Article Text

Download PDFPDF

Psychiatry's new manual (DSM-5): ethical and conceptual dimensions
  1. J S Blumenthal-Barby
  1. Correspondence to Professor J S Blumenthal-Barby, Center for Medical Ethics and Health Policy, Baylor College of Medicine, One Baylor Plaza, MS 420, Houston, TX 77030, USA; jsswinde{at}


The introduction of the Diagnostic and statistical manual of mental disorders (DSM-5) in May 2013 is being hailed as the biggest event in psychiatry in the last 10 years. In this paper I examine three important issues that arise from the new manual:(1) Expanding nosology: Psychiatry has again broadened its nosology to include human experiences not previously under its purview (eg, binge eating disorder, internet gaming disorder, caffeine use disorder, hoarding disorder, premenstrual dysphoric disorder). Consequence-based ethical concerns about this expansion are addressed, along with conceptual concerns about a confusion of “construct validity” and “conceptual validity” and a failure to distinguish between “disorder” and “non disordered conditions for which we help people.”(2) The role of claims about societal impact in changes in nosology: Several changes in the DSM-5 involved claims about societal impact in their rationales. This is due in part to a new online open comment period during DSM development. Examples include advancement of science, greater access to treatment, greater public awareness of condition, loss of identify or harm to those with removed disorders, stigmatization, offensiveness, etc. I identify and evaluate four importantly distinct ways in which claims about societal impact might operate in DSM development. (3) Categorisation nosology to spectrum nosology: The move to “degrees of severity” of mental disorders, a major change for DSM-5, raises concerns about conceptual clarity and uniformity concerning what it means to have a severe form of a disorder, and ethical concerns about communication.

  • Concept of Mental Health
  • Psychiatry

Statistics from

The Diagnostic and statistical manual of mental disorders (DSM) is published by the American Psychiatric Association and provides criteria for the classification of mental disorders (‘psychiatric nosology’). It was first published in 1952. The second version (DSM-II) was published in 1968, the DSM-III in 1984, a ‘revised’ version of DSM-III in 1987, the DSM-IV in 1994, a ‘revised’ version of DSM-IV in 2000, and the DSM-5i was just published in May of 2013. The evolution of the DSM-5, including the rationales for all proposed changes, can be followed at The table of contents for the DSM-5 was released in January 2013 (

In this paper, I examine three important issues that arise from the new manual: (1) ethical and conceptual issues relating to the relatively significant expansion of psychiatric nosology; (2) the role of claims about societal impact in rationales for changes in nosology; and (3) ethical and conceptual issues relating to the move to ‘degrees of severity’ of mental disorders (the move from ‘categorisation’ nosology to ‘spectrum’ nosology).

Expanding nosology

Psychiatry has once again broadened its nosology to include human experiences not previously under its purview. A growing number of phenomena that were once clinically unremarkable are now labelled as mental disorders and will likely be treated pharmacologically. Examples include: the removal of the bereavement exclusion criterion for diagnosis of major depressive episode, which allows grieving persons to be diagnosed with major depression; the new disorders of premenstrual dysphoric disorder, disruptive mood dysregulation disorder, illness anxiety disorder, hoarding disorder, excoriation (skin picking) disorder, binge eating disorder, language/speech/social communication disorders and minor neurocognitive disorder; and the ‘conditions for further study’ of persistent complex bereavement disorder, attenuated psychosis syndrome, internet gaming disorder, caffeine use disorder, non-suicidal self-injury disorder, suicidal behaviour disorder and neurobehavioral disorder associated with prenatal alcohol disorder. Other conditions for further study that were proposed by the working groups in the draft version of the DSM-5 until a couple of months before the final table of contents was released, but ultimately not included, are mixed anxiety/depression, hypersexual disorder and paraphilic coercive disorder.

Consequence-based ethical concerns about expansion

There are at least five consequence-based ethical concerns about the expansion: (i) over-diagnosis/false positives in practice; (ii) risks and costs associated with pharmacological management of new conditions; (iii) medicalisation of phenomena that results in a shift to individual responsibility and neglect of larger structural issues; (iv) trivialisation of the concept of mental disorder/decrediting of psychiatry; and (v) treatment or eradication of phenomena that are desirable or valuable in some way.

