Commentary: An exemplar of progress in understanding complex disorders - reflections on what we have learned about eating disorders (⁾

Abstract

A number of recent advances in eating disorders research have helped clarify the nature of risk for the development of such disorders. Culbert, Racine, and Klump (2015) provide an empirical and thoughtful review of these recent advances. The authors identified empirically established risk factors in each of several categories of risk for eating disorders: genetic influences, neurotransmitter activity, hormones, personality, and sociocultural influences. We highlight three implications of their review. First, the review can serve as an important asset to eating disorder researchers, both substantively, by providing a comprehensive account of empirically supported risk processes; and methodologically, by highlighting good standards of evidence for acceptance of a candidate risk factor. Second, eating disorder risk is increased by both transdiagnostic and eating disorder-specific factors; there is a need to understand how these types of factors transact with each other. Third and most important, we highlight the importance of Culbert et al.'s advocacy for the development of theoretical models, and empirical tests of those models that specify transactions among different types of risk factors, such as those based on genetic, neurobiological, personality, and social processes.

We are delighted to be able to comment on the accompanying contribution by Culbert, Racine, and Klump (2015) to this journal. The authors have provided a valuable service to eating disorder researchers by conducting an extensive, empirically driven integrative review of risk factors for eating disorders. As we describe below, we believe this review article makes a number of important contributions to the fields of child mental health, eating disorders, and, more broadly, to efforts to develop truly integrative models of risk for psychopathology.

Importance of disordered eating in child mental health

There has been a growing recognition of the importance of eating disorder behaviors as threats to child mental health and to the public health more broadly. This journal has played an important role in this development, by publishing theoretical accounts of eating disorder risk (Pearson, Riley, Davis, & Smith, 2014) as well as this review, which identified risk factors that met clearly defined standards of empirical evidence. Other reviews (Bakalar, Shank, Vannucci, Radin, & Tanofsky-Kraff, 2015) and theoretical accounts (Pearson, Wonderlich, & Smith, 2015) have also appeared recently in other outlets.

One reason for this increased appreciation for the problems associated with disordered eating has been the recognition that eating disorder behaviors exist on a continuum. Elevations on dimensions of symptoms predict the development of diagnosable disorders (Culbert et al., 2015) and are associated with significant dysfunction, even in the absence of a diagnosable disorder (Bakalar et al., 2015). Researchers and clinicians have come to appreciate that the harms associated with disordered eating are far more extensive than may be implied by the low base rates of diagnosable anorexia nervosa, bulimia nervosa, binge eating disorder, and other, residual diagnostic categories. It was thus a valuable contribution that Culbert et al. emphasized eating disorder symptom dimensions in their review.

Striking advances in understanding the etiology of disordered eating

Culbert et al. made a valuable contribution to eating disorder research by reviewing and evaluating putative risk factors on the basis of the degree to which they meet recognized standards of evidence for classification as a risk factor (Kraemer, Kazdin, Offord, Kessler, Jensen, & Kupfer, 1997). A clear conclusion from their paper is that eating disorder research has made remarkable progress over the last 15 years in understanding the risk process; several factors met the empirical standards they used for recognition as established risk factors. As we discuss further below, an important aspect of their work is the identification of risk factors spanning genetic influences, neurotransmitter activity, hormones, personality, and sociocultural influences. We emphasize three implications of their review.

Valuable direction for eating disorders research

The first implication is the valuable direction for eating disorder research provided by their review. Researchers can and should take advantage of this review by developing models that integrate their own research with the established risk factors identified by Culbert et al. as well as with well-validated theoretical integrations of such risk factors (Pearson et al., 2014, 2015). In addition, the use of established standards of evidence for identification of risk factors by Culbert and colleagues provides something of a template for investigators who are investigating promising, but not yet fully validated, candidate risk factors.

Transdiagnostic and eating disorder-specific risk processes

A second implication of the Culbert et al. review is that risk for eating disorders is increased by two broad categories of factors. The first is transdiagnostic factors, i.e. factors that increase risk for multiple forms of dysfunction. Examples of identified transdiagnostic risk factors include variations in the neurotransmitters serotonin and dopamine, and personality risk factors such as negative urgency. Processes involving these neurotransmitters and negative urgency along with other risk factors the authors identified clearly influence a wide range of dysfunction, including classic externalizing disorders, such as substance use and aggression and classic internalizing disorders, such as depression (Carver, Joorman, & Johnson, 2008; Cyders & Smith, 2008; Smith, Guller, & Zapolski, 2013).

