"The Emperor's New Drugs" by Irving Kirsch

A persuasive debunking of anti-depressants, with eye-opening coverage of the placebo effect

They say that the Origin of Species is "one long argument." Irving Kirsch may not share Darwin's eloquence, but in The Emperor's New Drugs he shares his passion for persuasion. Thanks to its wide scope, smooth delivery, and mastery of the data, this book is about as persuasive as a popular science book can be.

"The belief that antidepressants can cure depression chemically is simply wrong." So Kirsch claims. A claim like this raises a host of questions. Some are easy to answer: why would drug companies exaggerate the value of their pills in an anti-depressant market worth $19 billion a year? Why would regulatory agencies that are partly funded by drug companies play along with these exaggerations? Other questions are harder: if antidepressants do not cure depression chemically, how do they do so? And if the answer is "the placebo effect", how can the placebo effect be so strong as to convince millions of patients, thousands of doctors, and dozens of editors, that antidepressants are more than just glorified sugar pills?

Some of the tough questions turn out to have simple answers. The reason everyone was duped by the chemical-imbalance theory of depression, says Kirsch, is that the theory itself was based mainly on the (supposed) effectiveness of chemicals in treating depression. Some answers rely on clever reasoning. Clinical trials show that antidepressants are actually more effective, by a small but significant amount, than placebos. Kirsch explains this deftly as an "enhanced placebo effect": patients who detect the side-effects of antidepressants know that they are on active drugs, raising their expectations about the treatment and enhancing the placebo effect.

What all Kirsch's answers have in common is thorough attention to the relevant data (published and unpublished) and a keen nose for interpretation. Kirsch marshals an impressive range of evidence to back his case: the bibliography runs to 25 pages and consists mainly in articles from top medical and psychology journals. If he does not have a study or meta-analysis to back up a claim, he says so. And he knows that a striking anecdote is just a striking anecdote, even if it punches for his own team.

The book a good first course in scientific method, and a key lesson in the Kirsch curriculum is that data alone does not put a hypothesis to the test. Data, plus a dose of careful interpretation, is the only real medicine in science. Drug companies did not falsify the reports of individual patients or doctors. Nor did they (usually) fudge individual studies. The devil was not in the details but in the grand design, the way they selected out negative studies and re-hashed positive ones. In lifting the lid on the cover-up, Kirsch gives a running response to those say that meta-analysis, as a scientific technique, is indefensible; nay, says Kirsch, it is indispensible.

The question every reader will have is partly an ethical question: given that antidepressants would no longer be effective if everyone knew they were only placebos, should the "dirty little secret" be made public? Kirsch, true to form, answers this question with a patient summary of studies and meta-studies. In doing this, he does not ignore the ethical core of the question. The aim is to cut through the empirical flesh to make the core issue as clear as possible. Kirsch shows that if you cut deftly enough, the core issue might not be ethical at all. This book asks: if we can show that psychotherapy is cheaper, safer, and more effective in the long run than anti-depressants, what ethical argument could possibly warrant the continued prescription of anti-depressants? Good question.

When it comes to writing clear prose, it is not always advantageous to be a scientist. But for Kirsch, it is so. He is no wordsmith (or doesn't want to be), and if you are looking for blazing rhetoric then this not the book for you. But if you want to understand what a balanced-placebo test is and why it works, how neurotransmitters are meant to explain depression, and the difference between "response-rate" and "average improvement" in clinical trials, Kirsch is a lucid guide. His prose might read dryly for some. But the result is that if you can understand a bar graph, you can understand this book.

What Kirsch lacks in verbal charisma, he makes up for in arresting content. His chapters on the placebo effect make for fascinating reading. My favourite is the man who swallowed all his pills and collapsed in a heap on his GP's floor -- only to find, when he came round, that he had overdosed on fake pills. Equally striking are the basketball-players whose knee problems were fixed using placebo surgery, and the angina, dermatitis, and electro-shock victims who were all cured or assuaged by the power of belief. Kirsch describes lots of experiments designed to tease out the details of the placebo effect. The methods are clever, and the results run a skewer through our intuitions about physiological cause and effect. The results seem like voodoo, but the methods do not. As this book reminds us, implicitly but forcefully, it is the methods that matter.

Kirsch is sometimes not as methodical as he might be. One defence of anti-depressants is that they have both a placebo effect and a real chemical effect, but that these two effects are not additive. Kirsch describes how this hypothesis might be tested, but admits that no such tests have been done. He tells us that drug companies, who would otherwise sponsor such tests, are running scared. Fair enough; but the fact remains that the tests have not been done. Also, a bullet-point summary would be useful to tie up the threads of evidence against anti-depressants; sometimes the same thread turns up in widely separated chapters, making it hard to keep track.

