Finding Out What Works Best for YOU - Not the Mythical Average Depressed Patient
Yesterday, I advanced a case for making a clear distinction between agitated and vegetative depression. In a nutshell, we are talking about too much vs too little, as in too much emotion (anger, sadness) vs too little (psychic numbness), or too much thinking (eg over-ruminating) vs too little - on and on. We have a major paradox in action here: Even though we are talking about two conditions with entirely opposite symptoms, we would acknowledge both as depression. Depression is that kind of illness.
The problem is that psychiatry does not acknowledge this distinction. The standard depression screening tests (such as the HAM-D and MADRS) are designed to measure for severity of depression - that is all - while the DSM is designed to assess whether you have depression or not, little more. (Yes, the DSM does list subtypes such as “atypical depression,” but these subtypes are ill-defined and subservient to the “depressive episode” symptom list, not independent of it.) Thus, a typical antidepressant clinical trial will recruit patients who meet the standard DSM criteria for depression and whose HAM-D scores are at least 20. A successful trial - one that would impress the FDA - would have a result in which at least half the patients taking the test drug had their HAM-D scores reduced by at least half.
Here’s the catch: Nearly all the results are the same - about 50 percent of the patients in the antidepressant group getting 50 percent better. We never - ever - have a wildly successful trial (such as 80 percent of the patients getting 80 percent better) or a spectacular failure (such as 20 percent of the patients getting 20 percent better.
In fact, you can make a case that it doesn’t matter what the test drug is - the result will always be pretty much the same. Indeed, Irving Kirsch of Harvard has made a bullet-proof case to this effect. In the 15 years since he reported his first results, no academic researcher has succeeded in shooting holes in it. What Dr Kirsch did was analyze all the antidepressant clinical trials in the FDA database, not just the published ones (which of course are biased in favor of positive results). The numbers don’t lie: There was no clinical difference between the antidepressant results and the placebo results.
Here is the money quote from his 2003 analysis of 47 antidepressant trials: “Overall, the drug/placebo difference was less than 2 points on the HAM-D.”
The result represents some kind of weighted average or mean. In some trials, there is a clear separation from the placebo, which the drug companies take to the bank. In most, though, we are talking about a virtual dead heat.
Does this mean that antidepressants are essentially placebos with side effects? Years ago, I emailed Dr Kirsch with this question and he responded in the affirmative. But there is another interpretation, namely that clinical trials have no idea what they are measuring. This is the view of Gordon Parker of the University of New South Wales, who argues that depression is an umbrella diagnosis that covers a whole host of conditions. In an article published in 2004, Dr Parker pointed out that when we indiscriminately lump all “depression” patients together, with no regard to critical distinctions, it is no surprise that clinical trial results always come out pretty much the same.
This applies whether we are testing medications or talking therapies. You can almost post the results in advance - 50 percent of the patients in the treatment group getting 50-55 percent better vs 40-45 percent of the patients in the placebo group. What does that tell us? Next to nothing.
Dr Parker has directed a lot of his attention to parsing out the differences between “melancholic” and “non-melancholic” depression. These distinctions are beyond the scope of this piece (later, I promise). What is important is that Dr Parker has taken a depressed population and divided them into sub-populations. A recent pilot study of his affords an excellent example:
In the study, Dr Parker identified a group of melancholically depressed patients, 18 which were then treated with antidepressants and 11 with cognitive-behavioral therapy. There was no placebo group. After 12 weeks, the melancholically depressed patients on antidepressants showed a 61 percent improvement in their HAM-D scores vs 34 percent taking cognitive therapy.
Keep in mind this was a pilot study, which means we should look to what the results suggest, rather than what they may prove. Nevertheless, the suggestion is fairly eye-popping, namely: When we study a particular sub-population of depressed patients - not the entire population - the trial results begin to tell us something rather than nothing. Here we see a result in which one treatment was shown to be twice as robust as the other - not about the same, not marginally better.
Had this been the usual comparison between two groups of “depressed” patients, both treatments would have shown the same results. Past CBT and antidepressant trials on “depressed” patients bear this out.
Let’s quickly review:
- We tend to regard depression monolithically, as a single illness requiring the same treatments. One size, in effect, is supposed to fit all.
- Not surprisingly, the monolithic view leads to a state where one size fits no one. This is why clinical trial results are so disappointing.
- We would be better served by slicing and dicing a huge and diverse depression population into various discrete sub-populations. Yesterday, I proposed an agitated/vegetative distinction. Dr Parker offers melancholic/non-melancholic. Another differentiator would involve stress and trauma (Charles Nemeroff of the University of Miami has done important work here). On and on. Pilot studies bear out the value in slicing and dicing.
- Research into sub-populations would take a lot of the guesswork out of treatment. We know - on average - that antidepressants are not particularly effective in treating “depression.” But for certain people, they may prove immensely beneficial. The catch is we haven’t conclusively identified these people.
- Likewise, we would be able to identify who should NOT be taking an antidepressant or receiving CBT in certain situations.
To sum up, the quest is to find out what works best for YOU, not some mythical average patient. We have a long way to go.
As always, you - patients and loved ones - are the real experts. I am looking forward to hearing from you. Comments below ....