Monday, June 22, 2015
There has been an increasing debate about whether doctors are using "evidence-based medicine" or instead are merely going by conventional wisdom or listening to pharmaceutical company sales pitches. Often, as I have discussed several times, the so-called "evidence base" consists of randomized controlled studies (RCT's), while clinical experience is written off as "anecdotal."
An anecdote is a report of a single, or at most a few, specific incidents, such as a patients' seeming to get better after taking a medication, along with the interpretation of the event by an allegedly biased observer. Besides the fact that one or two cases may not be representative of a psychiatric condition, and that other causes for the observed effect have not been ruled out, the description provided by the source of the anecdote may be incomplete or incorrect. And his or her conclusion may be logical, or it may be fallacious in any of a variety of different ways.
Widespread clinical experience — or the clinical experiences of a wide variety of practitioners who are known to do careful diagnostic work-ups and to follow patients closely — is somehow also written off as "anecdotal," but this is just nonsense.
First of all, as I discussed in a previous post, in psychiatry there are almost no objective diagnostic blood tests or direct measurements of brain function that can tease out the difference between neuropathology and normal neural plasticity in response to environmental factors. Therefore, conclusions drawn in RCT's are based entirely on the self-report data from subjects or by the potentially biased observations of the experimenters. They are just as much anecdotal as widespread clinical experience in that sense.
Actually, widespread clinical experience is better than most RCT's in determining the efficacy of drugs. It employs a sample size far larger than all the RCT's on a treatment put together. Also, there are several important limitations with RCT's.
In a paper by G. Parker and S. McCraw (Acta Psychiatrica Scandinavia, 2015: 1-10) about the drug Lamictal (lamotragine), they discuss the disconnect between the reported efficacy of a new psychotropic medication as quantified in RCTs and its actual effectiveness as observed in the 'real world' of psychiatric practice:
"Most commonly any such disconnect is one of degree, with the medication being progressively clinically judged as less or more effective. Less commonly, the medication may be judged to have a different therapeutic 'signal' than for its formal indication. Two examples of the latter are the selective serotonin reuptake inhibitors (SSRI's) which seemingly modulate emotional dysregulation as much as they have antidepressant propensities, and some atypical antipsychotic drugs having utility in augmenting antidepressant medications and in having mood stabilizing propensities rather than being confined to the management of psychotic conditions.
Any such disconnect can reflect multiple factors, as considered now in relation to antidepressant medications. An antidepressant's 'efficacy' is evaluated from RCTs with the 'control' treatment being either a placebo or a comparator drug. In such trials, the duration is often limited to several weeks, the sample constrained by inclusion and exclusion criteria (e.g. non-suicidal, no substantive co-morbid conditions) that do not hold in clinical practice, and which may, depending on recruitment strategies, be weighted to those with potentially spontaneously remitting and/or less severe conditions.
RCT-evaluated efficacy may he overestimated using a lower-than-usual dose of any comparator medication, or underestimated if the antidepressant is prescribed at what might be later determined to be a suboptimal dose. Biases may emerge if the sponsoring developers provide data only on trials generating superior findings. "
Richard Horton, editor of the medical journal The Lancet, recently observed, “The case against science is straightforward: much of the scientific literature, perhaps half, may simply be untrue. Afflicted by studies with small sample sizes, tiny effects, invalid exploratory analyses, and flagrant conflicts of interest, together with an obsession for pursuing fashionable trends of dubious importance, science has taken a turn towards darkness.”
In studies used to evaluate medications for treatment of various mood disorders, there is another glaring problem: the use of inadequate and incomplete diagnostic evaluations of the subjects. I have written several posts about the nonsensical expansion of the diagnosis of bipolar disorder, in which researchers lump together Bipolar I with the bogus disorder Bipolar II.
Readers of this blog know my attitude about the latter: In my practice since I started training, I have seen four patients who actually met the DSM criteria for this supposed disorder when they were interviewed carefully and followed closely, and all four responded to lithium, suggesting that they were just milder cases of Bipolar I. And most patients who were given that diagnosis by previous psychiatrists did not have bipolar disorder at all but had the mood instability characteristic of certain personality disorders. Some studies have shown this to be true.
Worse yet, some studies include patients diagnosed with unspecified bipolar disorder and "bipolar spectrum disorder." Neither is defined clearly nor is there any agreement on exactly with what criteria these diagnoses are to be made. So the RCTS's are basically comparing apples to oranges.
This particular problem is not discussed clearly in the Parker and McCraw article. In fact, they conclude that the anticonvulsant drug Lamictal (lamotrigine) is probably not effective in bipolar I disorder, but probably is effective in bipolar II disorder - which probably means it may help the affective instability of patients with personality disorders. This puts this drug as a possible alternative to SSRI's, which the authors themselves note "... seemingly modulate emotional dysregulation." Emotional dysregulation and affect or mood instability are basically the same thing.
(Combining SSRI's and a long acting benzodiazepine such as clonazepam is even more effective in "raising the bar" on the strength of environmental stimuli required to set off an episode of affect dysregulation, as well as in decreasing self-injurious behaviors like cutting. There are no RCT's on this combination for these symptoms; the drug companies won't do them because most of these drugs are generic and the drug companies probably know that it is effective but do not want docs to know this. However, widespread clinical experience by those doctors who know the difference between bipolar disorder and borderline personality disorder backs me up on this).
The history of the FDA approval for the drug Lamictal for bipolar disorder is bizarre. Its manufacturer first touted it as a treatment for the depressive episodes in bipolar disorder, although later studies showed it to be ineffective in the acute phase of bipolar depression. Then, when it got the FDA approval for psychiatric use, rather than being given the indication for prophylaxis for (prevention of) episodes of bipolar depression, it was given a general indication for bipolar disorder as a whole. This, even though there was zero evidence that it prevented episodes of mania, or that it was useful in acute mania.
In fact, most of the studies in bipolar disorder with the drug were based on a rather flawed outcome measure. The drug was found to delay, but not to prevent, the emergence of another bipolar episode. Whether the episode was a depressive episode or a manic episode was not specified in the majority of these studies!
Furthermore, lithium — the standard prophylactic treatment—when used at the dosages which lead to the correct blood levels of the drug in patients who respond to and can tolerate it (about 80% of bipolar I patients) completely prevents the re-emergence of manic episodes. In those cases, the measurement "time until the next manic episode" would essentially be the patient's entire lifetime.
Delaying episodes of bipolar disorder is better than having them more frequently, but why would you use a drug to do that when there is another drug that can prevent them from happening completely?