NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Forum on Drug Discovery, Development, and Translation. Transforming Clinical Research in the United States: Challenges and Opportunities: Workshop Summary. Washington (DC): National Academies Press (US); 2010.

Cover of Transforming Clinical Research in the United States

Transforming Clinical Research in the United States: Challenges and Opportunities: Workshop Summary.

Show details

5Clinical Trials in Depression

Depression is a chronic disease characterized by recurrent episodes that interfere with daily life and normal functioning, exacting large costs for both individuals and society. James McNulty, Vice President of Peer Support at the Depression and Bipolar Support Alliance, presented data from the World Health Organization (WHO) revealing that depression is the primary cause of disability in the United States and Canada for individuals aged 15–44 (WHO, 2002). Indeed, depression and other mental illnesses result in a greater loss of healthy life years to disability and death than cardiovascular disease, cancer, or diabetes (WHO, 2004). The onset of mental illness occurs primarily at a young age—by age 24 in 75 percent of cases (Kessler et al., 2005)—but can strike at any age. Regardless of age at onset, a study by the Council of Medical Directors of the National Association of State Mental Health Program Directors showed that individuals who receive treatment for a serious mental illness still die 25 years earlier than the normal population (NASMHPD, 2006). Disconcertingly, similar statistics are not available for those who do not receive care for a mental illness.

The neuroscience knowledge base underlying the study of depression has been growing since the emergence of biochemical pharmacology and molecular technologies in the 1970s and 1980s. Over this same period of time, pharmaceutical companies, the National Institutes of Health’s (NIH’s) National Institute of Mental Health (NIMH), and patient advocacy groups have aggressively pursued new treatments for the disease. The success of selective serotonin reuptake inhibitors (SSRIs) and the many structurally similar drugs that followed improved the lives of many patients. However, William Potter, most recently Vice President of Translational Neuroscience at Merck Research Labs, Merck & Co., Inc., explained that a truly novel antidepressant has not been introduced in the last 40 years. According to Potter, the period of SSRI development established a level of comfort in the mental health community that may have temporarily hindered the development of new and better antidepressants. Today, significant effort is focused on understanding the challenges to developing novel antidepressant therapies and designing the informative clinical trials necessary to test the effectiveness of new discoveries.

This chapter begins with a patient’s perspective on clinical trials in depression. Next, the commercial contract research organization (CRO) model for conducting clinical trials in depression is described. A discussion of the unique issues in conducting clinical trials in depression is then presented. Finally, the chapter summarizes workshop participants’ discussion of specific ways to develop informative clinical trials to accelerate depression research.


As a patient advocate living with the psychiatric diagnosis of bipolar disorder, McNulty shared his perspective on mental health research and the role of patients in clinical trials. Having participated as a subject in both industry- and NIH-sponsored studies, he received SSRIs in the clinical trial setting and experienced life-changing improvements in his condition due to these breakthrough drugs. McNulty described depression as a protean disease—extremely variable and readily assuming different shapes and forms. In his own life, depression had devastating effects, including the loss of his family and business, as well as a period of homelessness. Noting that there is a significant human dimension to the disease, he explained that his success in battling depression has been due only partially to medications. This is an important point to note because, in McNulty’s experience, scientists can become excessively focused on data and lose sight of the real-life manifestation of depression and its effects on individuals and families.

At the age of 20, McNulty experienced his first major depressive episode. In 1985, he received his first diagnosis of mental illness. At one point, he lied about which medication he had previously used so as to become eligible for a clinical trial. Then, after being diagnosed with bipolar disorder type II, he began a period of years during which he searched for stability with medications. His involvement in an NIH-sponsored trial and eventually an Institutional Review Board (IRB) led him to a career focused on national mental health policy and clinical research as the vehicle for answering questions of great importance to the field of mental health.

