grief], some of the primary care physician organizations, including my own American Academy of Family Physicians, might vote formally to dissociate their members from DSM. This would be an unprecedented move as up till now, starting at least with DSM-III, all medical specialties have been willing to accept as gospel the disease classification proposed by the APA.

This is one of Frances's primary concerns, that over-diagnosis is damaging the field of psychiatry.

Worse, if Brody is right, the de-legitimization of the DSM may be damaging mental health care generally. Providing effective mental health care to the population requires close cooperation between mental health specialists and primary care clinicians (like family physicians). There are not enough psychiatrists to care for the mentally ill. Some rural areas have no psychiatrists at all. The way to make this work is collaborative care, in which primary care physicians work directly with patients in consultation with a mental health specialist. This will not work if the disciplines cannot agree on a common language to describe cases.

Or it may be that pressure from family physicians and other peers will encourage the APA to reconsider the design of its diagnostic system.

 

Andrew Gelman on Gary Gutting

post by Bill Gardner

I criticized a recent blog post by Gary Gutting on the supposed over-valuing of social science in policy decision-making. Andrew Gelman is also skeptical that social science is over-valued ("Responding to a bizarre anti-social-science screed"):

In no area are scientific conclusions “the primary guides to setting policy.” Political and business leaders rule policy; the rest of us can just supply advice. It’s not like physicists are in charge of energy policy (if so, I expect we’d have a big fat carbon tax), nor are biologists in charge of the teaching of evolution in many states. So I’m not quite sure what Gutting is talking about.

The only place where I see social scientists controlling policy is (some) economists’ influence over economic policy... But this can’t be Gutting’s point: if it were, he’d talk about economists, not social science in general. But in that case I just don’t get it. It’s not like there are a bunch of number-crunching sociologists running around telling the government what to do!

The death panel libel again

post by Bill Gardner

The US Preventitive Services Task Force has issued a new recommendation about screening for prostate cancer.  Dr. Michael LeFebrvre, The co-chair of the task force, offered this summary:

Prostate cancer is a serious health problem that affects thousands of men and their families. But before getting a PSA test, all men deserve to know what the science tells us about PSA screening: there is a very small potential benefit and significant potential harms. We encourage clinicians to consider this evidence and not screen their patients with a PSA test unless the individual being screened understands what is known about PSA screening and makes the personal decision that even a small possibility of benefit outweighs the known risk of harms.

Cancer screening with the prostate specific antigen (PSA) test has been rigorously studied in hundreds of thousands of patients; the USPSTF recommendation was based on those data.

The Wall St. Journal has published an editorial by Tom Perkins, the founder of Kleiner Perkins and a cancer patient. He summarizes the report as follows:

A recent announcement by the U.S. Preventative Health Service can rather simply be summed up: Most men eventually get prostate cancer, but most don't die from it; those who do are mostly over 75 years of age, so that ends their continuing burden on the public purse. Further, early and prolonged testing is expensive, and can lead to medical complications from biopsy examination.

Mr. Perkins is lying. The recommendation makes no reference to the expense of any medical procedure or any burden on the public purse. The recommendation discusses only the benefits and harms of the use of the PSA test as a screen. Perkins concludes that

A highly taxed and highly regulated economy leads to "Death Panels," like the U.S. Preventative Health Service.

Apparently placing a defamatory claim in quotes makes it ironic or something. You can rationally disagree with the USPSTF. The way to do it is with data.

UPDATE: Go to Carey Goldberg's blog to see a graph that does a terrific job of explaining why current data indicate that the PSA screen provides few benefits and substantial harms. (h/t Austin Frakt)

Allen Frances on over-diagnosis in psychiatry

post by Bill Gardner

Anyone with a clinical or policy interest in mental health -- and this will include anyone concerned about US health care costs -- should listen to this lecture by Allen Frances on the problem of over-diagnosis in psychiatry (see also here).

