Normal and Abnormal Anxiety in the Age of DSM-5 and ICD-11

Introduction

Several developments have left psychiatry in an ongoing state of flux (Brüne et al., 2012). First, basic theory and methodology have undergone significant shifts in recent decades. Psychoanalytic concepts and practices that dominated in the mid-20th century were displaced by cognitive-behavioral ideas and trials in the 1980s, while clinical neuroscience is at the forefront of the 21st century psychiatric research agenda. Second, key controversies persist on fundamental issues, including the nature of mental disorder, defining what is and what is not a mental disorder, the optimal approach to psychiatric diagnosis, the causes of psychiatric disorder, and the best approach to optimizing treatments.

In this article, we consider the challenge of differentiating normal from abnormal emotions, particularly normal anxiety from its disorders, in an effort to shed light on these conceptual controversies. We assess how well the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the 11th edition of the International Classification of Disease (ICD-11) meet this challenge in the light of modern emotions theory. We suggest that one core problem is expecting too much from psychiatric diagnoses; most are not diseases with specific etiologies, they are clinical descriptions of observed syndromes intended to guide treatment choices (Nesse & Stein, 2012; Sartorius, 2015).

A related fundamental problem is trying to define abnormal emotions based only on considerations of severity and duration. The rest of medicine often examines the context to decide if responses like cough and pain are normal or abnormal responses. Such protective responses are aroused by abnormalities, but doctors may recognize that they are not themselves abnormal unless they are manifest in the absence of an appropriate arousal cue. In psychiatry, however, there is the risk that anxiety may at times be classified as pathological without regard for the situation.

DSM-5 and ICD-11

DSM-5 initially aimed at a thoroughgoing revision of DSM-IV, with goals of grounding diagnoses in neuroscience, and moving from a categorical to a dimensional system (Regier, Narrow, Kuhl, & Kupfer, 2009). However, those goals were not reached; no sensitive and specific biomarkers for mental disorders have yet been found, and clinicians continue to prefer categorical diagnostic constructs. Nevertheless, some changes in DSM-5 do reflect these original aspirations; for example, obsessive-compulsive disorder (OCD) and other anxiety disorders are now classified separately because they show substantive biological differences (Stein, Fineberg, et al., 2010), and because separating them may encourage better diagnosis and treatment of a spectrum of conditions that is often underdiagnosed or inappropriately treated (Phillips et al., 2010).

ICD-11 is intended to maximize clinical utility (Reed & Ayuso-Mateos, 2011). It recognizes that patients with mental disorders frequently present in primary care, where they are often evaluated by nonspecialist clinicians. Indeed, compared to the ICD-11, some of the operational criteria of the DSM-5 are pseudoprecise; for example, while the exact cut-point of 6 months duration for symptoms of generalized anxiety disorder (GAD) may contribute to diagnostic reliability, it is not surprising that GAD patients with slightly shorter symptom duration do not markedly differ from those with slightly longer duration (Kessler et al., 2005). At the same time, ICD-11 has accepted many of the tenets of DSM-5; for example, like DSM-5, ICD-11 will have separate sections for anxiety disorders, obsessive-compulsive and related disorders, and trauma- and stressor-related disorders.

Consider a patient who suffers from obsessions about physical and moral uncleanliness, and repetitive compulsions to pray. Such an individual could have met diagnostic criteria for OCD in earlier editions of the DSM (i.e., DSM-III and DSM-IV), as well as in early editions of ICD (i.e., ICD-9 and ICD-10). Subtle differences in the newer editions arguably reflect research advances; for example, from DSM-IV onwards, the diagnostic criteria for OCD emphasize that compulsions can be purely mental. In addition, as noted before, both DSM-5 and ICD-11 will classify OCD in a new section on obsessive-compulsive and related disorders (OCDRs). However, these are relatively small changes. Psychiatric diagnosis is still based largely on clinical observations, rather than on laboratory testing, with fuzzy thresholds between normality and abnormality (Stein, Phillips, et al., 2010).