Kirsch could do better to explain the weirdness of the placebo effect. How can the mind restore the cartilage in a bad knee just by expecting the knee to be cured? Kirsch suggests that this is easily explained as a purely physical causation, the brain acting on the knee. But it seems just as weird for the brain to cure cartilage as for the mind to do it. It also seems weird to say that the brain "expects" something. Lastly, it seems weird for positive expectations to have a positive effect. Why don't positive expectations just make the brain more complacent, and therefore idle? Kirsch seems complacent about explaining the placebo effect, even if (or because) the evidence for its existence is overwhelming.

These quibbles do not threaten Kirsch's argument. In the epilogue Kirsch says he enjoys "rocking the boat." And the evidence suggests he has knocked antidepressants into the water. He reports a recent survey of UK clinicians showing that almost half will (or have) changed their practices because of Kirsch's work. He has also made waves in the murky waters of drug regulation, helping to bring about proper tracking of drug trials. But he is a placid revolutionary, and his easy prose and wide knowledge make for a smooth ride – and a persuasive one.

[summary only]

[read more]

Another look at the people of the screen

Two points of method in the debate about online reading

Literacy, the most empowering achievement of our civilization, is to be replaced by a vague and ill-defined screen savvy. The paper book, the tool that built modernity, is to be phased out in favor of fractured, unfixed information. All in the name of progress.

This is Christine Rosen writing in The New Atlantis on the fate of reading in a digital age. Needless to say, it's not a very happy article. Rosen means the article as a counterweight to the work of "techno-utopians", who see the digitisation of books as an unalloyed good. But Rosen doesn't just balance things out -- she slides them into gloom. Screen-reading won't just be book-reading in a new medium, she says. It'll be a new activity altogether, where reading is shallow and aimless and readers are numb to their own humanity.

I'll look at Rosen's main worries in the next post. But first, here's a couple of general points about the debate, inspired by Rosen's article: compare like with like, and don't let differences of medium obscure similarities of function.

Compare like with like

To be fair, Rosen scolds a "technophile" (Steven Johnson in the London Guardian) for failing to compare like with like -- for comparing computer programmers with book readers in terms of their likely economic success in the digital climate. As Rosen points out, "Johnson would have done better to compare obsessive novel writers and obsessive computer programmers...Most of the people immersed in screen worlds are not programmers. They are consumers who are reading on the screen, but also buying, blogging, surfing, and playing games."

That's fair enough, but it rubs both ways. Rosen is as misguided as Johnson when she compares the book-reading community with the entire web community, in all its moods and activities. It's more plausible to compare the web community to the entire non-web community, where the latter includes (along with book-readers) its fair share of shallowness, idleness, and wasted time. The online community, like the offline community, has its disciplined and intelligent suburbs and its more scruffy one. The real question is whether the internet will give a net outflow from the former to the latter.

So if Rosen is right, and cognitive content of normal reading has been replaced by, and not just supplemented by, a "vaguely defined screen savvy", that's something to worry about. But if we do an equivalent of quality reading on the internet now, as before, there's no immediate concern. Importantly, this means that we may be OK even if we spent the majority of our internet time in the aimless, shallow, "power search" mode that Rosen laments.

Different medium, same function

Rosen scolds Johnson again for failing to distinguish between the values of the digital world and values within the digital world. Computer programmers might earn a lot in the "digital climate", she says, but that's no use if the climate itself is wrong. And there's truth in that. But the overall lesson of Johnson's article, in seems to me, is not just that techies achieve techie goals. It's rather that they achieve goals that everyone wants to achieve -- like professional success and political awareness. Activities in the digital medium can serve the same function as different activities in other media -- even if the difference of medium tends to obscure the similarity of function. In Johnson's words: "Are you not exercising the same cognitive muscles because these words are made out of pixels and not little splotches of ink?"

With this in mind, the following claim from the article looks fishy: "People who read regularly for pleasure are more likely to be employed, and more likely to vote, exercise, visit museums, and volunteer in their communities; in short, they are more engaged citizens."* Perhaps your average blogger is less likely to go to museums or help out in the local soup kitchen. But they become engaged citizens in ways that are peculiar to the internet -- online donations, political awareness, writing open source code, getting involved in decent online forums.

*To be fair, the claim comes from a National Endowment for the Arts (NEA) study, not from Rosen. And she counters it with Harold Bloom's skeptical view of the link between books and civic engagement.

[summary only]

[read more]