Frances argues that too many people with mild to moderate mental distress are diagnosed with mental illnesses. This is harmful to people who are diagnosed in error, it is harmful to the 5% of the population with severe mental disorders who have less access to services, it is harmful to the nation because of the rapid growth of spending on psychopharmacological medications, and it is harmful to psychiatry as a discipline. The principal causes of over-diagnosis are the gradual loosening of diagnostic criteria over time, and the unremitting pressure by the pharmaceutical industry to exploit that looseness.

These are familiar criticisms. Nevertheless, this lecture is important for two reasons. First, Frances is an exceptionally lucid and expert speaker. The second reason is who Frances is and what makes him an expert.

Allen Frances is an emeritus professor of psychiatry at Duke, where he was also chair. More importantly, he was the chair of the task force that wrote the fourth edition of the American Psychiatric Association's Diagnostic and Statistical Manual (DSM-IV) and carried out many empirical studies to validate the diagnostic categories. This is probably the most important job in psychiatry. The DSM defines psychiatric diagnoses and is, as Frances describes, a central part of the problem. Frances is not a skeptic about the reality or importance of mental illness. The lecture is his account of the unanticipated consequences of his own work. 

Gary Gutting on social science and policy decisions

Gary Gutting appears to think that policy makers give too much authority to social science in decision making. Is there really a problem of social science being over-valued in policy decisions? I doubt it. The consensus results of physical science on climate change are ignored in policy making. 

Neverthless, Gutting makes several cogent critical points about the limitations of social science. Using research on teaching as an example of policy-relevant social science, he questions whether any work on the effectiveness of teaching -- or, we are to infer, other social sciences -- is sufficiently established to support major policy decisions. He thinks not, because unlike physical sciences, these sciences are unable to make and confirm precise quantitative predictions.

The reason is that such predictions almost always require randomized controlled experiments, which are seldom possible when people are involved.  For one thing, we are too complex: our behavior depends on an enormous number of tightly interconnected variables that are extraordinarily difficult to  distinguish and study separately.  Also, moral considerations forbid manipulating humans the way we do inanimate objects. 

So the answer would be to find ways to do more experiments, right? But Gutting doubts this will help much because whereas a neutron is a neutron is a neutron, the causal complexity of human behavior and the diversity of human settings limit the range of cases to which a given experimental result will apply.

There is a lot of truth in this. But on the question of whether we should use social science in making policy decision, Gutting fails to give a clear answer:

My conclusion is not that our policy discussions should simply ignore social scientific research.

Really? Even though Gutting seemed to be arguing that social science was not sufficiently sound to support major policy decisions? Then how should we use social science in policy making?

Given the limited predictive success and the lack of consensus in social sciences, their conclusions can seldom be primary guides to setting policy.  At best, they can supplement the general knowledge, practical experience, good sense and critical intelligence that we can only hope our political leaders will have.

It is unclear what Guting wants if he doesn't think social science should be ignored. Social science can only "seldom be a primary guide to policy decision making." Does this imply that occasionally it can be a primary guide? If so, when? Perhaps when there is sufficient consensus and predictive success? Then what criteria do we use to identify sufficient consensus and predictive success? And when social scientific studies can't be the "primary guides to setting policy", are they to be ignored, or can they be secondary or tertiary guides? And so on. 

Gutting owes us an alternative to current uses of social science in policy making. If he thinks we should put more trust in politicians, what are the grounds for preferring the intelligence of political leaders to the results of social science? Presumably Gutting is under no illusion that politicians' predictions about policy outcomes are precise and well-confirmed. Concluding with a hope that those leaders will have "general knowledge, practical experience, good sense and critical intelligence" is intellectually irresponsible. Or if he meant it, insane.

My alternative would begin with the extensive literature on how scientific evidence can be synthesized and assessed for policy relevance (see here for work in medicine). We do not need to settle for empty faith in politicians.

Life in a civilized country

Tomorrow is Victoria Day, so this is a three-day weekend in Canada. Something interesting is happening. My colleagues are taking the time off. 