Turning to the first concern, the authors of the DSM-IV have critiqued the authors of the DSM-5 for expansions that they believe will cause harm from over-diagnosis and false positives in practice. In a recent article they wrote: Of the nearly 100 new diagnoses suggested for DSM-IV, we rejected all but a few—making calculated gambles by including Asperger's and bipolar II. Both of these were reasonable decisions that filled an important vacuum in the diagnostic classification. But each has also been wildly overused in ways that were never intended. We made only small cosmetic changes in the definition of attention deficit disorder…and the convergence of two anticipated events led to doubling of rates: (a) the FDA approved new drugs for ADD, and (b) drug companies were given permission to conduct direct-to-consumer advertising campaigns. DSM 5 threatens to provoke several more epidemics (eg, disruptive mood dysregulation, psychosis risk, binge eating, mixed anxiety depression, and minor neurocognitive).1

The removal of the bereavement exclusion criterion for the diagnosis of major depression is another example where concerns of over-diagnosis and false positives have been raised, since many of these cases will probably be dealt with in primary care settings by practitioners not specifically trained in psychiatric diagnosis.2 ,3

A second consequence-based ethical concern about the expansion is that treatment usually involves pharmacological management—which carries known and unknown risks—and costs to the individual and the healthcare system. This concern is especially salient with respect to children in the DSM-5. As the DSM-IV authors note in their critique, ‘Some of the proposed new categories (eg, attenuated psychotic symptoms, mood dysregulation) will further promote what is already an alarming overuse of antipsychotic medication in children—sometimes causing obesity, diabetes, cardiovascular complications, and possible reduced life expectancy.’1

A third consequence-based ethical concern about the expansion is that, in becoming ‘medicalised’, these phenomena become problems of individuals that individuals bear the responsibility of controlling and managing, and this results in overlooking larger issues and societal structures. Carl Elliot develops a poignant analogy when he writes, ‘On Prozac, Sisyphus might well push the boulder back up the mountain with more enthusiasm and creativity. I do not want to deny the benefits of psychoactive medication. I just want to point out that [t]o see him as a patient with a mental health problem is to ignore certain larger aspects of his predicament connected to boulders, mountains, and eternity.’4 This concern reminds us of the larger structure, and indeed the world, in which mental disorders do or do not exist (eg, consider the proposed hypersexual disorder with pornography subtype, or the ‘condition for further study’ of internet gaming disorder). Another example that demonstrates this concern is the disorder of female sexual interest/arousal disorder. Perhaps there is indeed a ‘dysfunction’ in normal biological sexual functioning (eg, inability to attain or maintain an adequate lubrication–swelling response of sexual excitement) and that dysfunction may cause harm or distress to the woman, but one might inquire why the woman has the disorder (eg, perhaps it is due to lack of ability of her sexual partner, that this inability is due to misconstrued societal norms and messages about female sexuality, etc).

A fourth consequence-based ethical concern is that the expansion (especially with the addition of disorders such as internet gaming disorder) may trivialise the concept of a mental disorder and harm the credibility of psychiatry. This concern has been realised to some extent in the media coverage of DSM-5 (a simple search for DSM-5 using an internet search engine will reveal this occurrence).

The final consequence-based ethical concern about expansion is that it might lead to the pharmacological eradication of phenomena of value. I think that this concern is limited in its application given that most of the disorders involved in the expansion produce significant harms for patients and their family members, but it is worth mentioning, especially with respect to the removal of the bereavement exclusion criterion for depression. The basic idea is that phenomena labelled (correctly or incorrectly) as disorders will be viewed negatively—as things to be ‘treated’ and eradicated—and, in some cases such as grief, we will lose something of (instrumental or non-instrumental) value. Normal grief, for example, has instrumental value in that it allows people to come to terms with their loss, and it also has non-instrumental value in that it (along with love) is a part of the human experience.5

