SSRIs in The Atlantic: Forget the Science Bring on the Anecdotes

Jonathan Leo, PhD
22
152

The Atlantic web site has just published a strange piece on the efficacy of the antidepressants. When getting into a discussion with about antidepressants with a supporter of the drugs the first thing to figure out is: Are they going to talk about science or anecdotes? Ten-years ago the proponents usually talked all about science and evidence based medicine. If you brought up an anecdote you were considered non-scientific. But the tables are now turned. Now that the science is showing that the once presumed efficacy of the SSRIs is questionable, the proponents are relying more on anecdotes. Given the fact that the likes of Kirsch, Whitaker, and Angell have been focusing on the science it would seem that someone who disagrees with their arguments should really focus on the science and not just throw out another anecdote. Don’t look for a discussion about the science in the latest Atlantic article written by Maura Kelly.

Right after the title, before even getting to the article, there is a header that shows that the author doesn’t even really understand what the debate is about. Kelly says, “But just because some people are taking SSRIs that don’t need them doesn’t mean that the drugs don’t work.” No one – not Kirsch, Whitaker, or Angell, has ever equated the idea that lack of efficacy in an individual person means that the person is not depressed. They all acknowledge that depression is real and that treatment is appropriate. The problem is that the SSRIs don’t appear to work very well. To question the utility of SSRIs in no way automatically equates to the idea that depression is not real. How could such an obvious error in logic make it into the Atlantic.

The article is interesting if one looks at it as one person’s experience with depression and recovery, but it adds very little to the discussion about the efficacy of the SSRIs. It is simply a personal testimonial to one SSRI success story. Of course, the SSRIs have worked for some people, even in the Kirsch meta-analysis they worked for some people. Scientists argue about the number needed to treat for the SSRIs, or the percentages, but very few people say it is zero. So the fact that out of the millions who take SSRIs there are some success stories is not surprising. No one is arguing this. But, the reality is that they haven’t worked for everyone.

The subtitle of the Atlantic Piece is: Antidepressants really help people.   But if you really want to prove this then the way to do it is to talk about the science and not another anecdote.  Rather than try and stack up anecdotes on each side and see which side is higher – which is what is going on in the comments section of the Atlantic article – a scientist designs an experiment to move the discussion forward. This is exactly what the companies did with their controlled clinical trials which they submitted to the FDA to get their drugs approved. And at first glance the studies seemed to say that the drugs worked, so everyone trumpeted the studies, and the drugs got approved, and the marketing departments did their job, and all in the name of science the drugs became a financial success. Then along comes Kirsch who simply looked at the studies in a more detailed approach and showed that the initial conclusions were a little premature. Keep in mind all he did was look at the same studies which were submitted to the FDA. He did exactly what a scientist is supposed to do. His studies highlight the fact that the drugs don’t work for a whole lot of people. Kelly, is now saying to forget the science, they worked for me.

No doubt, that for every legal and illegal drug being sold on the market or on the street, one can find users who will highly recommend the drug. But is this a reason to take it? In fact, most of those addicted to an illegal drug probably first took the drug because they heard from a friend how great it was. But wouldn’t these people have been better off if instead of listening to anecdotes they listened to science? Should people pondering the use of an antidepressant not be told about the science?

Consider her proof that the drugs work. She doesn’t cite studies, or data, but mentions some experts: “Luckily, I’m not the only one convinced of the good they’re doing.” She goes on to cite several psychiatrists who wrote into The New Review of Books to criticize Angell. Dr. John Oldman wrote in and said, “The bottom line is that these medications often relieve the patient’s suffering,” and Dr. Daniel Carlat wrote, “on the whole, they work.” But missing from the Atlantic Piece is Angell’s reply, which starts off with: “… these letters simply assume that psychoactive drugs are highly beneficial, but none provides references that would substantiate that belief. Our differences stem from the fact that I make no such assumption.” From there, Angell basically destroys the letter writers’ arguments.

