The Institute of Medicine’s Forum on Neuroscience and Nervous System Disorders planned and held a public workshop June 16, 2009, that brought together experts from industry, academia, government, and advocacy groups to discuss issues directly related to a recent Food and Drug Administration (FDA) policy that all clinical protocols for products developed in the Division of Psychiatry Products (of the FDA) include a prospective assessment for suicidality.2 Given the focus of the Forum, participants were charged with examining and discussing currently available data, data analysis, and the future of potential partnerships that will be or are being impacted by this announcement as it relates to clinical trials involving the nervous system. Discussions centered on the critical areas of further examination and analysis needed.
The planning committee’s role was limited to planning the workshop, and the workshop summary has been prepared by the workshop rapporteurs as a factual summary of what occurred at the workshop.
The definition of “suicidality” used for the purpose of the workshop is completed suicide, suicide attempt, or preparatory acts toward imminent suicidal behavior. Please refer to Box I-1 and the paragraphs before it for more information.
BACKGROUND ON THE ISSUES
Suicide is a serious, yet tragically underrecognized, public health problem that claims approximately 30,000 lives each year in the United States. Cast in its societal and historical context, the rate of suicide for the past century has been two to three times that of homicide. Worse yet, suicide is one of the foremost causes of death among children and young adults (IOM, 2002). Depression and other mood disorders confer the highest risk of suicide (Fawcett et al., 1991; Goodwin and Jamison, 2007; Harris and Barraclough, 1997). It has been estimated that 35 to 50 percent of depressed children make a suicide attempt, and 2 to 8 percent commit suicide over a 10-year period (Fombonne et al., 2001; Kovacs et al., 1993; Rao et al., 1993; Weissman et al., 1999).
Given the severity of the problem, the public health profession, after decades of effort, has developed successful programs to reduce suicide among high-risk groups (Brown et al., 2005; Guzzetta et al., 2007; Knox et al., 2003; Lauterbach et al., 2008; Rutz et al., 1989, 1992; Szanto et al., 2007). Meanwhile, the public health profession also has wrestled, since 1991, with the controversial possibility that the very medications given to combat depression and other mood disorders paradoxically may contribute to the risk of suicide ideation (i.e., suicidal thinking) and suicidality (i.e., suicidal attempts, preparatory behaviors, or suicide completions) in a minority of cases. Researchers and regulators began in earnest to examine the evidence systematically through randomized and controlled trials.
As more data and analysis came to light in 2004, the FDA issued its first set of warnings by calling attention to an increased risk of suicidality among children and adolescents taking antidepressants. The FDA required that labeling of specific antidepressants carry black box warnings, intended to alert physicians and patients to increase monitoring of troubling symptoms. One concern at the time was the surge in antidepressant prescriptions by primary care physicians with insufficient oversight of patients (IOM, 2002). The United Kingdom’s drug regulatory agency, the Medicines and Healthcare Products Regulatory Agency (MHRA), used its Committee on Safety of Medicines to explore the issue in 1998 and eventually banned the use of antidepressants3 for children and adolescents with mild cases of depression. In 2003, the MHRA stated that “on the basis of this review of the available clinical trial data, CSM has advised
that the balance of risks and benefits for the treatment of major depressive disorder (MDD) in under 18s is judged to be unfavourable for sertraline, citalopram and escitalopram and unassessable for fluvoxamine. Only fluoxetine (Prozac) has been shown in clinical trials to have a favourable balance of risks and benefits for the treatment of MDD in the under 18s” (MHRA, 2003).
At the time of FDA’s regulatory actions, in 2004, many public health professionals feared that a black box warning would deter providers from prescribing antidepressants, which in turn could lead to more suicidality from untreated or undertreated depression. One study found a 20 percent reduction in physician prescribing in the United States from 2003 to 2005 and an increase in the youth suicide rate from 2003 to 2004 by 14 percent (Gibbons et al., 2007). New diagnoses of depression by primary care providers were reduced by 44 percent in pediatric populations, based on time series analysis (Libby et al., 2009).
Public health professionals agreed that the rarity of completed suicide presented a daunting methodological challenge: No single randomized clinical trial (RCT) of an antidepressant or any other medication, however well designed, has the power to detect such a rare event as completed suicide. This point was borne out by finding no completed suicides among nearly 4,600 pediatric subjects in clinical trials submitted to the FDA. The suicide rate among children and adolescents, ages 5 to 17 years, is approximately 1.8 per 100,000 (CDC, 2006). Another problem arose from confounding of variables, observed Robert Gibbons, University of Illinois at Chicago and workshop co-chair, who remarked that “depression is related both to suicide and to treatment for depression … suicide [also] can lead to treatment, the very same treatments that are suspected of potentially producing suicide.” In addition, patients with suicidality are normally excluded from clinical trials. Gibbons questioned if, given all these methodological factors, plus others discussed subsequently, the study population is representative of the general population of patients being treated and whether an association exists that is free of bias and includes adequate control of confounding.
