The Hidden Gorilla

Three weeks ago What would Batman do Now covered the issue of suicide in the military – an issue that had Batman missing in action, and the Joker suffering the adverse effects of psychotropic drugs. Then along came James Holmes to the premiere of Dark Knight Rises in Aurora.
Most drugs that can cause suicide, including the antidepressants, mood-stabilizers, antipsychotics, smoking cessation drugs and others, can also cause violence. The akathisia, psychotic decompensation, or emotional disinhibition these drugs trigger that lead some to suicide, lead others to violence (see Healy et al 2006).

There is some awareness that these drugs can cause suicide but considerable resistance to the idea. There is less awareness and even greater resistance to the idea that they can cause violence. Treatment induced violence lies in a medical blind spot – no doctor wants to contemplate the possibility that she may have had a role in the deaths of innocent third parties.
This may be the grim prospect facing Dr. Lynne Fenton. Dr Fenton we are now told had been seeing James Holmes, the killer at Dark Knight Rises in Aurora, and had seen him just a week before the killings. Given the current reliance of American medicine on medications it seems likely that medications are involved in the Aurora case.
For many the instinctive reaction to Holmes will be that he is either mentally ill, evil or a street drug addict. This makes sense. Violence is one of the associations we all make to the ideas of evil, mental illness and illicit drug use. In contrast most of us know people on antidepressants none of whom are violent. This makes it difficult to accept a link to prescription drugs. For many even raising the idea that Holmes may have been crazed by a prescription medicine is likely to sound deranged or the excuse of a bleeding heart liberal.

But in fact there is a great deal of publicly available clinical trial (Hammad 2004, p40-41) and other data highlighting the risks of violence from psychotropic drugs. There is far more hidden data. There is in fact no other area of medicine in which there is so much hidden data on a risk that has consequences for the lives of so many innocent third parties.
With each “outing” of suppressed data lately companies have been beating their breasts about the lack of transparency “in the past” and have committed themselves to greater transparency. Here’s a chance for our major companies to prove things have changed by making the data on hostility, aggression and violence on their drugs publicly available. These data might tell us something about who is at risk, and allow us to better manage these risks. If there were a conspiracy to keep the details of all plane crashes out of the public domain, would airlines or the authorities have any incentive to make travel safer?
Instead, we are likely to see a vigorous marketing of articles that deny the possibility of a link. It takes really great science to overcome our biases. But if an article fits in with our biases (our associations), almost anything can be published, and doctors can be depended on to treat it as respectable science.
While 9 + of us out of 10 find the idea that an antidepressant might have caused Holmes to behave the way he did unbelievable, those whose lives have been touched by these issues are in a completely different position. News of another mass shooting immediately raises the suspicion that an antidepressant or related drug will be involved. And as, Rosie Meysenburg has shown on SSRI Stories, the drugs are all too often involved.

The drugs have been involved so often in campus or mall shootings that for some the surprise is that the medication question is so slow to get asked, as Peter Hitchens who is not a bleeding heart liberal has pointed out. What political considerations keep the NRA out of the debate? When Batman tells America “no guns, no killing”, there must be a temptation to respond “no drugs, no killing”.
But if Holmes turns out to have been on a drug that can cause violence, it is a quite separate matter to establish that in his case the drug he was on did contribute to what happened. It may not have. Without details of the case it is difficult to offer a view.
But this will not stop the debate in the public domain about an easier question for drug companies to control – do psychotropic drugs cause violence. And here, even though in some jurisdictions companies are legally obliged to say their drug can cause violence, a recent article in Psychopharmacology by Paul Bouvy and Marieke Liem denying the possibility of a link is certain to be marketed heavily.