Kelly also cites Peter Kramer who attempted to salvage the reputation of the SSRIs in the New York Times. According to Kramer, the main problem with the Kirsch study is that he based his work on the FDA database, which is nothing more than a collection of faulty trials. So because the trials were faulty we can forget the science and go back to anecdote. However, in the past, when the trials were in vogue, Kramer had a different stance. In his 2005 book Against Depression, he said “Antidepressants have proved effective in large scale multi-site trials whose results have been published in either way, pro or con.” In other words, when people thought the clinical trials supported the use of the SSRIs they were good trials. But now that a more in-depth analysis of the trials has revealed that there is little difference between drug and placebo, the trials should be considered irrelevant?

On the surface, instead of moving the discussion along in a scientific manner, the Atlantic piece simply harkens back to the days of anecdotes. At a deeper level it is highlighting the fact that the scientific justification for the SSRIs is on shaky ground.

22 COMMENTS

  1. interesting issues around ssri’s.. you go the doc they tell you ah this will work we will just increase your seratonin levels… 10 or so ssri’s later with a mirad of metabolic immune and cardio disorders and a lot of research you find they dont actualy do much to seratonin directly but rather alter the metabolisum of the endocannabinoid and opioid receptors.. http://www.springerlink.com/content/2127100x043n3674/
    this also goes a long way to give evedence to harms across all major systems that run on endocannabinoids and endovaniloids… pity the doctors either fail to tell you or simply do not know the full actions of ssri’s

  2. Marvellous post – thankyou.

    I have seen articles like this before, they are by people who use the drugs and who like the effect. They feel threatened by people questioning the use of the drugs on a societal level and confuse it with questioning people’s use of them individually.

    When I read the article she said she had a really bad patch bought on by insomnia. Insomnia will tend to make people’s emotions go a bit haywire. She said she had tried lots of other ways of treating the insomnia but none of them seemed to be about looking at what the psychological drivers were. Not that I can say that this would have worked but for me when I can’t sleep it is usually because someone has driven me up the wall.

    The drugs may work for her due to placebo or it may provide an effect she likes but as you point out this is not enough to provide general prescribing guidelines, for that you need a broader scientific base

  3. People willing to drug their brains when no physician has ever even looked at their biology and proven it disordered or diseased, are not interested in real science.

    It’s ‘sciency’ enough for them that a dude in a white coat told them they were feeling sad because they had a brain disease and bad genes while he remained seated behind a desk and didn’t even come near their body to test or measure anything.

    To the extent that they ‘work’, it is never just the drugs ‘working’, it is the twin pillars of fraud and the power of suggestion ‘working’ on these people.

    1) I believe I have a brain disease that makes me sad.

    2) I pop a pill that I believe ‘fixes’ my broken brain function.

    3) The combined ritual of believing I am a helpless and powerless passive victim of an active disease process, and my bottomless trust in the credentialed wizards of neurochemistry in the white coats that never even looked inside my biology, allow some believers to magically start feeling less sad after a while engaged in the cleansing ritual of swallowing the church of psychiatry’s daily eucharist wafers.

    But simply swallowing the eucharist wafers is not enough by itself, first one has to swallow the creation story of one’s own despair. It all depends on how hard to swallow you find these ‘revelations’.

    The large scale study back and forth on SSRIs is like a debate on whether the glucose or carbohydrate action on the biology in real Catholic communion eucharist wafers is responsible for the dose response of Catholicism in feelings of connected to the body of Christ.

    Psychiatric transubstantiation, whether anecdotal or writ large in large scale studies, is never going to tell us why so many millions of people are so able to put their reason aside and simply take it for granted that a doctor who has never once looked inside their biology, can ethically have his or her patients leave their office with the received belief that their sadness and despair in life is a brain disease.

  4. Thanks so much for bringing up the Atlantic piece, even if my reaction differs. We need as many anecdotes on the positive experience from medication as negative, especially first hand ones that would be posted on this site. Even so, anecdotes only go so far because they are subjective and may not give the whole picture. But they are the truth for the person writing them.

    Science, unfortunately, is as limited as the anecdotes. Science only examines part of reality, and in limited ways. This is especially true about the brain, which is so difficult to examine. Then, if you add the conflict-of-interest of the pharmaceutical companies doing the studies instead of NIMH, it should be buyer (patient) and seller (prescriber) beware.

    We need both anecdotes and scientific studies, as much as we can get, from as many perspectives as possible. In the meanwhile, we should be humble and tentative in our conclusions. At least that is the stance I will take. Hubris almost always causes problems.