American Medical academics don't do this (I believe it is called 'relaxation'). I once attended a small conference about chronic illness where the organizers started with an ice breaker: "Tell us what you did on your vacation last summer." It was supposed to put us at ease and it was a total failure, because the modal response was, "I wrote a grant proposal." The medical research culture selects in that sort of personality, and rigorously selects out alternative patterns of behavior.

Today and here, however, everyone understands that we live in a beautiful country and we have been given beautiful weather and we can't take these to our graves so let's stand together, and open the iron cage. Except me. But I am giving it serious thought.

Blogging and the intellectual long tail

post by Bill Gardner

I spoke yesterday at a conference organized by the Agency for Health Research and Quality (AHRQ), making the case for why scientists should blog. One reason is that, as Austin Frakt has discussed, blogging is an excellent way to get scientific information into policy discussion.

Bringing more science into policy discourse is good citizenship. But what's in it for you?

Scientific blogging can enable you to find the invisible college of people who care about an intersection of disciplines that you care about, and you may have believed that you were the only one who cared about.

The graph is from Chris Anderson, who used it to discuss how the Internet facilitates the formation of a subculture around music. What was graphed in the original figure was sales of music, with songs ranked from left to right in order of popularity. On the far left will be Lady Gaga and Bruce Springsteen who are well served by mass market channels. On the far right will be musicians like my son, who plays traditional Celtic music at Speed Metal tempos and volumes. There is an audience for his work, but you need a channel with the reach of the Internet to find the possibly 200 people who are interested in feedback-distorted fiddle music.

For Something Not Unlike Research, it is the intersection of health care policy and political philosophy. Ohio State is an absurdly large institution, but I never found the colleague who felt that it was important to understand both John Wennberg and John Rawls. But trying to write about health care and social justice on the net attracted the attention of people like Paul Kelleher (and our great commenters), who not only understands Rawls but exposed me to more recent authors like Jonathan Wolff. The incentive to blog, and particularly to blog on specialist topics, is not that it is likely to make you famous or even widely read. Rather, it increases your opportunities to converse with colleagues who share your core interests. 

Becalmed waiting for SCOTUS

post by Bill Gardner

I have the sense that the health blog conversation is in a doldrums. After the struggles of 2009 and 2010, the ball is in the Supreme Court, as it were. Blogging about health seems less urgent, perhaps because I do not expect those ladies and gentlemen to change their views based on whatever data I can assemble in the next month.

Is persuading the decision makers ever the point of blogging? Perhaps it is for Krugman. For the rest of us, it is just an opportunity to participate in the political discourse, and if possible to inform it with science. This is not a new idea. Pericles again:

We alone regard a man who takes no interest in public affairs, not as a harmless; but as a useless character… The great impediment to action is, in our opinion, not discussion, but the want of that knowledge which is gained by discussion preparatory to action.

Blogging is changing the world in ways that we cannot easily measure. Consider how women used to learn to mother. You consulted your mother / grandmother / aunt / sister, possibly your priest, or read a book by an elderly male pediatrician. Now there are tens of thousands of talented peers writing everyday about what marriage and family mean. Culture matters, for health as much as family, and we can shape it.

On slippery slope arguments that (sometimes) work

post by Bill Gardner

Austin Frakt has a post on bad slippery slope arguments, of which there are many. A recent example is the argument at the US Supreme Court about the Affordable Care Act, where Justice Scalia asked,

Could you define the market — everybody has to buy food sooner or later, so you define the market as food, therefore, everybody is in the market; therefore, you can make people buy broccoli...

The basic strategy of argument is "if you move a step in direction X, you will inevitably end at location X*, which is horrible." Why stepping toward X leads you to X* is unexplained, so the argument seems to have only rhetorical force.

But there is at least one slippery slope argument which sometimes makes better sense. You can be in a situation where circumstances press you to step in direction X. But the situation has positive feedback, so that moving a step toward X generates consequences that increase the pressure for further movement. The natural and social worlds are full of examples of positive feedback, such as fission warheads, avalanches, cattle stampedes, and bank runs.