Bouvy and Liem’s article has much in common with recent articles by Robert Gibbons in Archives of General Psychiatry (see Coincidence a fine thing & May Fool’s Day). These articles may have no links to or input from industry, but they fall on the fertile ground of a distribution system complete with public relations companies geared up to make sure that messages like this get picked up and equally that messages about problems that treatment may cause do not get heard.
When it comes to Adverse Drug Reactions (ADRs) on prescription drugs, there is no such thing as an academic debate with equal airtime for both sides, although Psychopharmacology have published a response to Bouvy and Liem’s article unlike Archives of General Psychiatry which has refused to publish responses critical of Gibbons’ articles.
Bouvy and Liem correlated data on lethal violence in Holland between 1994 and 2008 against sales of antidepressants. The drug sales went steadily up and the number of episodes of lethal violence fell, leading the authors to claim that “these data led no support for a role of antidepressant use in lethal violence”.
This is a marvelous example of what is called an ecological fallacy. An ecological fallacy is when someone claims that if an increase in the number of storks parallels an increase in the number of births that storks must be responsible for births.

The best known example of storkology in recent years were the graphs produced by tobacco companies showing rising life expectancies and even reduced deaths from respiratory illnesses in line with rising cigarette consumption. These were produced as part of a Doubt is our Product strategy to deny the risks of smoking.
Recent sightings of storks include claims that increased SSRI use is linked to falling national suicide rates. The articles making these claims offer data from the late 1980s but disingenuously omit some key facts. One is the fact that suicide rates in most Western countries were falling before the SSRIs were launched. Another is the fact that both suicide rates and antidepressant use rose during the 1960s and 1970s when antidepressants were being given to the most severely ill people at the greatest risk of suicide. This was when suicide rates should have fallen if antidepressants have any effects on national suicide rates (Reseland et al 2008).
Autopsy (post mortem) rates are also left out. The more autopsies done the more suicides and homicides are detected. Autopsy rates rose in the 1960s and 1970s and fell from 1980 before antidepressant consumption began to escalate dramatically. The rise and fall in autopsy rates perfectly mirrors the rise and fall in suicide rates.

For the purposes of this argument, let’s assume the data on episodes of violence in Holland that Bouvy and Liem use is correct. This may not be the case – British national suicide rates are no longer dependable. The national figure is in essence set by a bureaucrat in London, who has scope to make the rate rise or fall as needed. Let us also assume declining autopsy rates play no part.
Before considering what else could be involved, let’s look at the shape of the argument and ask why Psychopharmacology would take an article like this. First alcohol use has increased in Holland during this period but no-one is making the argument that increased alcohol use has led to a decline in acts of lethal violence or the further Bouvy and Liem argument that this means alcohol cannot cause violence. Why not? Because, we associatealcohol with violence.
SSRIs cause growth retardation in growing children. The clinical trial data show this retardation and the labels for the drugs mention it. During this period SSRI consumption among children has increased in Holland but the Dutch have become the tallest people in the world and are getting taller. Where is the article saying that the increasing height of the Dutch proves that SSRIs don’t retard growth?
In the case of violence, the published trials show antidepressants cause it, probably at a greater rate than alcohol, cannabis, cocaine or speed would be linked to violence if put through the same trial protocols that brought the antidepressants on the market. The labels for the drugs in a number of countries say the drugs cause violence. And there is at least one clear and well-known factor, just like autopsy rates, that can account for the findings – young men. Violence is linked to young men, and episodes of lethal violence are falling in all countries where the numbers of young men are declining.

Whatever Psychopharmacology were doing taking an article like Bouvy and Liem’s making claims that run counter to the warnings that are already on the drugs, without warning their readers that this was the case, from here on the game for industry is about managing associations. From conmen to hypnotists to Batman, the trick is to hold the audience’s focus so they miss something much more important in their peripheral vision field. This is what public relations companies excel at.

In the case of prescription drugs, the key people are doctors, the Watsons. Always one step behind the smarter Holmes. While it would be nice to see Watson turn the tables for once, in this mystery Holmes has the last line once again. It’s elementary My Dear Watson. For an ADR you need A Dr.