    • Hey-Hey, patients who’ve already been through the mill with psychiatric drugs and treatments really don’t need to be told, yet again, how wonderful the drugs and treatments are.

      We get it that some people swear by them.

      Efficacy is one question: Do the drugs help people?

      An equally important question: Do the drugs hurt people?

      Only if BOTH questions are answered honestly can a clinician come to a reasonable assessment of risk-benefit for any one person.

      The second question, bearing on patient safety, is not addressed by “positive” anecdotes or even “positive” studies. Misadventure must be examined systematically on its own, to determine causation and frequency.

      What is psychiatry doing to track and analyze adverse outcomes from psychiatric treatment?

      It’s not from lack of p.r. efforts that psychiatry looks bad. It looks bad because its research base is shoddy and it has a lot of dissatisfied customers who will not shut up. Clinical practice manufactures more injured patients every day. You can’t kill bad word of mouth.

    • Hey-Hey, I agree that there will never be enough science for us build to develop a fully truthful understanding of everything based on it. We’re not supposed to though. Science is used to test hypotheses in order to build or puncture explanatory theories of some aspect of reality. The point being, the more we do it, the more honest we do it, and the more open and public the process, we can get to both profound and useful understandings all of sorts of phenomena.

      I agree that we should be wary of for-profit science intended to try to sell drugs. I certainly agree that patients and precribers should be wary of the science of the pharmaceutical companies. But, it really doesn’t take a sophisticated methodologist to read a study, assess the quality of its design, the rationale of particular statistical tests on particular variables, what important parts of the experience of taking drugs is left out, the accuracy of its results, and how honestly the conclusions are reflective of the actual data. I anticipate the pharma industries are going to try to skew their data as positively as possible, that is their job. It is prescriber’s jobs (and now also the FDA’s, insurers, governments, etc.) to be critical of their studies. On the other side, the pharmaceutical companies will not invest the millions of dollars needed to test a drug unless their are somewhat confident they can show significance greater than placebo on a particular outcome measure. This makes me wary of off-label use of medications because if it was generally likely to be even marginally effective beyond placebo, the drug companies would try to get the indication. Psychiatrists (unlike primary care providers) especially are freer with off-label prescribing, even though there is no FDA approval for use, and no/limited academic or government supported studies to demonstrate efficacy worth risk of harm.

      I agree that anecdotes are important, but just like the science, they need to be put in context and critically examined. Anecdotes also don’t make a convincing case for ongoing use of non-indicated psychiatric medications anymore than the current scientific literature.

  5. Speaking as a psychiatrist and an observer of the behavior of my fellow psychopharmacologists (a label I wear only reluctantly, myself), I find this science-vs-anecdote issue fascinating.

    In an attempt to be “scientific,” we (and the FDA) uphold the sacred p value as the sole, valid indicator of whether a drug “works.” And we’ll cite studies till we’re blue in the face about which drug showed a statistically significant signal over placebo, etc, to back up our prescribing habits.

    But we are also just as quick to brag about the 1 or 2 “success stories” we had with a particular off-label medication, or about the clever combination of 3 or 4 meds (for which there is no scientific support, and never will be) that “cured” a particularly complicated case.

    It seems like a double-standard to be able to move so easily between hard (i.e., “real”) science and personal (i.e., biased) observations. But when the former is often pseudoscience, and the latter is often nothing more than faith, it’s no wonder we pick and choose the logic that minimizes our own cognitive dissonance the most.

    • How would unintended consequences show up, if not as anecdotes? Isn’t that the way many adverse effects becomes recognized?

      Efficacy trials aren’t designed to capture adverse effects, they record them only incidentally. The STAR*D study switched 4,000 people on and off antidepressants yet recorded not a single incidence of withdrawal symptoms — a result that is not credible.

      It would be unethical to design studies to evoke and observe most adverse effects.