To help overcome methodological hurdles in measuring suicidality in central nervous system clinical trials, one obvious approach is to increase the number of patients under study by pooling each study’s findings. Yet pooling is most methodologically rigorous when the outcome measures across studies are standardized, which was not the case—studies had varying definitions of outcome measures. Affirming the depth of the problem of measuring and tracking suicidality poses a significant challenge, given the low sample sizes and methodological
problems in pooling patients together, said William Potter of Merck Research Laboratories, co-chair of the workshop. He stressed that quality control, standardized measures, and training are essential for pooling studies. Age group is also a crucial issue. Studies in adults may be inapplicable to children and adolescents, who often manifest or express symptoms of a psychiatric disorder in ways that diverge from adults. Thus, diagnosis itself may be problematic.
Common definitions of suicidality are essential in clinical practice and in defining outcome measures in clinical trials, Potter commented. Referring simply to “suicidality” can become complicated because different researchers may exclude various behaviors in the definition. The FDA uses C-CASA (the Columbia Classification Algorithm for Suicide Assessment) terms to define “suicidality” as attempt or completion as well as thinking about suicide (suicidal ideation), or preparatory acts toward imminent suicidal behavior if intent is to die (Posner et al., 2007). Others exclude ideation, thereby defining suicidality more restrictively as an attempt, preparatory acts toward imminent suicidal behavior, or completion.
The following definition was used for the purpose of the workshop for consistency (Box I-1). One practical reason for excluding suicidal ideation in the definition of suicidality was that one of the key goals of the workshop was to determine whether suicidal ideation has predictive power as a surrogate measure for suicidality. When a speaker uses a different definition of suicidality, it is noted in the text.
Definition of Terms Used During Workshop
Suicidal ideation refers to thoughts of suicide.
Suicidality, for the purpose of this workshop, refers to completed suicide, suicide attempt, or preparatory acts toward imminent suicidal behavior.
ABOUT THE FORUM AND WORKSHOP
As background, the Forum was expressly created by the IOM to bring together the public and private sectors, among other stakeholders, to discuss topics of critical and overarching importance, particularly ones that stimulate partnerships to accelerate understanding and treatment of nervous system disorders. The Forum has been convening stakeholders, sponsoring workshops, and producing short-turnaround workshop summaries since its inception in 2006. This summary of the workshop on measuring suicidality during clinical trials, which was held on June 16, 2009, offers guidance and insights from participants at the workshop. In accordance with IOM policy, the workshop is designed to seek different views but preclude participants from making explicit recommendations or reaching consensus.
The Forum planned and held the public workshop to identify effective methods to predict suicidality, in the near term,4 during the conduct of clinical trials (seen Appendix B for full workshop agenda). Speakers and attendees included experts from industry, academia, government, and/or advocacy groups. They were asked to present data and stimulate discussion about the goals set by the steering committee as listed in Box I-2.
Workshop Goals for “CNS Clinical Trials: Suicidality and Data Collection”
The overall purpose of the workshop on central nervous system (CNS) clinical trials is to examine what methods are best used to determine treatment-emergent suicidal behavior during the conduct of clinical trials. More specifically, attendees were asked to:
An important set of discussions grew from the workshop goals and was heavily emphasized at the meeting. Participants reviewed the classification instrument known as C-SSRS, the Columbia Suicide Severity Rating Scale, which does not have as its purpose the prediction of future suicide attempts or completed suicides but rather is used to systematically detect, or ascertain the occurrence of, and document events of suicidality as defined by C-CASA, which is discussed in Chapter 1. This subtly distinct focus allowed participants to emphasize the importance of capturing suicidal occurrence and non-suicidal occurrence. Elaborating on the specific goal of predicting suicidality, however, Gibbons asked whether suicidal ideation is an appropriate surrogate endpoint for suicidal behaviors and completion. “What information do we need,” he asked, “ … to determine whether or not suicidal thoughts, which are so ubiquitous among depressed people, are in fact good predictors of suicidal behavior and completion?”
Given the nature of the subject, it is important to remind readers that this workshop summary is a record of what occurred at the workshop. Many important discussions are needed in order to fully explore the statement of task, and while many of these issues were brought up and discussed, there are also a number that were not. This summary is by no means a complete review of the extent to which emergent suicidal ideation may or may not predict the occurrence of suicidal behavior. What is contained in this summary is a review of the presentations and discussions that took place during the workshop.