Conceptual concerns about expansion

There are at least two conceptual concerns about the expansion: (i) a confusion of ‘construct validity’ and ‘conceptual validity’; (ii) a failure to distinguish between ‘disorder’ and ‘non-disordered conditions for which we help people’. To take the first point first, philosopher and practitioner, Jerome Wakefield, has argued that much of the expansion is due to confusion of construct validity with conceptual validity. The expansions may meet the criteria for having construct validity (ie, they pick out one kind of condition with a distinctive aetiology), but many do not meet the criteria for conceptual validity (ie, correct application of the concept of disorder). He gives the fictitious example of ‘youth identity disorder’ (anxiety about life choices) having construct validity but not conceptual validity.6 According to Wakefield, the correct application of the concept of disorder is ‘harmful dysfunction’, where ‘harm’ is the value component and ‘dysfunction’ is the factual component based on failure of normal variation in biologically designed functioning. Both components are necessary (and sufficient) for disorder. For example, adultery is harmful, or negatively socially valued, but the desires underlying it are not outside the range of human biological design. Thus, it is not a disorder. The same is true of bereavement, so the criticism goes.6

Interestingly, if one looks at the criteria used by the DSM-5 committee, the focus does seem to be on construct validity. Criteria for including something as a mental disorder must involve sufficient empirical evidence that the disorder in question (a) is distinct from other disorders, (b) has antecedent validators such as familial aggregation, presence in diverse populations and environmental risk factors, (c) has concurrent validators such as cognitive and temperament correlates, biological markers and a certain comorbidity profile, and (d) has predictive validity with respect to diagnostic stability, predictability of the course of illness, and response to treatment.7

To be fair, the definition of mental disorder in the DSM does refer to components of conceptual validity: ‘A mental disorder (a) is a clinically significant behavioural or psychological syndrome or pattern that occurs in an individual, (b) is associated with present distress or disability or with a significantly increased risk of suffering death, pain disability, or an important loss of freedom, (c) must not be merely an expectable and culturally sanctioned response to a particular event, for example, death of a loved one, (d) is a manifestation of a behavioural, psychological, or biological dysfunction in the individual, and (e) neither deviant behaviour (eg, political, religious, or sexual) nor conflicts that are primarily between the individual and society are mental disorders unless the deviance or conflict is a symptom of dysfunction in the individual.’8

However, the DSM developers do not rely heavily on the definition of ‘mental disorder’ in making decisions about psychiatric nosology. The text of the DSM itself recognises this, noting, ‘No definition adequately specifies precise boundaries for the concept of ‘mental disorder’’, and ‘The concept of mental disorder (like many other concepts in medicine and science) lacks a consistent operational definition that covers all situations.’8 And DSM-IV authors note ‘[W]e have reviewed dozens of definitions of mental disorder (and helped to write the one in DSM-IV) and must admit that none have much practical value. Historically, conditions have become mental disorders by accretion and practical necessity, not because they met some independent set of abstract and operationalised definitional criteria.’1 While the messiness of the concept of mental disorder is undeniable, analysis of its application should still be part of all decisions regarding changes in nosology. Construct and predictive validity should not be the only criteria for inclusion.

A final conceptual concern about the expansion of the DSM-5 is a potential failure to distinguish between ‘disorder’ and ‘non-disordered conditions for which we help people’.6 Many of the appeals for inclusion of new disorders or expansion of criteria for existing disorders (discussed more in the next section) involved claims about how inclusion would allow patients to access treatment that helps mitigate harmful or distressful symptoms. Yet, the problem with this reasoning is that, just because something is causing harm and that harm could be alleviated with treatment, it does not mean that that thing is a disorder (eg, pain during childbirth). This is an important point about bereavement.1 A common rationale offered in defence of the removal of the bereavement exclusion criterion for the diagnosis of major depression is providing grieving people easy access to treatment to reduce suffering or suicide risk. Yet, it is not necessary that we diagnose them with a disorder to provide them with help in terms of prescription medication, counselling, monitoring, etc.

Summary: concerns about expansion

In sum, there are significant consequence-based concerns about the relatively significant expansion of the DSM found in the DSM-5, and the messy conceptual terrain of the definition of, or criteria for, a mental disorder does not help and is (as I think most would now admit) heavily value-laden. Thus, decisions about changes in DSM nosology should be made considering (in a robust, evidence-based and democratic way) the values–risks and values–benefits of those changes. Changes should only be made when (a) there is convincing empirical evidence supporting the scientific claims made (if the ‘experts’ cannot even agree what the evidence points towards, then that is a red flag—as was the case in many of the work groups for DSM-5), and (b) a thorough, democratic, evidence-supported (not merely speculative) risk–benefit analysis comes out in favour of the benefits. These guidelines are similar to the widely accepted guidelines for any change in practice or policy guidelines proposed by David Eddy in a classic JAMA article.9 Many of the DSM-5 expansions do not meet these criteria.

