Something Not Unlike Research

Primary care physicians, patients, and routine PSA testing

The US Preventive Services Task Force (USPSTF) has issued a controversial recommendation against routine screening for prostate cancer using the prostate-specific antigen (PSA) test.

Following the release of a draft version of that recommendation, Craig Pollack and his colleagues asked 114 primary care doctors whether they intended to change their practice. The interesting case here is the intentions of the 56 (49%) doctors who agreed with the USPSTF, because there is no reason to expect that those who did not agree would change their views. Of those,

even among those clinicians who agreed with the draft recommendations, fewer than half (41.1%) stated that they would no longer order routine PSA screening or be much less likely to do so. 

Should this bother us? If the USPSTF is right, which is of course arguable, then continued routine PSA screening will harm a significant number of men and benefit very few.

However, framing this issue simply as a matter of physician decision leaves the patient out of discussion. Dr. Michael Barry believes that "there will be informed patients who will still want a PSA test, due to risk factors or other personal reasons." Perhaps the physicians who agree with the USPSTF are thinking of such patients. Should a physician who believes that the test is on balance likely to be harmful accommodate such a request? Surgeons do not perform amputations of healthy limbs, even though there are patients who want these procedures.

Let's suppose that the physician was able to explain to the man the likely odds of benefit (extension of life as a result of prevention of prostate cancer) and harm (incontinence, sexual dysfunction, surgical risks), and that there were no circumstances that identified the man as being at elevated risk for prostate cancer. Nevertheless, the patient might have a consuming fear of dying of cancer. Is this irrational? Cancer is a difficult way to die. Perhaps the patient believes that the reassurance he would receive from a negative result would outweigh the risk of harm from a positive result. I have chosen not to be screened, but I think that getting a PSA screen is within the space of plausible choices.

So if having fewer routine PSA screens is a desirable goal, and we do not necessarily think that doctors should refuse to supply them, what needs to change? Of course, the primary issues is that most people do not understand the odds of harm and benefit consequent to a screen. If physicians could find ways to convey this information to patients, and refrained from simply ordering the test without fully informing patients, I believe there would be far less demand for the screen.

 

As if a prisoner were a human being

I had the wonderful fortune to attend my daughter's graduation from Bard College this weekend. Bard is a liberal arts college and most of the students come from affluent families.

But not all. Bard runs a Prison Initiative at the Eastern Correctional Facility in New York State. The men and women in this program receive a Bard education, including the completion of a senior thesis. Ordinarily, and for the obvious reason, the Prison Initiative holds its commencement in the prison. But one of the prisoners was released (after serving 11 years) in time to complete his degree on the main campus, so he participated in that commencement.

He received several minutes of standing ovation when he walked across the stage. It was almost as if he was a human being, and his intellectual and personal development actually mattered.

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.