It is valuable for eating disorder researchers to recognize that a number of eating disorder risk factors also operate in relation to other disorders, so they can take full advantage of advances made in other fields in understanding shared risk processes. As much progress as has been made in recent years in understanding risk for eating disorders, it is clear that future progress will occur far more rapidly if eating disorder researchers make a priority of becoming aware of, and learning from, the successes and failures of research on risk factors that are shared with other disorders. To investigate the process by which eating disorders develop is likely to involve investigation of mechanisms by which shared risk factors influence the emergence of other disorders.

The second category of risk processes are those that are likely specific to disordered eating, such as learning about and internalizing social norms that emphasize the value of thin body types. An important challenge for eating disorder researchers implied by this review is to develop models that integrate broad, transdiagnostic risk processes with those specific to disordered eating. Of all the forms of dysfunction that become more likely with, for example, elevations in traits such as perfectionism, negative affect, or negative urgency, what factors result in the dysfunction becoming expressed as disordered eating?

Culbert et al. mention models that have begun to address this problem; there is a clear need for a greater focus on this issue. Here, too, awareness of other fields and collaboration with researchers in those fields will prove helpful. For example, it is likely that a team of eating disorders and substance abuse researchers would make rapid advances in developing models that clarify: (a) which risk processes are shared between the two problems; (b) which are unique to one of the problems; (c) the nature of the transactional process between shared and disorder-specific risk factors; and (d) what influences the process by which a shared risk factor comes to be associated with a disorder-specific risk process and just one of the two disorders. There are, of course, many possible outcomes that could follow from collaborations of this kind.

Integration across genetic, biological, personality, and sociocultural risk processes

The third, and perhaps the most important, implication of Culbert et al. is the need to develop models and conduct empirical research that leads to a true integration of genetic, hormonal, neurotransmitter, personality, and sociocultural risk processes. At this point in the development of our field, it seems clear that risk is influenced by each of these dimensions. What is missing is the development and evaluation of models that explain risk by integrating across them. We very much appreciated the thoughtful way in which Culbert et al. highlighted the likely transactional processes across risk processes. An important theme of their review was the need to think about risk in an integrative way and conduct empirical tests of transactions among risk factors.

As the authors note, it is rare to find either risk models or empirical studies that offer integration across, for example, genes, hormones, personality, and psychosocial learning. The need to conduct truly integrative research is a shared one within the eating disorders field. Genetics researchers risk conducting very large genome-wide association studies without adequately or accurately representing the many factors that transact with genetic effects. Psychology researchers risk studying personality, or applying laboratory manipulations to study putative risk factors, without investigating the relationship between psychosocial risk and heritable tendencies, neurotransmitter activity, or hormone levels. Social researchers risk studying sociocultural and societal factors, such as the idealization of the thin body type, without adequate consideration of factors that influence responsivity to the environment or moderate such influence.

It is of course unrealistic to expect all empirical studies to test risk processes integrated across multiple levels of explanation. Instead, we argue, and we believe Culbert et al. argue as well, that the field will benefit from application of three principles. First, rapid advances in our field are most likely to come from the development of models that explicitly recognize the many factors that influence risk and offer theoretical models of how those factors transact. Second, there is a clear need for more empirical work designed to test transactional models. Third, although most studies will not, realistically, integrate across more than two or three risk factors, it is nonetheless important that the theory driving such studies be integrative in nature. Such theories are likely to prove most useful and generative of subsequent important advances.

It is of course the case that the need for integration across multiple levels of explanation exists in many fields within psychopathology research. We believe that eating disorders researchers have an opportunity to provide a model for successful integration, because our field is characterized by established risk factors across so many such levels. Indeed, the nature of eating pathology means that any comprehensive risk account will necessarily integrate biological and medical considerations with psychological processes and social influences. In a very real sense, the eating disorders field is well-positioned to lead the way in the development of truly integrative accounts of dysfunction. Thus, Culbert et al.'s emphasis on transactions among risk factors and other recent theoretical work (e.g., Pearson et al., 2014) also emphasizing such transactions perhaps anticipate the next generation of advances in understanding eating disorders. The many advances reviewed by Culbert et al. have set the stage for the development of fully integrative theories of risk. We anticipate that a similar literature review 15 years from now will involve evaluation of multiple, competing integrative theories of risk. Perhaps the advances we anticipate will prove useful for researchers developing similar models for other forms of dysfunction.