      Of necessity, many adverse effects are initially reported anecdotally. For example, see http://www.hbs.edu/research/pdf/09-122.pdf from Harvard Business School:

      “In 1999, the FDA approved Merck’s arthritis drug Vioxx. Through clinical drug trials, Vioxx had been subjected to rigorous tests and careful research. Soon after approval, however, Merck began to receive anecdotal evidence about patients who suffered heart attacks while taking the drug. Of course, many people suffer from cardiovascular disease and heart attacks are multiply determined. How then, should executives at Merck respond? Since Vioxx had already been subjected to rigorous testing, should the “anecdotal evidence” be dismissed? Or, should Merck conduct additional research exploring the possible link between Vioxx and heart disease? And, even before there is definitive proof linking Vioxx and heart disease, is it reasonable for doctors to become more wary of prescribing the drug?

      Locke and Latham (2009) deem anecdotes unworthy of academic attention….

      We profoundly disagree. We think that qualitative analyses, case studies, journalistic accounts, and anecdotes should all be used to raise questions, focus attention, and develop ideas that should be subjected to rigorous, causal analyses.

      We believe that our disagreements with Locke and Latham highlight not only our differences about goal setting, but also about what constitutes good scholarship….”

      And we all know how the Vioxx story turned out!

    • Dr Steve,

      There are other ways to analyze data beyond searching for significance. Also, a p-value doesn’t have all that meaning out of context. I agree we have made p-values sacred, when by definition it is only supposed to tell you how confident you can be that the difference in group means isn’t based on chance alone. This doesn’t mean much on its own. How studies designed are important. What variables are being manipulated matters. How outcomes are measured is important and the value of changes of that measure should be subject to criticism and its own analysis. Who is participating in the study matters and how representative are those participants (and how representative are they construed to be). I can go on.

      I believe quality science is essential for supporting and improving medical intervention. Accepting poor science in order to minimize dissonance (or to save face, keep a rationale to be employed, etc.) does not justify care, make care “good,” and can lead to a lot of harm.

      Psychiatry doesn’t have to be a choice between latching to pseudoscience or rigid faith. What the science does tell us is that how psychiatry is currently practiced has little basis in science. To me, this is an indication for recognizing all sorts of theories have not held up to testing, and that some more thinking and observing has to be done before psychiatry can reemerge as actually helpful. There is no guarantee for this, but it scientifically justified and medically responsible.

  6. Yes, I think we do need discussion about both positive results and negative on this site to try to get to some truth. Nowadays, I think we hear more of the negative than postive.

    In working with patients, then, there is a terrible dilemma. You can’t really rely on what studies we have or anecdotes. So what is the best we can do right now, providing someone wants help? Give careful informed consent and use each case as a case study of one (a time-tested alternative or addition to anecdotes and large studies), and then not prolong treatment unnecessarily.

    • Are there not many, many sites and publications extolling the positives? Why does this site have to provide the same information as a corrective? To be “fair and balanced”?

      Every single injured patient on this site represents someone whose trust in psychiatry has been betrayed — except for those who were forcibly treated. They may not have had any trust in it to begin with.

      How can you give sufficient information for informed consent when you don’t know what the harm quotient is?

      All you can do is make vague personal assurances that everything will be all right.

      Yet when things go wrong for the patient, what does the psychiatrist have to offer? How many have studied reversing iatrogenic damage? How many try to solve the problem by loading on more drugs?

      Prescribing psychiatric medications is shooting in the dark. The only honest informed consent foundation possible: “We don’t know how they work. They help some people and hurt others. I can’t predict which way it will go for you. I won’t be able to fix you if it goes wrong. Do you want this pill now?”

  7. The real question is, “What do we tell people who say, ‘My meds work.’ ”

    Because every person who has ignored me uses this line. When I try to tell them about emotional distress causing “symptoms” instead of some unproven disease, they don’t want to hear it. When I tell them maybe their meds just work because the withdrawal symptoms are nasty, they don’t want to hear it.

    How do you approach med users who love their meds? Anyone figured out effective techniques yet? The trouble is that the distress vs. disease model and the meds as painkillers only explanation involves a lot of science and digging. It’s very tough to communicate in the 30 – 40 seconds we usually get as advocates.

    • “What do we tell people who say, ‘My meds work.’?”

      I agree, this is incredibly frustrating. Even though there is no way to prove that a given medication “works” (particularly in the long-term), patients and providers want to believe this, so meds are continued. As we all know, some patients end up suffering greatly as a result.

