 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
COMMENTARY |
Department of Statistics
Carnegie Mellon University
Pittsburgh, PA 15213
Office of Clinical Sciences
Columbus Children’s Research Institute
The Ohio State University
Columbus, OH 43205
Abbreviations: FDA, US Food and Drug Administration • RCT, randomized, controlled trial
In this issue of Pediatrics, Leslie and colleagues1 review the evidence, deliberations, and recommendations of the US Food and Drug Administration scientific advisory committees that led to the black-box warning on the use of antidepressants for children and adolescents. Their article raises a number of critical questions about how we got to this point, including questions about federal drug-regulatory processes and the motivation (or lack thereof) of pharmaceutical companies to design, implement, and report scientifically rigorous trials. However, no question is more compelling than the one faced by many primary care clinicians: How should young patients who are currently on, or potentially in need of, antidepressants be treated?
Although the increased risk of suicidal behavior and ideation for children and adolescents who use antidepressants is statistically significant, unlike Leslie and colleagues, we would not characterize this association as causal for suicidality. In our opinion, the evidence on which the committees based their assertion of a causal link does not yet meet the standards usually required for establishing a cause-and-effect relationship in medicine. Nevertheless, the increase in risk looks real, the black-box warning is in effect, and the critical question that confronts us now is: "Where do we go from here?"
We suggest several areas in which new research is necessary to advance our understanding of the safe and effective use of antidepressants in children and adolescents. First, we need to identify subgroups of children, based on pretreatment risk factors for suicidality, for whom the use of antidepressants might be contraindicated. Second, we need to identify mechanisms of action that explain the relationship between antidepressant use and the increased risk of suicidal behavior and ideation in children. Finally, even without knowledge of the mechanisms by which risk is increased, we need to identify ways to implement close monitoring and careful dose titration of patients, especially in primary care settings to avoid adverse effects and relieve symptoms. In general, not nearly enough research has been done on the delivery of treatments and services for young persons on psychiatric drugs.
Leslie and colleagues identify flaws in the current system of medication testing and approval and suggest regulatory changes that include encouraging larger, longer, and more representative randomized trials. It is not clear to what extent such information would have altered the regulatory review of the efficacy and safety of antidepressants in children, even if all the trials were uniformly well designed. Although the pivotal randomized trials required by the FDA for approval of new drugs are the time-honored gold standard for documenting efficacy, individual randomized trials likely have little value in identifying safety concerns, particularly for rare or delayed-onset events, nor are they efficient in identifying subpopulations at special risk or benefit. Additionally, the generalizability of such trials to routine community practice is a critical issue. The patients that participate in these studies are typically not representative of the general patient population, nor are the research protocols representative of the complex situations seen in routine practice. It seems that answers to the sorts of questions we are asking today are unlikely to come from single trials, even if they are larger.
Where, then, will evidence come that will help us get beyond the current black-box warning? At least part of the answer is likely to be found in the expanded use of methods for research synthesis. It is perhaps self-evident, but significant nonetheless, that the FDA turned to meta-analysis to quantify the risk of suicidality that results from antidepressant use.2 Research syntheses, such as meta-analysis, play a major role in the formal evaluation of scientific evidence by attempting to integrate empirical research for the purpose of generalizations. Perhaps the single most important strength of research synthesis over individual studies is the ability to investigate the robustness of a relationship across study-level covariates. Nevertheless, combining information from a number of similar randomized trials, as done in the FDA meta-analysis, often has some of the same limitations as evidence generated from a single randomized trial, such as restrictive study inclusion/exclusion criteria, limitations on generalizability, and narrowly focused questions.3
Although meta-analyses have typically focused on combining data from a single study type, such as randomized, controlled trials (RCTs), we believe that evidence that will inform the types of clinical research questions that have been posed here will come from the synthesis of information from multiple data sources including randomized and nonrandomized studies. Examples of the latter include administrative databases (such as claims data), epidemiologic studies, and health-survey data. Such data sets contain tremendous amounts of information on large numbers of patients in much broader settings than found in the typical RCT. Valuck et al,4 for example, using insurance claims and prescription fills, followed over 24000 adolescents newly diagnosed with major depressive disorder to investigate the relationship between antidepressant use and the risk of suicide attempt. It is interesting to note that after adjusting for confounders, these investigators found no such association. Experience with combining information from observational studies with results from randomized trials is limited and not without challenge. However, advances in statistical methods, such as the use of Bayesian hierarchical models5 and methods that capture the diverse strengths of the different study designs while minimizing their weakness by adjusting for selection and confounding effects,6,7 have begun to address a number of practical problems in the implementation of this more inclusive synthesis paradigm. Still, more methodologic work is needed.
Once again, depressed and anxious youth have limited options for treatment. To get beyond the black-box warning on the use of antidepressants for children and adolescents, we have suggested several open research questions to pursue. These are research questions that won’t be answered by RCTs alone. Integrating results from efficacy, effectiveness, practice, and service-system research together means combining data not just from different studies but from different types of studies. Because observational studies are often the only feasible way to collect data on rare events or primary care practices, they are likely to be essential, in combination with RCT data, in answering these questions. This approach promises to elucidate what is safe, what works for whom, and when and where it works. Additionally, combining information from a multitude of data sources holds great promise for better informing policy makers and regulators in their assessments of the efficacy and safety of health interventions. Finally, and at least as important, the use of research syntheses promises to provide the evidence base needed to inform primary care clinicians on how to better manage their patients.
 |
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Address correspondence to Kelly J. Kelleher, MD, MPH, Office of Clinical Sciences, Columbus Children’s Research Institute, 700 Children’s Dr, J1401, Columbus, OH 43205. E-mail: kellehek{at}ccri.net
No conflict of interest declared.
|
REFERENCES |
|---|
 TOP REFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. Heneghan, A. S. Garner, A. Storfer-Isser, K. Kortepeter, R. E. K. Stein, and S. McCue Horwitz Use of Selective Serotonin Reuptake Inhibitors by Pediatricians: Comparing Attitudes of Primary Care Pediatricians and Child and Adolescent Psychiatrists Clinical Pediatrics, March 1, 2008; 47(2): 148 - 154. [Abstract] [PDF]
|
|
|
|
 |
|
 E. E Kaizar, J. B Greenhouse, H. Seltman, and K. Kelleher Do antidepressants cause suicidality in children? A Bayesian meta-analysis Clinical Trials, April 1, 2006; 3(2): 73 - 98. [Abstract] [PDF]
|
|
|
|
 |
|
 L. M. Wegner Pediatricians and Antidepressant Medications: Black Box or Black Hole? Pediatrics, July 1, 2005; 116(1): 233 - 235. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2005 American Academy of Pediatrics. All rights reserved. |
||
|
||