The role of social reasons and changes in nosology

The second major issue that arises from the new manual is the role that claims about societal impact do play and should play in rationales for changes in nosology. I believe that this issue has become even more central in the DSM-5, because its development was the first to take place in the era of the internet and open public commentary and communication (DSM-IV was developed in 1994), and that this will be a major issue for development of future editions as well. For example, DSM-5 developers (to their credit) posted draft diagnostic criteria on a website ( during three open comment periods. In total, more than 15 000 comments were submitted, about 800 general and the rest pertaining to specific working groups working on specific diagnostic categories.10 It was via this mechanism that many opinions, concerns and claims about societal impact came to the attention of the DSM-5 developers.

Indeed, several changes in the DSM-5 involved claims about societal impact in their rationales. Examples (further detailed in table 1) include advancement of science, greater access to treatment, greater public awareness of condition, loss of identity or harm to those with removed disorders (what anthropologist Paul Rabinow refers to as harm to ‘biosociality’),11 stigmatisation, offensiveness and forensics. This raises a challenging question about the role that claims about societal impact should play in DSM development. They clearly play some role, as ‘[T]he DSM project includes not only the text of the DSM and its architects but also the way in which the text is received and interpreted by clinicians, and the lay public.’12

Table 1

Typology of roles of claims about societal impact (CSI) in changes in psychiatric nosology

In order to help clarify what roles they might play, I have identified four importantly distinct ways in which claims about societal impact might operate (table 1). First, claims about societal impact might call for a change in terminology or language use. Second, claims about societal impact might add to claims about scientific evidence that support a change in nosology (addition/removal of a disorder or change in diagnostic criteria). Third, claims about societal impact might substitute for missing scientific evidence supporting a change in nosology. Fourth, claims about societal impact might be in tension with or subtract from scientific evidence supporting a change in nosology.

Changes in scientific/clinical terminology (category 1 of my schema) will be the least controversial category to deal with. For the most part, most of the terminology concerns brought to the attention of the DSM-5 developers were noted and changes were made. ‘Conversion disorder’ has added (in parentheses) the name ‘functional neurological symptom disorder’, ‘gender identity disorder’ has been renamed ‘gender dysphoria’, ‘mental retardation’ has been renamed ‘intellectual development disorder’, ‘social phobia’ has been renamed ‘social anxiety disorder’, ‘premature ejaculation’ has added (in parentheses) the name ‘early ejaculation’, and ‘vaginismus/dyspareunia’ has been renamed ‘genitopelvic pain/penetration disorder’. One terminology concern that did not result in a change was ‘borderline personality disorder’ (a term that some found to be pejorative or stigmatising; John Oldham MD, MS, Past President of the American Psychiatric Association and member of the Personality and Personality Disorders Work Group for dsm-5. 2013).

Changes where claims about societal impact support or add to claims about scientific evidence (category 2 of my schema) are also uncontroversial. So long as scientific evidence for change exists, then it seems to be of little concern if there are additional claims about societal impact supporting the change. There is, however, a fine line between that category and the category where claims about societal impact substitute for missing scientific evidence (category 3 of my schema) supporting a change in nosology. As an outsider, a non-scientist, I cannot tell whether the examples I am about to mention fall into category 2 or category 3, but it is important that developers are clear about which category they fall into, since category 3 is problematic because social impact is by itself not a sufficient reason to add a new disorder or alter diagnostic criteria (conceptual and construct validity supported by scientific evidence are also required). The examples are the new disorders of ‘premenstrual dysphoric disorder’ (rationale noted that inclusion would help facilitate research and the development of treatments),13 ‘hoarding disorder’ (rationale noted that inclusion would increase public awareness, improve identification of cases, and stimulate research and treatment development),14 and ‘disruptive mood dysregulation disorder’ (rationale noted that inclusion would allow these severely impaired youth to have a ‘home’ in the DSM, enabling them to receive intensive attention and services, and fostering future research).15 Examples also include the section III ‘conditions for further study’ of ‘neurobehavioral disorder associated with prenatal alcohol exposure’ (rationale noted that inclusion would help individuals with these impairments to have access to appropriate mental health services, which may help prevent delinquency, school failure and drug and alcohol abuse)16 and ‘caffeine use disorder’ (rationale noted that inclusion would increase awareness among healthcare providers and the general population that some caffeine users want to stop but cannot, which might lead to development of treatment and intervention strategies, and to more people seeking treatment).17