      But a large number of people (perhaps the majority?) stay on meds indefinitely without even knowing that they don’t need them. Do these people “suffer” too? Maybe not in a physical, biochemical, objectively verifiable way (eg, by a lab value). But I maintain that the dependence– even if it’s purely psychological– on a drug, a diagnosis, or monthly visits to a doctor for the rest of one’s life, constitutes a form of spiritual suffering. If nothing else, it takes the power away from the individual and puts it in the hands of an entity (the doctor, an insurance company, the government, Pharma) that doesn’t always have that person’s best interests in mind.

      • A lot of people are on meds for a long time because when they tried to go off, they experienced withdrawal symptoms.

        Their doctors tell them that was evidence they’d need the medication to correct their flawed brains for the rest of their lives. The patients become convinced this is so.

        According to the CDC http://www.cdc.gov/nchs/data/databriefs/db76.htm :

        “More than 60% of Americans taking antidepressant medication have taken it for 2 years or longer, with 14% having taken the medication for 10 years or more.”

        That 14% should be studied to find out why they are on medication and whether they have any health problems that can be attributed to long-term use.

    • I say, “I’m very glad that you find them helpful. Unfortunately, not everyone has had that same experience. I’m not saying that some people can’t be helped with medication, I’m saying that we need to look at who is or is not helped and why.” This helps with anyone who is merely concerned that their own experience is being invalidated. For those who can’t hear that, I’d say you are in the realm of telling an avowed Christian that God is a social creation. They won’t be able to hear it, no matter what you say. So I try to spend my time on those who aren’t religiously committed to the DSM and biological psychiatry. Otherwise, it’s just a waste of breath.

      — Steve

  8. The fact in the article is that Kelly first got on SSRIs, then got off due to intolerable side effects, switching to a non-SSRI, and has not returned to SSRIs since. After 2 years, she got off medications. She herself apparently did not buy that she needed to permanently be on medications to correct a medical illness, and did not want to continue on medications.

    She returned to the non-SSRI after a year, because she became distressed and sleepless due to adverse events in her life. Kelly has elsewhere published accounts of childhood trauma and various psychological difficulties, including depression, that she directly attributed to these experiences, not to biological brain disorders.

    One question unanswered in the article is, if her Wellbutrin failed her, would she return to SSRIs, given the side effects that caused her to get off them? Didn’t those side effects constitute harm?

    Another question is, why does Kelly assume that she is immune to the placebo effect? Wellbutrin may be succeeding as a placebo merely in not having intolerable side effects.

    My last post was apparently seen as disrespectful of her motives, so I hope this is more appropriate. (I have no problem with you removing it if it still seems unproductive for the purposes of the website.) My aim is to draw attention to the fact that she herself, even anecdotally, is not an SSRI success story and her anecdote tends to undermine any notion of a biological origin to her mental distress.

  9. A much weaker drug, Altostrata? What do you mean?

    Actually, I asked Mr. Whitaker why Wellbutrin was not covered in his book and what he might know about it? Apparently, there were not enough relevant information available. Wellbutrin is supposed to be unique, but not without problems of its own. I don’t think it is “weaker” in any sense of that word. We need to find out more about it. It at a superficial level seems to work more on dopamine, rather than serotonin. It’s also used for ADD, smoking cessation, and sometimes even for cocaine addiction.

    • “Apparently, there was not enough relevant information available. Wellbutrin is supposed to be unique, but not without problems of its own…We need to find out more about it.”

      Isn’t this an indication for more cautious or limited use? How about finding out more about it BEFORE prescribing.

    • The activity of Wellbutrin is more diffuse than SSRIs; its noradrenergic activity is prominent, causing sleeplessness, fast heartbeat, and appetite suppression.

      While side effects are not absent, they tend to be less severe than SSRIs, see http://www.ncbi.nlm.nih.gov/pubmed/11793630

      Buproprion is not as difficult to discontinue, appearing to cause less physiological adaptation than SSRIs.

      Given its lesser firepower, it is not appropriate for “bridging” strategies to assist SSRI or SNRI withdrawal.

      A relative lesser strength does not mean it cannot be effective to alleviate low mood or act as a placebo as well as an SSRI. The methamphetamine-like effect may be welcome by many people.