Tuesday, May 26, 2009
Stimulant drugs like methylphenidate and dextro-amphetamine for ADHD are among the most well-established therapeutic drugs in all of psychiatry (see for example my article, 40 years of methylphenidate treatment of ADHD, J of Attention Disorders, 2002, Vol. 6, Supplement 1). One reason for this general acceptance is a large number of well-controlled trials repeated by many different investigators over the years. But more importantly, frequent clinical use over many years supports the conclusion of the efficacy and safety of these drugs. Though not without some controversy and certain drawbacks, on the whole it is safe to say that most reputable scientists and clinicians find these stimulants to be of substantial value in treatment of an otherwise very refractory disorder. But as the success of these drugs has encouraged pharmaceutical companies to seek even more effective drugs, or to market drugs that provide an alternative for those patients who prove refractory to the stimulants, many novel chemical agents are appearing in new drug trials. In addition, potent new anti-psychotic drugs are being touted as safe and effective. Are the drug trials for these newer drugs adequate, or are there serious loopholes in the methodology that allow ineffective or dangerous drugs to be approved? Here's what often happens in the development of a new psychoactive drug for ADHD. Someone has an idea that a particular molecule, perhaps from a failed trial of some other condition like depression or Alzheimer's or schizophrenia might work with ADHD. An "open label" trial by a proponent is found to produce some favorable response in a few patients by a welll-known clinician treating ADHD, suggesting to the pharmaceutical company that it is worth doing a more carefully controlled trial. (Of course it is financially profitable for the clinician to find positive results because further trials will potentially bring in big bucks, and the pharma company tends to trust the clinical judgments of well-known clinician experts in the field.) Sometimes a well-meaning investigator finds that although the main experiment failed to confirm the efficacy of the drug, a few patients showed a dramatic improvement, warranting a further look. One of my favorite maxims is that "looking for a subgroup in an experiment is a sure sign of a failed experiment." Pursuing a small lead in a failed experiment is like betting on a nag that put up a good effort but lost. But pharmaceutical companies have experts in drug trial methodology who also know that they must conform to the standards set by the FDA, and eventually must fund large trials that employ a standard double-blind design, using approved measures. Typically, this means that patients are randomly assigned to a treatment group and a placebo control group, and assessed with a symptom list of some sort representing the current diagnostic standards, such as a DSM-4 checklist. After getting measures before treatment starts ("baseline") subsequent checklists or scales are filled out based upon an interview with the patients (or in the case of children, their parents and/or teachers). So a parent might respond to questions about hyperactivity, inattention, defiance, or social behavior for example; and the investigator will then fill out his or her own scale, or perhaps a "global" judgment of overall improvement or change. Sounds reasonable. But wait, there are some issues here! What if the parent or teacher or the patient is able to "peek through the double blind," that is to guess which treatment they are receiving? Not that they would consciously cheat, but because it is well-known that the first thing a patient or parent wants to do is please the doctor, or to do what they think the investigator or doctor wants to find. This is called the "demand character" of an experiment, which every experimenter learns to expect and tries to avoid. But how can the patient or parent peek through and know which treatment is being used? Easy. Every drug produces some subjective effect, perhaps a side effect like stomach pain or nausea, or a racing heart, or a thousand other subtle or obvious bodily changes. When the patient experiences one of these effects, they will often respond in the way their brain expects. This is called the "cognitive potentiation of a bodily feeling." Let me illustrate from a very early experiment carried out by Stanley Schacter. Subjects in a drug experiment were told they would receive a drug that would cause some changes in their mood. One group was told that the drug (epinephrine) would make them feel euphoric, and another group was told that the drug would cause them to feel depressed. Now epinephrine is a stimulant that typically will increase heart rate, but the behavioral effect that occurred in this experiment depended upon which expectation the subjects had about the drug, and they acted accordingly. Even though epinephrine produces a number of peripheral signs of stimulation, patients interpreted those signs as if they got a depressant, not a stimulant. Another favorite experiment was with medical students who were asked to drink coffee and then have measures of heart rate, respiration, and skin temperature taken. One group saw the experimenter brew the coffee from a brand coffee jar (Folger's as I recall), and the other group saw that the jar was labeled as decaf. In fact, both groups got only decaf. But the supposedly "loaded" coffee group had increases in heart rate, respiration and finger temperature, while the "decaf" group showed no changes. This study illustrates how the subjects' expectations can manipulate their own bodily responses. Returning to ADHD, we now see that when an investigator uses the report by a patient, a parent, or a teacher, there is the possibility that "improvement" will simply be the desire to please the investigator as informed by their own knowledge of the "active" treatment condition. Thus, the double-blind is leaking all over the place. But there is an even more deadly threat to the integrity of these experiments. In many cases it is the most senior member of the investigating team, the accredited specialist or doctor, who records the signs of improvement. This is so because there is always the possibility of significant "adverse events" or AEs caused by the drug, and the senior investigator must be vigilant in detecting and reporting these AEs for the safety of the patient as well as for eventual calculation of the risk and benefit of the drug. So in these trials, the dose of the drug will usually be adjusted to get a good response without undue side effects or AEs. But if the doc is knowledgeable about the side effects, isn't he or she also then apprised of which drug the patient is receiving? Obviously. Well then, even the honest doc might unconsciously want to please their sponsor by finding positive results. And of course if it is an openly greedy doc, what would you expect? Funny, though, when I raise these objections in pharmaceutical meetings about the drug results, nobody seems to care. The drug companies persist in believing that the investigator is objectively recording their observations as far as improvement is concerned, while nevertheless being apprised of the side effects. And the investigators themselves never question their own objectivity despite the clear knowledge of who is being treated with what. These problems might not matter much with the proven stimulants because the efficacy is often so dramatic; but when a more subtle drug effect occurs, who knows what might happen? In fact, I know of more than one such drug that got approved but began to falter after a longer period of clinical use showed the true lack of efficacy. The old maxim in psychopharmacology is worth remembering: "Use a new drug quickly, while it still works." This is not to allege conscious impropriety. Investigators may truly believe they are objective and drug companies are simply following the models set up as appropriate by the FDA. But there are some simple solutions to these dilemmas, but not without cost. One solution is to use an "active placebo," where a dummy drug is used as a control, producing some bodily effects without much behavioral effect. But this solution has its own problems and is seldom used. Another solution, and the most logical one, is to have one well-trained physician recording and judging the side effects (and therefore changing the dosages during the adjustment phases of the experiment), and a different experimenter/physician recording the improvement. This requires careful scripting of the protocol for inquiring about improvement, and careful shielding of one investigator from the other colleague. I often wonder how much subterranean motives influence the investigator, who stands to profit from a successful trial, and the drug company which seeks FDA approval and a rich reward for their investiment. In general I choose to believe in the honesty and integrity of both the docs and the companies. But I also believe in the deeper effects of "demand qualities of the experiment," and the power of placebo when potentiated by cognitions, therefore that we must employ the most rigorous protections possible in our experiments on behalf of our trusting and troubled patients.