Referencing the policy debate that surrounds funding for mental health services and the allocation of scarce resources, McNulty noted that it is very expensive not to treat mental illness. Although the true cost of mental illness involves societal and personal costs that are not easily captured in financial terms, the sum of the societal and personal costs of failing to provide care to those who suffer from mental illness is probably greater than the cost of providing the care.

According to McNulty, patient advocates should be involved in developing and conducting clinical trials. Bringing this expertise into the clinical trial process at an early stage could help avoid some of the pitfalls that hinder trials today. An example is the informed consent process. McNulty serves on the board of directors of the Association for Accreditation of Human Research Protection Programs—an organization that resulted from recommendations in the IOM study Preserving Public Trust: Accreditation and Human Research Participant Protection Programs (IOM, 2001b). As a member of the IOM study committee for that report, McNulty joined in recommending that the informed consent process for clinical trial participants be simplified and streamlined. According to McNulty, however, the consent process has been hijacked by lawyers who have rendered consent documents unintelligible to both patients and researchers. The need to simplify the informed consent process was echoed by a number of participants throughout the workshop.

Aligning industry efforts more closely with the real-world needs of patients is another area that could benefit from more patient input. McNulty described his experience in which a pharmaceutical company asked a patient group what the ideal antidepressant would be like. He said that for most patients, the ideal antidepressant would be one that restored their life to a presickness state. When doctors were given the opportunity to answer the same question, some responded that sexual functioning was not important to their patients taking antidepressants. McNulty and the patient group clarified that, of course, sexual functioning recedes in importance when other major symptoms are considered, but it is not unimportant. Such divergences of opinion need to be illuminated early in the process of drug development.


As an executive with the largest global contract research organization (CRO) in the world, Amir Kalali shared his perspective on the role of CROs today. Drawing on his experience in randomizing thousands of patients into global clinical trials, he discussed why, in his view, trials are increasingly conducted outside the United States (see also the discussion of this issue in Chapters 3 and 4).

Clinical trials conducted according to the CRO model are not specifically designed to produce results that can be translated into useful information for clinical practice. Rather, most CRO-run preapproval clinical trials are conducted with the goal of regulatory approval, and complex protocols that involve a number of study questions are often a recipe for failure.

In the area of psychiatry, Kalali stressed the importance of conducting scientifically robust and efficient clinical trials, a theme repeated frequently throughout the workshop. More than any other area, psychiatry has been scrutinized with respect to the drug development process and the role of pharmaceutical companies in the marketing of drugs. High-profile attention has surrounded a number of issues, including antidepressants and suicide, the safety of newly marketed drugs, the dissemination of negative clinical trial data, the lack of evidence-based drug development in psychiatry, and the globalization of clinical trials in this area. These issues have contributed to a decline in the public’s trust of the pharmaceutical industry and the clinical research enterprise. Kalali cited the high-profile withdrawal of pharmaceutical products from the market, as well as scientific misconduct, primarily at academic institutions. For these reasons, clinical research, especially in psychiatry, is under increasing scrutiny.

In addressing the issue of globalization, Kalali spoke to the concern about the applicability of global trial results to the U.S. population by noting that for decades medicines were tested only in America and Western Europe yet used around the world. He highlighted the benefits he perceives in conducting global clinical trials:

  • wider, early patient and physician access to novel therapies;
  • shortened drug development times due to more rapid patient recruitment;
  • reduced drug development costs (i.e., the ability to develop more drugs);
  • the generation of data to address ethnic diversity;
  • study personnel with higher qualifications;
  • accelerated local product approval;
  • the availability of drug-naïve patients;
  • improved patient retention rates;
  • better medication compliance rates; and
  • stability of the patient population, facilitating long-term follow-up.