Critiques of Psychiatric Nosology

Given the enormous effort made to update DSM-5 and ICD-11 in the light of relevant literature, it is notable that few appear satisfied with the revised versions. The history is extraordinary. The DSM-III was created to make diagnosis objective at a time when psychiatry’s status as a medical specialty was in jeopardy. It was at the time viewed by some as a temporary system that would soon be replaced by diagnoses based on biomarkers and the specific brain abnormalities that were assumed to cause most disorders. The failure of three decades of intensive research to find reliable biomarkers for any major psychiatric disorder has been perceived by many as problematic. However, instead of considering the possibility that the assumption was incorrect, many have blamed the DSM, positing that some other new categories will map to specific neurophysiological abnormalities.

Indeed, research leaders at National Institute of Mental Health (NIMH) have developed the Research Domain Criteria (RDoC), as an alternative approach to classification. The RDoC describes phenomena on different dimensions, and seeks to understand them in terms of underlying neurocircuitry and neurotransmitters (Cuthbert & Insel, 2013). In this view, a translational approach is key; behavioral dimensions and their biological mechanisms can be characterized in animal models, and applied to human emotions and emotional disorders. Notably, the RDoC framework also emphasizes how genes and environment interact to account for animal and human phenotypes. Its emphasis on psychobiology may advance research, however, it is not a substitute for a diagnostic system.

While framed as a radical new approach, the RDoC runs the risk of doubling down on the bet that we can find specific brain causes for most mental disorders. Mental phenomena obviously arise from brains, so it can seem that there is no alternative possibility. This is, however, very different from the rest of medicine, where defenses such as fever and cough are readily distinguished from diseases that cause them, such as pneumonia. A more genuinely medical approach would use this principle to recognize the special status of emotional disorders as compared to other psychiatric disorders. The rest of medicine also recognizes syndromes such as congestive heart failure that can have many interacting causes. A more genuinely medical approach would accept the reality of clinical syndromes and look for causes without expecting to find specific causes for each disorder.

The RDoC framework remains at an early stage of development, and it is therefore difficult to evaluate its eventual impact on the field. Furthermore, the RDoC is not intended as a substitute for the DSM and ICD. At the same time, caution about this approach seems warranted (Stein, 2014). First, a clear goal of medical and psychiatric classification is clinical utility; this is only partly related to underlying pathophysiology. In psychiatry, as in the rest of medicine, many entities are syndromes that have clinical utility despite having multiple causes, blurry boundaries, and absent biomarkers. Second, given that multiple mechanisms play a role in producing psychiatric signs and symptoms, privileging one particular diagnostic validator, such as “circuit-based behavioral dimensions” may ignore much relevant information. Science has progressed from Hippocrates’s account of the “humors,” to neurocircuitry-based theories, but a century from now current circuitry concepts may be considered rudimentary and based on naïve reductionist notions of causality. Third, given the multiple mechanisms underlying psychiatric complaints, and the many considerations relevant to treatment decisions, we should be cautious in our expectation that diagnostic criteria or thresholds will ultimately be based on behavioral dimensions or biological markers. In medicine and psychiatry, deciding on whether and how to intervene necessarily requires a complex assessment of a range of factors, including understanding the function of symptoms, their social context, and treatment risks versus benefits (Kessler et al., 2003).

Another influential perspective in current debates comes from advocates of global approaches to mental health. Writings in this area have emphasized that most psychiatry research has been undertaken in high-income countries even though the vast bulk of those with mental disorders live in low- and middle-income regions where most people who suffer from mental disorders do not receive adequate treatment. It is important that classification approaches be useful in all settings. In this view, the key translational step is not from the bench to the bedside, but from the clinic to the community; evaluations and interventions need to be formulated in ways that they can feasibly and affordably be scaled up (Jacobs & Patel, 2014).

Consider, for example, a patient who has obsessions about cleanliness, washing compulsions, and motor tics. Research suggests that OCD patients with and without tics may have somewhat different clinical features, including different treatment responses (Leckman et al., 2010). Furthermore, genetic vulnerability to OCD is related to vulnerability to tics in some families. Still, the various perspectives (DSM-5, ICD-11, RDoC, global health) discussed here may be complementary rather than contradictory, when considering an individual patient. DSM-5 and ICD-11 provide a useful clinical approach, drawing a pragmatic boundary between normality and abnormality, with DSM-5 providing a tic specifier for OCD. From the perspective of the RDoC framework, future research on such individuals needs to address the neurobiology of the relevant domain (e.g., cognitive and motoric control), which cuts across this symptomatology. From the perspective of global mental health, the key issue is whether it is feasible and cost-effective to scale-up interventions such as exposure prevention and habit reversal across a range of OCDRs. This complementarity is arguably possible because of the somewhat different aims of each of these systems; DSM-5 and ICD-11 focus on providing a pragmatic guide to clinical decision-making, the RDoC framework intends to advance translational neuroscience and to develop new treatment targets, while global mental health emphasizes the importance of implementation science.

Emotions and Nosology

An approach that emphasizes the complementarity of different approaches to evaluation may, however, be a premature conceptual compromise. Can contemporary knowledge about emotions provide a more scientific basis for psychiatric nosology? Certainly, we now know a great deal about the psychobiology of emotions such as fear. Indeed, one domain defined by the RDoC framework is that of “negative valence,” which includes acute threat (or fear) and potential threat (or anxiety). The RDoC framework does not, however, specifically emphasize the utility of these negative emotions in certain situations (Nesse & Ellsworth, 2009), nor does it emphasize how natural selection shaped mechanisms that regulate expression of negative emotions according to the “smoke detector principle.” A signal detection analysis reveals why many “false alarms” are products of a normal brain; the costs of the negative emotion are small compared to the costs of not experiencing the emotion in a situation where it may offer huge benefits (Nesse, 2005). Social anxiety, for example, is an aversive emotion that can give selective advantages in certain social situations; it often seems excessive because its costs are immediately and personally experienced, while its benefits are delayed and indirect.

The nature of “normality” has long been debated in the philosophical and psychiatric literature. Importantly, an evolutionary perspective implies that many distressing and even disabling emotional experiences can arise from normal brains. It also explains why many emotions that are normal in their expression are nonetheless problematic or excessive in the individual instance, and thus appropriate targets of treatment, in the same way that normal pain is an appropriate target for treatment. Thus our intention is not to equate normality with statistical norms; for example, while tooth caries may be ubiquitous in the population, we still want to conceptualize it as a disease (Nesse, 2001). Similarly, our intention is not to conflate normality with adaptability; a very low threshold for “false alarms” may offer selective advantages under certain circumstances (e.g., a dangerous environment), yet be associated with suffering and impairment, and so constitute a disease (Stein, 2008).

A Medical Model of Nosology

Physicians in specialties other than psychiatry often distinguish between evidence of bodily malfunction and symptoms that are normal protective responses. Seizures, paralysis, and dyskinesias arise from abnormal bodily mechanisms, but cough, pain, and fever are normal protective responses shaped by natural selection in conjunction with regulation systems that express these responses in situations where their benefits are likely to exceed their costs (Nesse & Stein, 2012). In patients with fever, physicians avoid symptomatic treatment until they have understood more fully its underlying cause, which can then be directly targeted. There are some exceptions to this rule; when fever accompanies seizures in young children, symptomatic treatment is administered, and in some cases physicians do erroneously prescribe symptomatic treatments for fever and so “mask” key underlying processes with potentially negative consequences.

Capacities for protective responses, including negative emotions such as anxiety, disgust, and low mood, also exist because of their evolutionary advantages (Nesse, 1990; Nesse & Stein, 2012). Fear, for example, adjusts a range of aspects of physiology, cognition, behavior, and motivation in ways that increase ability to cope with dangerous situations and so offered our ancestors selective advantages. The survival value of emotions is confirmed by the existence of regulatory systems that closely govern them; such systems could evolve only if the responses were adaptive under particular conditions. It is also confirmed by the negative consequences that may ensue when normal defenses are blocked; clinicians must weigh the potential benefits of medications that suppress cough and anxiety, versus the possible harms of cough suppressants (when infected lung airways need to be cleared) and of anxiolytics (when real danger requires appropriately high anxiety).