The classical example of a slippery slope in politics, as it were, is Pericles' speech to the Athenian Ecclesia urging war with Sparta over a seemingly peripheral issue. It begins, "Athenians, I say, as I always have said, that we must never yield to the Peloponnesians."

For some time past the designs of the Lacedaemonians have been clear enough, and they are still clearer now. Our agreement says that when differences arise, the two parties shall refer them to arbitration, and in the mean time both are to retain what they have. But for arbitration they never ask; and when it is offered by us, they refuse it. They want to redress their grievances by arms and not by argument; and now they come to us, using the language, no longer of expostulation, but of command. They tell us to... rescind the decree respecting the Megarians... I would have none of you imagine that he will be fighting for a small matter if we refuse to annul the Megarian decree, of which they make so much, telling us that its revocation would prevent the war. You should have no lingering uneasiness about this; you are not really going to war for a trifle. For in the seeming trifle is involved the trial and confirmation of your whole purpose. If you yield to them in a small matter, they will think that you are afraid, and will immediately dictate some more oppressive condition.

In other words, the attempt to appease the Spartans will not only cause us harm, but will also embolden them to demand even more.

Now, does positive feedback apply to the constitutionality of health care reform? I don't think so. Scalia may believe that if the government can mandate the purchase of health insurance, it will be emboldened, like the Spartans, to demand further control over economic life. But if there was positive feedback here, wouldn't we expect that the many societies that implemented nationalized health care systems after World War 2 would have accelerated towards totalitarianism? In fact, after a rapid expansion of governmental control of the economy, there was a world wide reversal in the 80s and 90s. So the idea that there is positive feedback loop driving the expansion of social democracy doesn't seem to fit the evidence, but the idea itself is not empty.

Acoustic modems? Seriously?

post by 3 of 9

Yes, 3 of 9 of the Borg (you may remember my sister, 7, she and I have lots of jokes involving prime numbers, write me if you are interested). I am guest-posting to comment on the current state of real-time medical monitoring available to you, the unassimilated.

In particular, let's look at the tech that Dr. Gardner is using to manage his atrial fibrillation. One of the atria of his heart has a tendency to contract prematurely, upsetting destabilizing his heart rhythm. This can trigger an episode of very rapid but ineffective beats, i.e., fibrillation. 

This problem can be moderated, sometimes, using anti-arrhythmic medications. The problems with these medications are that a given drug will work for only some patients, and sometimes a drug that seemed to work for a while then seems to stop working. So what they wanted to do for Dr. Gardner was to give him the next line of therapy: a different anti-arrhythmic drug, propafenone (rhythmol). Rhythmol, however, has side-effects, and is therefore taken with another drug to counter the side effects of the first drug. And the second drug, in turn, may cause... you see where this is heading.

However, the actions of anti-arrhythmic drugs can be affected by diet, exercise, and stress. This, Dr. G argued, is a perfect opportunity for personalized medicine. What humans ought to do is use on-board technology to monitor heart rate and blood pressure, and use a diary to look for patterns of behavior that might be changed to reduce the episodes of fibrillation. For example, cutting all stimulants (even tea, OMG) out of his diet has seemed to help. But has it really? We need individual data to tell. The key idea in personalized medicine is not genomics, it's continuous real time monitoring.

But, to make this work, the technology has to work. And, unassimilated humans, your technology, at least at this Enormous MidWestern University Hospital, truly sucks. Dr. Gardner now walks around with a wearable heart monitor attached to leads on his chest. Where, however, do the data go? Bluetooth to the Macintosh? Um, no. Well, then, presumably you can put it on a flash drive? No ports on the device. What you do is, first, find a land line (ask your mother). You then put the mouthpiece of the telephone handset on the monitor's on-board accoustic modem (ask your grandmother) to send it to a company in California. This is a ten minute process that has to be performed after every three episodes. Can Dr. Gardner get these data back to look at them himself? No indication of how this could be done.

There must be better equipment somewhere here on earth. Or perhaps you could ask the Borg.