The final and most challenging category is the category where claims about societal impact are in tension with or subtract from claims about the scientific evidence that has been provided in support of a change (category 4 of my schema). A good example of this in the development of the DSM-5 is the removal of Asperger's disorder as an independent mental disorder. The work group felt that the scientific and clinical literature pointed to a spectrum disorder that encompassed both autism and Asperger's rather than each disorder having independent construct and clinical validity. The tension they faced was the claim (from some of the public comments) that this change would result in loss of identity or ‘fall from grace’ for people with Asperger's. While the work group felt that it was important to consider the potential harms to people with the diagnosis of Asperger's, they ultimately did not think that they should over-ride their scientific reasons for the change.

Consider a second example: in DSM-IV, one criterion for diagnosis of antisocial personality disorder was that the individual be at least 18 years of age (John Oldham MD, MS, Past President of the American Psychiatric Association and member of the Personality and Personality Disorders Work Group for DSM-5. 2013). The work group felt that this was an arbitrary point and that scientifically the clause should be removed. However, there were forensic reasons for leaving the clause in. Namely, the courts were watching the DSM-5 development closely, since, if the clause were removed, young adolescents who displayed requisite antisocial traits and were on trial for a crime could be labelled as having a mental disorder and that could be seen as either exculpating them from blame for their act or as generating reasons for the view that they would be of special danger in the future and should receive heavier or more restrictive sentences. The work group did not want the DSM criteria to be doing the forensic work in question when it came to young adolescents, and so it decided to leave in the ‘over the age of 18’ requirement for antisocial personality disorder even though it is now removed for every other personality disorder in DSM-5. This is a case in which the scientific conclusions were over-ridden by claims about societal impact.

What are we to make of these cases? Are there any circumstances in which concerns about societal impact might justifiably over-ride scientific conclusions? I would argue that there are, since the DSM is not just a scientific manual, but also a manual for the clinical care of patients (which involves fiduciary obligations to protect and promote patient well-being), but that any claims about societal impact of the removal or addition of a disorder (or change in criteria) must be supported by empirical evidence (about the nature, scope, severity, probability, and reversibility of the harm) as much as possible and not merely be speculative or hypothetical, and that the conversation and risk–benefit analysis should not only be conducted by a small group of psychiatrists devoted to a particular disorder or category of disorders, but involve other stakeholders as well. This adds to the moral and political legitimacy of any changes made.

Summary: the role of social reasons and changes in nosology

DSM-5 developers incorporated several claims about societal impact into their rationales for various changes in nosology, alongside claims about scientific evidence. These sorts of consideration need to be carefully delineated, for some changes involving them are benign or prima facie desirable, whereas others are problematic or raise questions about moral and political legitimacy. I have offered a taxonomy of the various roles such claims might play as a starting point to guide developers (table 1), and have stressed the importance of these claims being evidence based rather than merely hypothetical or speculative.

Categorisation nosology to spectrum nosology

The final major issue arising from the new manual that I will discuss in this paper is the move to degrees of severity—a move that the DSM-5 seeks to make for all mental disorders. Previous DSMs had a multiaxial approach in which axes I and II were where mental disorders were coded as present or absent, axis III was where general medical diagnoses were coded, axis IV was where psychosocial and environmental problems were noted, and axis V was where global assessment of functioning was noted. The DSM-5 is dropping the multiaxial classification system and instead adding dimensions of severity for each mental disorder (ie, ‘dimensional assessments’).18 The overall aim is to have less of a categorical approach and more of a spectrum/degrees of severity approach to mental disorders.19 This major change for the DSM raises concerns about conceptual clarity and uniformity about what it means to have a severe form of a disorder, along with ethical concerns about communication.