According to Kalali, the quality of clinical trials conducted globally is very high. The average level of education and expertise of study personnel abroad is higher than that in the United States, and this greater expertise also comes at a lower cost in the global market. Kalali also pointed to notable differences between U.S. and global study populations. For example, given the large number of chemically similar antipsychotics on the market today, most individuals with schizophrenia in the United States have tried a number of medications to treat the disease. If the seventh in a string of antipsychotics with a similar mechanism of action is developed, there is little reason to believe it will be effective for individuals with schizophrenia who have not responded to previous drugs. In contrast, a patient population of individuals with schizophrenia in Ukraine has likely been exposed to various drugs with different mechanisms of action (e.g., haldol, chlorpromazine). As a result, the Ukrainian population might be responsive to the new antipsychotic drug.

Kalali explained that there is no inherent interest in conducting research outside the United States. The United States is the largest pharmaceutical market in the world, and companies are unlikely to abandon U.S.-based clinical trials. The U.S. Food and Drug Administration (FDA) requires that, to gain access to the U.S. market, global clinical trials include a separate U.S. population. This requirement is aimed at the development of U.S.-based data on the safety and efficacy of a drug for the population in which it will be marketed. In selecting the best clinical trial sites, however, it no longer makes sense to choose 50 U.S. sites, 20 of which could be inadequate and fail to enroll patients, when there are significant advantages to choosing sites globally.

Califf responded by questioning whether it is appropriate to market a new drug in the United States that has been tested on a very different patient population, for example, in Ukraine. In addition, he suggested that the solution to improving clinical research in the United States is not to move clinical trials abroad. Rather, U.S.-based global companies, such as Quintiles and many other CROs, should help fix problems with the current U.S.-based clinical trial system.


According to Madhukar Trivedi, Professor and Chief of the Division of Mood Disorders in the Department of Psychiatry at the University of Texas Southwestern Medical Center at Dallas, two of the greatest challenges in depression research are (1) the lack of a definitive marker for diagnosing depression—a pathophysiologic “smoking gun,” as he described it; and (2) the fact that a significant number of trials in depression are not focused on answering the most important clinical questions—that is, there are many uninformative trials in the field.

As was discussed in Chapter 4, the successful classification of acute coronary syndromes allows clinical investigators to identify the appropriate patient population for a study quickly and easily. Similar to the difficulties seen with the diagnosis of heart failure, depression currently lacks a robust mechanism for diagnosis. The current standard for measuring depression in clinical trials is the Hamilton Depression Rating Scale (HAM-D), a rating system based on the subjective determination of the diagnosing physician or other qualified personnel. The result, according to a number of workshop participants, is the identification of patient populations inadequate for distinguishing treatments for depression.

Trivedi further explained that there is very little scientific evidence regarding which patients will respond to which of 25 to 30 treatments for depression. Treatments are similar but may be different in important ways. The result is substantial variation in clinical practice patterns for depression. It is unclear in many cases which antidepressant a clinician should choose for a patient, as well as how long to treat the patient with a given drug.

Identifying a Study Population for a Heterogeneous Disease

Presenters discussed the issues involved in identifying the correct study population for a successful clinical trial in a heterogeneous disease such as depression. First, as suggested above, the diagnosis of depression is less technical and more subjective than that of other diseases. Moreover, changes in the way depression has been diagnosed over time call into question the use of large, historical databases. There is significant uncertainty as to whether a population diagnosed with depression in 1990 would be comparable to a population diagnosed with depression in 2000. Potter further explained how two patients diagnosed with depression at the same time might have very little in common. The criteria for diagnosing depression require that the patient have a depressed mood or a markedly diminished interest in pleasure (anhedonia). Hypothetically, one depressed patient could have a depressed mood as well as weight gain, hypersomnia, and recurrent thoughts of death, while another could have anhedonia, weight loss, and insomnia. Both of these patients would be diagnosed with depression and yet have no symptoms in common, and both could be enrolled in a clinical trial for depression. This treatment of a heterogeneous disease as if it were homogeneous is one reason clinical research in depression struggles to distinguish among antidepressants.

Potter described the difficulty of using entry criteria for clinical trial participants based on the severity of their disease. The standard minimum criterion for enrolling an individual in a clinical trial for depression is a total HAM-D score greater than 18. Conventional wisdom in the psychiatry field is that increasing the severity criterion creates a study population with more severe cases of depression, which in turn reduces the placebo response rate. Potter described his research to better understand the effects of the severity-based entry criterion on trial outcomes. He and his colleagues studied the relationship between patients’ and physicians’ ratings according to the HAM-D over the course of a 6-week trial. Potter said it was not surprising to find in the first week of the trial that physicians rating patients for admittance to the trial gave them at least the minimum score to get them into the study. When patients were asked to rate themselves (via telephone) during the first week of the trial, the HAM-D scores ranged from high to low and were of a normal distribution. As the trial continued, the patient and physician ratings increasingly converged. Kalali explained that after their high initial ratings designed to get patients into the trial, physicians began rating the condition of patients accurately. The result was a large drop in the HAM-D scores, which made the placebo appear effective.

Potter’s research highlights the difficulties inherent in setting a severity criterion for entry into a clinical trial in which the measure of disease is subjective and easily manipulated. The result can be to introduce bias into a study and create significant statistical problems due to the skewed distribution of the patient population. Kalali added that HAM-D is an incomplete measure in that it does not include anhedonia, cognition, or the painful physical symptoms that are important in depression. Potter noted that while research is being conducted to understand the issues surrounding the use of severity criteria for trials in depression, solutions to the problem have yet to be developed.

Trivedi estimated that in the last 3 to 5 years, more than 300 randomized controlled trials (RCTs) for antidepressants have been conducted. He described the patient populations studied in these trials as “symptomatic” volunteers. His example of a symptomatic volunteer is an individual who responds to an advertisement for a clinical trial in depression and who would otherwise not seek treatment for depression outside the trial setting—that is, the clinical trial is the patient’s only interaction with the treatment setting. This unique, symptomatic patient population in which antidepressants are tested is different from the depressed population that will eventually be treated with the drug in the real-world clinical setting. Patients in clinical trials for depression often are not chronically ill and rarely have comorbidities, whereas it is well known in psychiatry that patients with major depressive disorder frequently have a number of comorbidities. In addition, there is overwhelming evidence that a patient with depression is at increased risk for cardiovascular disease, diabetes, and a number of other conditions. Thus, excluding patients with comorbidities from depression trials, as is most often the case, diminishes the applicability of the trial results to the real-world population.

Kalali also addressed the existence of “professional patients” in psychiatry. These patients seek to enroll themselves in multiple trials or study sites at once as a source of money and medicines. Kalali said active efforts to screen out these professional patients have been necessary. In one trial involving 300 patients, for example, 30 were found to be randomized to the same study by separate study sites.

Clinical Trial Methodologies and Placebo Response

The effect of placebo response rates in clinical trials for depression has added a layer of complexity and difficulty to the process of designing trials and interpreting the trial results. Placebo response refers to a patient’s clinical improvement in response to an inactive substance (e.g., sugar pill). Most clinical trials in depression are placebo-controlled. The ethical issue of whether placebo-controlled trials should be conducted when effective therapies are available remains contentious. McNulty noted that while placebo-controlled studies are not required by the FDA, it is difficult to design a trial the FDA will accept without including a placebo study arm.

Placebo response rates in clinical trials for depression have been increasing, but variable, over time. The paper cited by Potter (Walsh et al., 2002) reports that the response to placebo across trials varied significantly—from approximately 10 percent to more than 50 percent—and was frequently substantial: in approximately half of the studies, 30 percent or more of patients assigned to placebo exhibited a clinically significant improvement. In addition, over the course of 2 decades (1980–2000), the proportion of patients responding to placebo increased at the rate of approximately 7 percent per decade (Walsh et al., 2002). The proportion of patients responding to active medication over this time period showed a similar increase. It should be noted that placebo response is a significant issue in the design of trials in many different disease areas.

According to Potter, the variability in placebo response rates over time has made it difficult to plan large clinical trials in depression. In designing clinical trials, it is customary for researchers to look to prior experiences with similar trials in the medical literature to determine how to power the study statistically and develop a target sample size. When there is such dramatic variability in placebo response rates across studies, researchers are left with little information with which to construct an informative, adequately powered study.


A number of workshop participants noted that the current state of research in depression is marked by an inability to effectively distinguish one antidepressant treatment from another and identify the patient populations best served by a particular drug. In discussing how best to advance clinical research in depression, presenters and audience members raised a number of issues and possibilities.

Combination Therapy: Antidepressants and Psychotherapy

Deborah Zarin, Director of, National Library of Medicine, highlighted nonpharmacologic interventions for depression (e.g., psychotherapy) and studies that have shown such interventions to be at least as effective or sometimes more so than pharmacologic treatments. She suggested that future depression research further explore such interventions or combinations either alone or in combination with pharmacologic interventions to develop the best treatments. Potter questioned whether it would be better to focus new research efforts on identifying a meaningful distinction between two antidepressants before trying to develop the optimal combination of psychotherapy and antidepressant. Kalali referred to Quintiles’ efforts in conducting two of the largest trials testing the combination of psychotherapy and medication but cited the limited availability of psychotherapy in the United States as a major barrier to its widespread use. In addition, he pointed out that psychotherapy has been shown to be successful in treating mild to moderate depression, but that many individuals with more severe forms of depression would not be candidates for psychotherapy.

Trivedi suggested that depression-focused psychotherapy should be included in research efforts to understand treatment effectiveness. However, many of the same issues affecting medication research also plague psychotherapy research. In response, Zarin suggested that studies of antidepressants could be improved by successfully characterizing the intensity of the patient visits that occur in a clinical trial; that is, there is an impact from the half-hour visits that take place during a clinical trial that goes beyond the effects of the drug being studied. This type of clinical management should be characterized further and could explain some of the variability in clinical trial outcomes for antidepressants, according to Zarin. Potter agreed that more sophisticated tools for measuring what happens in patient visits are necessary. He also mentioned Eli Lilly’s effort to design tools for measuring the impact of clinical trial visits on depression treatments. In the end, the measurement tools varied significantly based on individual trial site characteristics and were determined to be too imperfect for practical use.

Accelerating Depression Research

Potter suggested that, despite the importance of creating new therapies for depression, investment in this area is no longer a top priority for some in the pharmaceutical industry because no path forward exists for obtaining clear, interpretable answers to essential research questions. The challenges facing depression research go beyond simply improving the efficiency of clinical trials and include gaining a deeper understanding of what constitutes useful clinical trial information, as well as how signal detection can be improved. Potter explained that moving beyond a single rating instrument (HAM-D) for depression will probably be necessary.

Also reflecting on current challenges, Trivedi suggested that the field of neuroscience has made a number of breakthroughs in recent years, but the area of depression research has yet to combine these breakthroughs in a meaningful way. Thus large investments will be required to combine clinical moderators with genetic, serological, and other biomarkers to map depression and develop more comprehensive markers for disease severity. Further research into various markers for depression could help distinguish which treatments work for which patients. In addition, Trivedi explained that the vast majority of clinical trials in depression are short term and focus on the first 8 to 10 weeks of treatment. Studying the long-term effects of treating depression, a chronic disease, could help accelerate the development of new therapies.

Trivedi also mentioned the importance of developing new animal models for studying depression subtypes. For instance, no animal model for treatment-resistant depression exists. Thus, if a new therapy for treatment-resistant depression were developed today, it would be studied with the same animal models used for other conditions.

In addition to developing new animal models to advance drug development in psychiatry, Kalali noted that it is important to increase collaboration among industry, academia, and government to advance the interests of patients and move the field of psychiatry forward at a time when the high rate of failure in drug development is driving investment away from psychiatry and toward easier targets and diseases. Kalali suggested that more pharmaceutical companies should share their data regarding the rate at which the placebo response diverges from the response to active medication (i.e., placebo separation data) in clinical trials. These data could improve overall understanding of placebo response in depression trials and help in developing new, more effective trial methodologies. As an example of a large collaborative effort, Kalali highlighted his work as Chair of the Evidence Based Methodology Initiative (EMI) of the International Society for CNS (central nervous system) Drug Development. The EMI is currently conducting a Cochrane-like review of the literature (published and unpublished) in clinical trial methodologies in CNS and assigning a level of evidence to each example.1 The goal is to create a better understanding of the gaps that exist in current methodologies and the areas that require more evidence. The EMI hopes to create a pathway for improving CNS clinical trial methodologies.

To accelerate the development of new drugs to better serve patients, McNulty believes there should be greater integration of clinical and basic science. For example, if neurologists discover interesting differences in the brain potentially linked to placebo response rates, they should have some way to work with a basic scientist to research the meaning of these differences and explore them for the benefit of patients. Drawing on a wide range of scientific expertise could be useful in accelerating drug development for depression. Even in the promising area of genomic research, answers have been limited, according to McNulty. For instance, the gene for Huntington’s disease has been known for a number of years, yet no new therapies to treat or cure the disease have been developed. Huntington’s is a single-gene disorder, whereas depression is likely a polygenic disease in which environmental factors and genes interact in a way that is even more complex than in Huntington’s. Thus, significant challenges exist in the application of advances in genomics to depression research.

The appropriate study size for depression research was debated during the workshop discussions. Potter mentioned that placebo-controlled inpatient trials of early antidepressant medications were small studies—approximately 50 patients per study arm. According to Potter, these small trials were predictive of the benefit many patients receive from the medications. Potter suggested that large trials are not necessary for signal detection as long as the trial includes the right patient population. Paul Hébert expressed surprise that sample sizes for depression trials are so small, considering depression is a disease affecting a large portion of society. He suggested that depression researchers embrace the idea of larger trials to distinguish among different treatment effects. Califf also expressed surprise at the extent to which the field focuses on designing smaller, more precise trials. He suggested that if the same data were examined in his field of research, the conclusion would be to conduct trials 10–20 times larger than they are today.

In addition to study size, workshop participants considered the appropriate length of a clinical trial in depression. Trivedi noted that in a chronic, long-term illness such as depression or diabetes, clinical trials that follow patients for a significant length of time are important. Long-term studies for chronic diseases require successful patient retention strategies to be effective since significant dropout rates over time can jeopardize the validity of trial results. To illustrate the power of large, multicenter, long-term clinical trials for evaluating depression therapies, Trivedi described the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial (see Box 5-1).

Box Icon

BOX 5-1

Case Study: STAR*D. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study was a large, federally funded clinical trial that tested the effectiveness of antidepressants in a population diagnosed with major depressive disorder. The trial (more...)

Finally, Potter explained that intention-to-treat analysis is useful and can be especially effective in depression research.2 According to Potter, this methodology allows outcome measurements to be continued at the end of the study even if the intervention does not continue.

Cochrane reviews explore the evidence for and against the effectiveness and appropriateness of treatments in specific circumstances to facilitate the choices of doctors, patients, policy makers, and others in health care (http://www​​/reviews/clibintro.htm).

In an intention-to-treat analysis, subjects who started in the trial are included in the final analysis regardless of whether they finished the treatment (i.e., dropouts are treated as if they finished the trial). The approach seeks to evaluate a treatment as it would be administered in the real world.



Cochrane reviews explore the evidence for and against the effectiveness and appropriateness of treatments in specific circumstances to facilitate the choices of doctors, patients, policy makers, and others in health care (http://www​​/reviews/clibintro.htm).


In an intention-to-treat analysis, subjects who started in the trial are included in the final analysis regardless of whether they finished the treatment (i.e., dropouts are treated as if they finished the trial). The approach seeks to evaluate a treatment as it would be administered in the real world.

Copyright © 2010, National Academy of Sciences.
Bookshelf ID: NBK50890
PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.3M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...