Systems can fail, giving rise to responses that are abnormal in any circumstance, such as seizures or delusions. High prevalence rates for anxiety and mood disorders indicate that the regulation systems governing emotions are especially prone to failure. Further, like pain and fever, anxiety and depression are relatively nonspecific responses that can be aroused by a range of diverse circumstances, so comorbidity and heterogeneity are expected. Most such failures are only partial; the relevant responses are excessive or deficient for the particular situation. However, in characterizing responses as too soon, too strong, or too prolonged, clinicians should rely on the same kind of functional understanding of normal responses that guides decisions in other areas of medicine. For instance, cough clears foreign matter from respiratory airways, so clinicians search for possible causes; cough itself is characterized as abnormal only when no such cause is found. Analogously, in psychiatry, emotional states should be classified as disorders only if they are excessive for the situation; deciding what is excessive requires knowledge about what situations normally arouse the symptom, as well as an understanding of the current situation. Wakefield has proposed that disease is a “harmful dysfunction,” and noting the difficulty in establishing the presence of “dysfunction,” First and Wakefield (2013) have posited that clinical considerations such as severity and duration are “pathosuggestive.” We agree with this approach, which also emphasizes the value of appreciating the context of symptoms, and would emphasize that an understanding of what is “pathosuggestive” is based on a good knowledge of the evolutionary-based concept that adverse emotions can be adaptive.

Like other medical and psychiatric symptoms, emotions arise from interactions of persons with situations (Nesse & Ellsworth, 2009; Nesse & Stein, 2012). Clinical assessments often focus on trait differences among individuals; most anxious patients are concerned about their chronic symptoms, rather than about their anxiety response to a particular spider. However, relieving a patient’s negative emotions often requires close examination of the motivational structure of his or her life; this includes the person’s goals, strategies, opportunities, and obstacles in a range of different areas. The complexity of an individual’s situation and interactions makes it difficult to generalize about causes; the important factors may differ from person to person, and even from episode to episode. Deciding how to use information about context is a challenge, but we have previously suggested that a simple approach would be to code the causes for each emotional state as none–mild–moderate–severe on two axes, one indicating the trait vulnerability, the other the intensity of current situations likely to arouse it. Thus, for example, in the case of the individual who experiences social anxiety when talking in her second language, described earlier, trait vulnerability may be moderate (given her history of shyness) and current situations may be moderate (given her move to an unfamiliar environment).

Towards an Evolutionary Perspective

A classical approach to diagnostic categories holds that they can be defined by necessary and sufficient conditions. Clearly psychiatric diagnoses are not that kind of entity (Stein, 2012). At the same time, this does not mean that we should regard diagnosis merely as a “labeling” process, reflective solely of societal power. The personal suffering and the socioeconomic costs of mental disorder are considerable, and even if they have blurry boundaries, these conditions clearly deserve recognition (Nesse, 2001; Wakefield, 1992). Furthermore, all psychiatric symptoms, even those that are normal emotions, are underpinned by specific psychobiological mechanisms; understanding those mechanisms can potentially lead to new treatments.

Any question in biology can be answered with respect to both the relevant underlying proximal (physiological) and distal (evolutionary) mechanisms (Mayr, 1988). Most medical and neuroscience investigations focus on proximate questions about what a trait is like and how it works and how it develops over the course of an individual’s life (ontogeny). However, a complete evolutionary explanation required answers to questions about how the trait developed over time in the history of the species (phylogeny) and what evolutionary factors shaped the trait (Nesse, 2013; Tinbergen, 1963). This evolutionary perspective contributes to practical questions in the diagnosis, understanding, and treatment of psychiatric disorders in several ways.

First, it helps to provide a framework for differentiating normal from abnormal emotions by calling attention to both the normal functions of emotions and the situations that arouse them. We have previously argued that characterizing an expression of emotion as abnormal without evaluating the relevant life context is akin to diagnosing chronic pain without assessing possible causes of tissue damage (Nesse & Stein, 2012). However, clinicians are not always able to differentiate normal from abnormal emotion based on the qualitative assessment of the emotion, or to assess reliably the extent to which context explains symptoms (Maj, 2011). Further, because normal and “excessive” emotions lie on a continuum, we are unlikely to find biomarkers that define such circuit abnormalities except in very severe cases.

Thus, for the foreseeable future, physicians will use pragmatic considerations, such as assessment of associated distress and impairment, and knowledge of available interventions, to make decisions about what is normal and what is abnormal. In generalized anxiety disorder, even when worrying is judged not to be “excessive,” there may still be considerable impairment (Ruscio et al., 2005). Similarly, in deciding whether or not to diagnose and treat an individual with social anxiety, the fact that such anxiety can be conceptualized as adaptive in certain circumstances should not necessarily delay treatment that can relieve symptoms that cause suffering and interfere with the person’s work performance.

A second issue is how to understand pathogenesis. We have emphasized the importance of considering both proximate and evolutionary explanations for vulnerability to psychiatric disorders. An evolutionary perspective also emphasizes individual variations and the multiple overlapping pathways that can contribute to conditions such as the anxiety disorders (Nesse, 1998). Thus, for example, genetic variations can influence vulnerability to anxiety disorders via many pathways, for example, preference for alcohol, nervousness that impedes making a needed major life change, or variations that make people unfriendly or isolated.

Fortunately, we now have a growing understanding of the proximal psychobiology of anxiety and related disorders that is being linked with an evolutionary understanding of the adaptive nature of fear and anxiety (Marks & Nesse, 1994; Stein & Bouwer, 1997; Korte, et al., 2005; Nettle & Bateson, 2013). For example, in the case of social anxiety disorder, there are now considerable data about the relevant neurocircuitry that underpins symptoms, and about genetic and environmental factors that increase risk for such symptoms. There is also a growing evidence base of relevant endophenotypes; behavioral inhibition may play a key role in the pathogenesis of social anxiety disorder. At the same time, based on an understanding of the different functions of social anxiety, we can expect that a range of different endophenotypes and pathways may ultimately lead to social anxiety disorder (Leary & Kowalski, 1995).

A third issue is how to optimize intervention. Evolutionary theory provides a theoretical foundation for psychotherapy by providing concepts for understanding emotion regulation, human motivation, and social conflicts—that is, an integrative psychobiological approach (Gilbert & Bailey, 2000; Troisi, 2012; Gilbert, 2015). DSM-5 and ICD-11 provide a pragmatic clinical tool for use in daily practice. RDoC constructs will hopefully guide more productive research on proximal mechanisms. The field of global health will appropriately emphasize the importance of feasible and cost-efficient interventions around the world. However, an integrative conceptual framework is needed for understanding the nature of normal emotions and of mental disorder, and for assessing the range of theory and data that are relevant to decision-making regarding psychiatric intervention.

Evolutionary theory does not provide any particular diagnostic tool, or any particular therapeutic intervention, but it does contribute significantly to this conceptual frame. Again, returning to the example of social anxiety disorder, an evolutionary perspective is helpful in explaining the nature of social hierarchies across species, the importance of appeasement displays in avoiding social conflicts, and the way in which a range of factors (e.g., introduction of new conspecifics) may disrupt social hierarchies. This kind of perspective is relevant to understanding the function of social anxiety in particular individuals at particular times, and so in providing the clinician an overarching conceptual framework within which to make pragmatic decisions about diagnosis and treatment.

Conclusion

Dissatisfaction with psychiatric diagnostic systems arises mainly, we believe, from unrealistic expectations. The failure to find biomarkers for DSM diagnoses should not be blamed on the DSM categories. The conviction that other categories will prove superior may be based on the false assumption most mental disorders arise from specific brain abnormities. Diagnostic categories are not diseases with specific causes, they are descriptions of clinically observed syndromes that are useful for research and treatment despite their limitations. Clinicians understandably draw a pragmatic distinction between normal and abnormal emotions based on multiple considerations including severity and duration. At the same time, we argue that psychiatric classifications must increasingly address issues of diagnostic validity, and for emotional disorders, this must be based on a theoretical framework that includes an understanding not only of proximal molecular and psychosocial mechanisms, but also of the evolutionary forces that shaped brain mechanisms that regulate the expression of emotions.