Issue: Evidence-based medicine that derives from studies of patient populations selected to gain marketing approval for new psychopharmacologic drugs is increasingly distinct from the experience of treating patients in clinical practice.
|Trying to use clinical
trial data as a beacon to guide the use of new drugs in
clinical practice can lead to errant expectations.
Consider the story of the police officer who encountered
a man looking for something near a street light after
"What are you
doing?" asked the police officer.
Just because the light of research shines on a population of patients in a clinical trial doesn't always mean it will illuminate patients in clinical practice. Why is this so?
You Can't Get There From Here
Getting from clinical trials to clinical practice can be difficult. Comorbidity of psychiatric disorders may be the rule rather than the exception in clinical practice today, yet such patients are generally excluded from clinical trials. In terms of health care utilization and disability, comorbid patients are clearly the most important to study, e.g., 14% of psychiatric patients have 3 or more comorbid disorders, yet they consume almost half of all medical care resources.1,2 These patients may be more resistant to drugs, represent a different genetic population (i.e., genotype), and require polypharmacy. Neglecting this population prior to marketing a new drug may be commercially motivated owing to methodological complexities in dealing with simultaneous and interacting conditions and treatments, fear of "niching" the drug out of the broad market, and the lack of leadership among many, but not all,1,2 experts who are often strangely silent about the need for such studies as part of the drug-approval process.
Come One, Come All Step Right up to the Clinical Trial
Currently, patients for clinical trials in psychopharmacology are often recruited from advertisements, sometimes with monetary incentives and no prior relationship to the investigator, and then "purified" of comorbidity and extreme severity for randomization into placebo-controlled trials. By contrast, in years past, patients in clinical psychopharmacology trials were frequently recruited from an investigator's clinical practice by utilizing diagnostic criteria that had not necessarily been well developed, especially for numerous anxiety disorders.3 Thus, patients who had a prior relationship with the investigator and who may have had unrecognized common comorbidities, such as social anxiety disorder, panic disorder, and other anxiety disorders, were probably included in clinical trials.
|As time has passed, not only
has the clinical trial population morphed, but so have
clinical investigators. On the one hand, patients have
become more sophisticated about their illnesses, often
self-diagnosing, and more knowledgable about treatments,
clinical research, and the placebo effect. They also may
have participated in prior trials and present at a milder
stage of illness in response to an advertisement for a
study. Often excluded are not only patients with
comorbidities, but also a host of other patient types:
depressed patients who are psychotic, bipolar, suicidal,
medically ill, taking concomitant drugs, and treatment
resistant; manic and schizophrenic patients who are too
ill to consent or who abuse drugs and alcohol;
Alzheimer's disease patients with early illness for whom
early drug intervention may be the most valuable;
children and adolescents; women taking estrogen
replacement therapy; and many ethnic groups.
On the other hand, clinical investigators have been increasingly recruited from both academia and clinical practice as clinical revenues have dried up and opportunities for clinical trials have multiplied. Raters of the symptoms of patients have multiplied in number, including many with little training, experience, or demonstrated reliability. Pressure to rapidly complete trials has resulted in the migration away from proven models of the past, in which a few investigators were held accountable for the trial results and many patients were enrolled per site, to a model with many investigators and only a few patients enrolled per site.3 Enrollment rates along with placebo-response rates have increased, leading to increased variance in results such as many failed clinical trials and positive clinical trials of patients not necessarily representative of those in a clinical psychiatrist's practice.
Pharmacogenomics May Be the Long-Term Answer
It is probably unrealistic to return to the "good old days," so how can a clinician apply clinical trial data to clinical practice? For one thing, the advent of agreed-upon conventions on how to use rating scales accompanied by formalized rater training and certification is improving interrater reliability as raters and investigative sites proliferate. For another, the placebo-response phenomenon is finally being seriously studied in relation to clinical trial populations, which may lead to enrollment of patients more representative of those seen in clinical practice. Ultimately, however, patients' genes may actually be the determining factor as to whether a drug will be approved as safe and effective (pharmacogenomics), i.e., the drug approvals of tomorrow are likely to be for psychiatric symptoms associated with a specific portfolio of genotypes and not just DSM-IV symptom clusters.4,5
|Indiana Jones, but not
Dirty Harry, May Be the Short-Term Answer
What is a clinician to do before pharmacogenomics is a reality? In other words, to what extent can the evidence from clinical trials help the prescriber treat patients largely excluded from clinical trials once a new drug is approved for marketing?
One possibility for the modern-day prescriber is to look at the data, identify the differences between evidence based on these patients and those in practice, learn the mechanism of action of the new drug, stay vigilant to case reports and early anecdotes, and then embark on a targeted prescribing mission, patient by patient--"making it up as you go," as did the successful and serendipitous adventurer Indiana Jones in Harrison Ford films. Such a prescriber will be skeptical of the generalizability of new data,5,6 knowing that some patients tolerate or respond better to one drug than to another, so that the median patient's response and tolerance to a drug in clinical trials are perhaps poor predictors of any given individual's response or tolerance in clinical practice. Best avoided is a reckless approach, forging ahead, dismissing the data and hoping to be "lucky," as did the renegade cop Dirty Harry in Clint Eastwood films.
The art of psychopharmacology derives from the science of psychopharmacology, but still requires wisdom, judgment, and experience to translate findings from clinical trials of a new drug into clinical practice.
1. Regier DA, Narrow WE, Rae DS, et al. The de facto US Mental and Addictive Disorders System. Arch Gen Psychiatry 1993;50:85-94
2. Judd LL, Kessler RC, Paulus MP, et al. Comorbidity as a fundamental characteristic of generalized anxiety disorder: results from the National Comorbidity Survey (NCS). Acta Psychiatr Scand 1998;98(suppl 393):6-11
3. Robinson DS, Rickels K. Concerns about clinical drug trials. J Clin Psychopharmacol 2000;20:593-596
4. Stahl SM. New drug discovery in the postgenomic era: from genomics to proteomics [Brainstorms]. J Clin Psychiatry 2000;61:894-895
5. Siever LJ. Antidepressant selection in the postgenomic era. Biol Psychiatry 2000;48:875-877
6. Stahl SM. Placebo-controlled comparison of the selective serotonin reuptake inhibitors citalopram and sertraline. Biol Psychiatry 2000;48:894-901
Brainstorms is a monthly section of The Journal of Clinical Psychiatry that provides updates of novel concepts emerging from the neurosciences that have relevance to the practicing physician.
From the Neuroscience Education Institute in San Diego and the Department of Psychiatry at the University of California San Diego.