Conceptual confusion about severity ratings

Intuitively, the move to capture severity rather than a simple dichotomous presence or absence of a disorder makes sense. But there seems to be a conceptual unclarity, or at least non-uniformity, about what it means to have a mild versus moderate versus severe form of a mental disorder. Take for example alcohol use disorder (side note that the abuse/dependence distinction has been dropped). A person has a ‘severe’ form of alcohol use disorder if he/she has all six of the criteria that mark the disorder. But take hypersexual disorder. While it was not ultimately included as a disorder, the authors indicated that a person would have the disorder if he/she had four of the five symptoms and would have a severe form of the disorder if he/she had more than 2 h of fantasies/urges and behaviour six or more times during the average week (compared with less than 15 min of urges/fantasies in the average day but no behaviour, which would indicate a mild form). Thus, one conception of mental disorder severity focuses on number of symptoms displayed and the other focuses on time spent exhibiting symptoms. Now, I want to be cautious, because the actual dimensional assessment scales for various disorders are still in the process of being developed, but, on the basis of the two examples I have just given, it seems that there are at least two different conceptions of severity operating. And we can imagine other conceptions that might operate as well —for example, impairment in functioning or likelihood of cure or successful management. It is interestingly paradoxical that part of the move from a multiaxial system to a degrees-of-severity system was a sense of arbitrariness in the view that when a patient has more than half (or some other number) of the symptoms or criteria for a disorder, they then ‘have’ that disorder, but the new system retains that approach to a large extent (ie, if you have some number of symptoms or criteria, then you have a ‘moderate’ form, but if you have some other number of symptoms or criteria, then you have a ‘severe’ form).

Ethical concerns about communication of severity ratings

Conceptual issues aside, the real question about degrees of severity is what they will mean to those to whom we communicate them—after all, however we choose to conceptualise ‘severe’, the main point of conceptualising it as severe is to deliver a certain message. What sort of message do we want ratings of severity to deliver—what is the intended and actual function of severity ratings? The message has several potential recipients: the patient him/herself; other psychiatrists who might treat that patient in the future; other caregivers who might help the patient in various capacities; and resource holders who control access to various services that could help the patient. Patients are obviously our central focus of concern, so let us examine the issue with respect to them first. I believe it is an open question whether severity ratings are intended to be, or will be, communicated to patients; or whether a patient will be told his/her diagnosis, with the severity rating being something that is just recorded in the medical chart for the psychiatrist to refer to for clinical management (eg, dosage of medication). From a normative standpoint, I believe that patients have a right to know their severity ratings (just as patients have a right to know their cancer staging), and this argument is rooted in autonomy, the importance of such information for informed decision-making, and the assumption that most ‘reasonable patients’ (to use a popular standard of disclosure) would deem such information to be relevant and material. However, the question of what exactly we want the patient to take from this rating is different. I can only imagine that we want the patient to take away an understanding of ‘how sick they are’ so that they may understand how important self-care is (be it monitoring or treatment—either psychoanalytical or psychotropic). What exactly these severity ratings do communicate to patients—what they mean to them and what effects they have—are interesting questions that deserve empirical study.

These severity ratings will also communicate things to other people beyond the patient. To other psychiatrists or other clinicians involved in the care of the patient, they will communicate how often the patient needs to be monitored, how much medication they need, how often they need psychotherapy, etc. All of this could be referred to as the ‘clinical utility’ of dimensional assessments of mental disorders. Actual and perceived clinical utility of the dimensional assessments will be an important area of empirical study as well—and will inform the resulting normative discourse. A third interesting question is the effect that severity ratings will have on allocation of healthcare services for those with the mental disorder in question (ie, the message that is communicated to the resource holders). Take for example the issue of organ allocation, where there has been longstanding debate about whether justice (or some other ethical principle) dictates the allocation of available organs to the ‘sickest’/those who need it most versus those who will benefit most (which may often be those who are not the sickest). While psychiatric services are by no means as scarce a resource as organs, they are scarce, especially in publically funded psychiatric hospitals or systems—not all patients receive all of the psychiatric services they need. Thus, how should (and how will) the severity ratings affect the allocation of psychiatric services? I do not wish to repeat the longstanding debates on this issue, but to note that it is both a normative and empirical question deserving attention as dimensional assessments begin to be integrated into psychiatric diagnosis. Indeed, this entire paper has been, in sum, an explication of the normative and empirical questions deserving attention in DSM development.


Lisa Valentine MD, John Coverdale MD, and Maria Merritt PhD for helpful conversations and feedback.



  • Funding Greenwall Faculty Scholars Award (2011-2014).

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i Note the change from Roman numerals to Arabic numerals. Find the explanation for this change here:

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles