Challenges Associated with Controlled Psychopharmacological Research Trials in Adolescents with Eating Disorders

Abstract

Introduction

Eating disordered populations present many unique challenges both to clinicians and researchers. Adolescents with eating disorders can be difficult to treat, and the challenges associated with research in this area can be significant.

Objectives

This paper was written with three main objectives in mind: to comment on some of the barriers impeding mental health research in general, to highlight challenges faced in the design and implementation of ED-specific studies, and to integrate personal insight into some of the many challenges that we have encountered during our experience with a randomized clinical trial examining the efficacy of olanzapine for the adjunctive treatment of youth with Anorexia Nervosa.

Discussion

It is hoped that providing information in this context will allow researchers greater insight into some of the many challenges that accompany study of this cohort.

Introduction

Assessment and treatment of adolescents with anorexia nervosa (AN) typically involves a multidisciplinary and multimodal approach. Controlled weight restoration in combination with psychoeducational and psychotherapeutic treatment provides patients with an optimized chance of recovery, although long term success and cure rates are difficult to predict, owing to insufficient long-term data (Berkman, Lohr, & Bulik, 2007). Medications such as anti-depressants and conventional neuroleptics have been previously studied in an effort to determine whether they affect weight restoration, reduce symptoms or improve recovery rates in adolescents with anorexia nervosa (AN) (Bulik et al, 2007). The outcomes in each of these respective trials have been discouraging. The emergence and use of “atypical antipsychotics” such as olanzapine over the last decade for the treatment of severe AN has provided a renewed sense of optimism in terms of pharmacological options for treatment of severe AN in adolescents (Malina et al, 2003; Mehler et al, 2001; Boachie, Goldfield & Spettigue, 2003).

In 2003, a working group convened within our center with a goal of designing and implementing a randomized trial evaluating the efficacy and tolerability of olanzapine in adolescent patients with AN. Over the last four years, our research team has encountered many obstacles. It is hoped that by describing some of these challenges, we will provide other researchers with insight into some of the complicated questions that need to be addressed when designing future studies.

The Challenges of Mental Health Research

Those of us who engage in Mental Health research in a clinical setting are only too familiar with its many challenges, not the least of which is the vulnerability of research participants with mental health illnesses. Obstacles to project design and patient recruitment are not new concepts in the realm of research. Both the National Institute of Mental Health and the American Academy of Child and Adolescent Psychiatry in the U.S. have addressed some of these challenges in recent forums (Agras et al, 2004; Hinshaw et al, 2004). However, neither of these documents is inclusive of the many obstacles that face the researcher in the field of Mental Health and to our knowledge, no specific Canadian document has been produced so far. In this section, we have attempted to summate some of the challenges we have encountered by addressing issues such as recruitment, consent and confidentiality, the challenges of studying psychoactive medications including the difficulties of identifying and measuring adverse effects, the role of research ethics boards and other regulatory bodies, ethical issues, measuring outcomes, as well as the role of potential side effects.

Recruitment

As noted above, one of the major obstacles to research in the field of mental health is recruitment. Large patient samples are a fundamental requirement of studies that set out to test important and clinically driven hypotheses. Hence, the ability of researchers to attract, recruit and retain patients into studies that require sufficient statistical power to confirm or negate a hypothesis is crucial. The AACAP Research Forum (Hinshaw et al, 2004) notes, “poor recruitment is the key limiting factor regarding the successful completion of many trials and surveys.” Factors affecting recruitment include low prevalence rates for certain mental illnesses (such as adolescent anorexia nervosa), and stigma, which can interfere with patients’ willingness to seek treatment. Patients often express fear at the thought of partaking in research trials, and the clinician-patient relationship can experience significant strain as youth may struggle with thoughts of whether researchers will be working in their best interests, or the best interests of the study. In our study, two of the most notable factors impacting our rate of recruitment to date have been related to the potential for side effects (see below) and the study design (randomized placebo-controlled trial): three potential candidates (and parents) have not been willing to accept the risk of receiving placebo, while others have refused because they cannot accept the risk of receiving a medication that may contribute to weight gain. This has resulted in a total enrollment of 6 patients (19 patients have met inclusion criteria) since active recruiting began over 18 months ago. Of these six patients, one dropped out during the second week of study due to medication intolerance. Of the remaining refusals, 9 have come directly from patients, and 4 from parents.

Consent and Confidentiality

Consent forms can be long and daunting, especially consent for trials involving the use of psychoactive medication in which every possible side effect is listed. These forms are often criticized by parents as being obscure, lengthy, and frightening (Benedetto & Jensen, 1997). Patients and their families often express fear that they are being used as “guinea pigs” when a new medication, or a medication for a new indication, is being tested. For example, in our placebo-controlled trial, as part of their informed consent, patients and parents must be informed that there is little current proof of the medication’s effectiveness in anorexia nervosa owing to the current lack of established evidence. They are then presented with a consent form with a long list of every potential side effect included in study involvement, ranging from local bruises attributable to blood-draws, through new-onset diabetes, to potentially fatal neuroleptic malignant syndrome. Needless to say, these issues, along with stressful and potentially dysfunctional intra-familial relationships (including splits between patient and parents, or parents themselves) have contributed significantly to our enrollment challenges.

The Role of Funding Agencies, Research Ethics Boards and other Regulatory Bodies

The challenges of identifying and securing funds for research purposes are significant. Indeed, for clinician researchers, the amount of time required to write grant applications in hopes of securing funding while at the same time attempting to balance a full clinical workload can be overwhelming. Without such funds however, large-scale studies typically become dependent upon public-private partnerships. Placebo-controlled trials that involve psychoactive medications under patent usually require participation from pharmaceutical companies in order to secure placebo medication, thereby creating potential challenges depending on what the outlined study objectives are (i.e. pharmaceutical companies may not wish to participate).

Each stage of research involving controlled pharmacological trials requires an abundance of planning, paperwork, communication, and most importantly time. There is little doubt that the role of ethical and regulatory bodies are essential in order to ensure the promotion and completion of safe and ethical research. Keeping this in mind, any changes regarding planning, design or implementation that are encountered need to be addressed and cleared by every board or body overseeing the safety of the project. Such holdups, albeit necessary, can significantly delay a study start date. In order to effectively address potential ethical or design issues, our research team invited outside clinicians and researchers to participate in a large interdisciplinary steering committee for the project.

Ethical Issues

Ethical problems often arise in Mental Health research, owing at least in part to the nature of the often stigmatized, debilitating, chronic course of psychiatric illness (Roberts, 2002). Children and adolescents, along with mental health patients have historically been considered a “vulnerable population,” and are commonly omitted from initial inclusion in trials studying new medications; controlled studies in these populations often lag significantly behind (Rudorfer, 1993; Roberts, 2002).

Researchers often struggle with the ethics of comparing a treatment group to a placebo or “no treatment” group. The ethics of placebo-controlled research in Canada has come under closer scrutiny in recent years (Sibbald, 2002). The National Placebo Intitative, (NPI) (a collaborative effort between the Canadian Institiute of Health Research (CIHR) and Health Canada published in 2004) states that a placebo comparator is acceptable in six different situations; two of which include the absence of established effective therapies for the population, and lack of existing evidence to support the net therapeutic benefit of available therapies. The value of placebo control in mental health studies can be significant, depending on the population under study. For example, early open label studies investigating the efficacy of tri-cyclic anti-depressants in adolescents showed promise, but overall outcomes were no different in placebo-controlled randomized trials (Rudorfer, 1993). How then does one set out to effectively study atypical anti-psychotic medication for eating disorder patients? Certainly one would think that the placebo-controlled trial would provide an optimal starting point. As noted by Strober (2005) however, it is no longer uncommon for research patients to refuse treatment at the point of randomization, a difficulty we have encountered on multiple occasions.

The question of prescribing practices for atypical antipsychotics within our program has also been discussed at various meetings. As noted above, the evidence for atypical anti-psychotic’s efficacy in adolescent populations with anorexia nervosa is lacking, although case reports and small open label trials have suggested the possibility of benefit (Malina et al, 2003; Mehler et al, 2001; Boachie, Goldfield & Spettigue, 2003).

There have been no published reports of Canadian prescribing practices for the treatment of anorexia nervosa, and in our communication with other centers, such practices vary significantly. Clearly, by implementing our RCT, we are interested in ascertaining whether olanzapine has any effect on the treatment of youth with AN. Ethically however, we have debated if and how our prescribing patterns should apply to patients that have either refused enrollment or are not eligible, given the current lack of well-designed high-powered studies in the area. We have used our steering committee to help discuss and debate some of these issues and have as well asked for consultation outside of the committee for especially challenging questions. Roberts (2002) does an excellent job of highlighting some of the many challenges associated with role conflict, including the need to recognize the dichotomous role of the clinician investigator in charge of looking after the best interests of his/her patient but also overseeing a clinical trial.

Side Effect Potential

Psychoactive medications have the potential to cause a number of unique side effects; each of these can cause significant challenges in a research domain. In the context of our study, it is known that olanzapine can be associated with a number of potential serious metabolic side effects including risk of sedation, weight gain, new-onset diabetes mellitus, and dyslipidaemia (Newcomer, 2005). One must of course consider that the majority of these studies have been completed in psychotic patients and not eating disordered populations. As well, many of the earlier studies did not control for nutrition, which would not be the case in any eating disorder research design. Thus, the question of how transferable the absolute risk is in a low-weight adolescent cohort becomes difficult to answer. One of the biggest recruitment challenges we have faced outside of medication administration questions has been associated with the potential for side effects, most notably weight gain and diabetes mellitus. In an effort to help quantify our experience to date, we have since undertaken a retrospective review of our experience with olanzapine in hopes of adding to the minimal amount of information presently available.

Despite the abundance of challenges already mentioned, there are others that relate to the measurement of extrapyramidal symptoms (EPS) in patients on antipsychotic medications. While more recent “atypical’ or second-generation antipsychotic medications are less likely to be associated with these disturbing symptoms, the risk is still present, and may be increased in young patients. These side effects can be ‘broken down’ into four different categories: acute dystonia, akathisia, Parkinsonism, and dyskinesia. We were unable to identify any one scale that measured each of these areas effectively; instead, we found only scales that individually monitor each of the four major categories. Pursuing this option would have added a significant amount of time to each of the required clinic visits; clinics that are already overbooked and running significantly overtime. For the purpose of our study, we chose the “Abnormal Involuntary Movement Scale,” or AIMS scale, and hope to capture any outstanding symptoms in our over-inclusive symptoms checklist.

Measuring Outcomes

A further obstacle to mental health research is the difficulty of defining what is being measured, and figuring out how to measure it. With regards to our trial, our impression (based upon anecdotal experience) has been that olanzapine decreases the agitation associated with eating and weight gain, and decreases the degree of rumination and resistance to treatment. It is important to remember that most scales are designed to measure thoughts and behaviours. Often, many of the challenges experienced in ED research lies in the ego-syntonic nature of the disorder and the fact that denial is often a core element to the presentation. Self-report measures are thus limited in their sensitivity and validity. There are well established and reliable measures of eating disorder thoughts and symptoms, however most measures have been designed for adults and then revalidated in youth. In our study, we have attempted to resolve these challenges by designing a clinician rating scale that can be used in conjunction with established self-report measures, thus overcoming the problems inherent in the denial youth are experiencing. It is also designed to measure the agitation and obsessional thoughts associated with eating disorders, as these are not well measured on other scales. This solution has of course required the development of a separate research project with time and financial commitments. Understandably, this venture compounded the delay in the implementation of our trial, and as well, has resulted in a number of other administrative and functional challenges.

ED-Specific Research Challenges

Much of the discussion above, as noted by our personal experiences, alludes to specific challenges in working with adolescent ED populations. In an attempt to quantify the reported experiences of others, the authors have set out to conduct a systematic review of the topic based upon current available literature. Although this work is not yet fully complete, it is clear in our initial survey of the topic that the methodological challenges present in the field of ED research are enormous.

Rather than attempt to inadequately summate all of the potential challenges in this section, we have instead decided to concentrate on one area in particular: the role of potential confounders in ED research. Examples of potential confounders include the role of malnutrition on treatment outcomes, the role of cultural diversity on classification, enrollment, and treatment outcomes, as well as potential similarities or differences in etiology, pathophysiology, and susceptibility to treatment between male and female patients (appreciating the incidence of eating disorders in males is even less) and patients with varying diagnostic classifications (i.e. patients with anorexia nervosa binge-purge subtype vs. patients with bulimia nervosa).

Another big challenge that researchers face in the area of ED research involves the establishment of inclusion and exclusion criteria. We cannot control which patients accept or decline participation, but it becomes important to maximize the ability to recruit and study diverse samples in an effort to produce outcomes that may be efficacious across heterogeneous populations. As a result, whenever clinically relevant outcomes are reported, the question of how representative the population is to “average ED populations” deserves attention. Clearly, depending upon the design itself, selection bias can be introduced before a trial even gets under way. In our study, for example, we have had 68% of eligible patients already refuse enrollment for a variety of different reasons. Is it possible that our enrolled patients have been “closer to accepting” treatment, or had less absolute fear of weight gain, and were therefore able to accept the potential benefit of drug vs. risk the potential of extra weight gain? If so, how representative is this population and how transferable would the potential results be? These issues deserve deeper exploration than can be afforded in this context.

Studying the efficacy of non-pharmacological treatment introduces a host of further obstacles, most significant of which is the challenge of standardizing treatment or therapy. How does one measure or account for differences in staffing experience, variability, or competence? And what of dimensional variables such as personality; surely these heavily influence the bond and therapeutic alliance a patient will form. How do individual practices at multi-center sites influence large-scale results? Other researchers have sought to examine these important questions and similar issues (Strober, 2005; Agras et al, 2004).

Summary

There is little doubt that research in any venue is extremely challenging, laborious, time-consuming, expensive, and potentially frustrating. Having the opportunity however to advance efficacious treatment options through the completion of sound research can be immensely gratifying. This article has outlined a number of challenges and potential considerations that should be accounted for before any large-scale ED pharmacological study is undertaken. Taking all of the above points into consideration, the most critical element of ED research lies in the initial planning stages. It will be important for Canadian researchers to collaborate in future venues, if we are ever to find answers to the elusive questions surrounding optimal treatment strategies for patients with eating disorders.